Question-Answering-with-Embedded-Context.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Text Embedding API\n",
    "\n",
    "Text embedding allows us to directly convert text documents to vectors with a simple API call with Open AI.\n",
    "\n",
    "Keep in mind, just like other Open AI services, it is not free and it is also important to note it has its own pricing structure (its typically much cheaper than GPT on a token basis, since the processing is simpler). You can view the pricing here: https://openai.com/api/pricing/\n",
    "\n",
    "## Imports"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "import openai \n",
    "import pandas as pd\n",
    "import tiktoken # https://github.com/openai/tiktoken"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 94,
   "metadata": {},
   "outputs": [],
   "source": [
    "openai.api_key = os.getenv(\"OPENAI_API_KEY\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### What happens when GPT doesn't know anything about a topic?\n",
    "\n",
    "For example, we know GPT is limited by its training data not being up to date to the present day (depending on the model, the cut-off can be very recent though). There are also limitations based on how esoteric the topic is. \n",
    "\n",
    "Let's ask GPT about a a \"unicorn\" company"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "---"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 188,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Pentera is a start-up company that provides software solutions to help organizations manage their employee benefits programs. The company has raised $3.5 million in seed funding from investors including Y Combinator, SV Angel, and Social Leverage.\n"
     ]
    }
   ],
   "source": [
    "prompt = \"What does the start-up company Pentera do and who invested in it?\"\n",
    "\n",
    "response = openai.Completion.create(\n",
    "    prompt=prompt,\n",
    "    temperature=0,\n",
    "    max_tokens=500,\n",
    "    model=\"text-davinci-003\"\n",
    ")\n",
    "print(response[\"choices\"][0][\"text\"].strip(\" \\n\"))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "While this may sound some what correct, the model is hallucinating! A common issue with LLMs, they are eager to please and with enough context they can make stuff up that sounds right, but actually isn't. In my personal research, it looks like Y Combinator did NOT actually invest in Pentera. Also Pentera isn't an HR company, Pentera is a penetration testing company that develops and provides an automated security validation platform to reduce cybersecurity risks.\n",
    "\n",
    "We could try to alleviate this issue with some prompt engineering:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 191,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "I cannot answer this question with 100% certainty.\n"
     ]
    }
   ],
   "source": [
    "prompt = \"\"\"Only answer the question below if you have 100% certainty of the facts.\n",
    "\n",
    "Q: What does the start-up company Pentera do and who invested in it?\n",
    "A:\"\"\"\n",
    "\n",
    "\n",
    "response = openai.Completion.create(\n",
    "    prompt=prompt,\n",
    "    temperature=0,\n",
    "    max_tokens=500,\n",
    "    model=\"text-davinci-003\"\n",
    ")\n",
    "print(response[\"choices\"][0][\"text\"].strip(\" \\n\"))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Alright, very interesting! How can we help the model? We could input some context from our own data. In fact, we have a data set about recent Unicorn companies.  "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Text Data\n",
    "\n",
    "Let's grab some text data and send it to Open AI to receive the embeddings back.\n",
    " \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 199,
   "metadata": {},
   "outputs": [],
   "source": [
    "df = pd.read_csv(\"unicorns.csv\") "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 200,
   "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>Updated at</th>\n",
       "      <th>Company</th>\n",
       "      <th>Crunchbase Url</th>\n",
       "      <th>Last Valuation (Billion $)</th>\n",
       "      <th>Date Joined</th>\n",
       "      <th>Year Joined</th>\n",
       "      <th>City</th>\n",
       "      <th>Country</th>\n",
       "      <th>Industry</th>\n",
       "      <th>Investors</th>\n",
       "      <th>Company Website</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>10/31/2022, 2:37:05 AM</td>\n",
       "      <td>Esusu</td>\n",
       "      <td>https://www.cbinsights.com/company/esusu</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1/27/2022</td>\n",
       "      <td>2022</td>\n",
       "      <td>New York</td>\n",
       "      <td>United States</td>\n",
       "      <td>Fintech</td>\n",
       "      <td>[\"Next Play Ventures\",\"Zeal Capital Partners\",...</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>10/31/2022, 2:37:05 AM</td>\n",
       "      <td>Fever Labs</td>\n",
       "      <td>https://www.cbinsights.com/company/fever-labs</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1/26/2022</td>\n",
       "      <td>2022</td>\n",
       "      <td>New York</td>\n",
       "      <td>United States</td>\n",
       "      <td>Internet software &amp; services</td>\n",
       "      <td>[\"Accel\",\"14W\",\"GS Growth\"]</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>10/31/2022, 2:37:04 AM</td>\n",
       "      <td>Minio</td>\n",
       "      <td>https://www.cbinsights.com/company/minio</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1/26/2022</td>\n",
       "      <td>2022</td>\n",
       "      <td>Palo Alto</td>\n",
       "      <td>United States</td>\n",
       "      <td>Data management &amp; analytics</td>\n",
       "      <td>[\"General Catalyst\",\"Nexus Venture Partners\",\"...</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>10/31/2022, 2:37:04 AM</td>\n",
       "      <td>Darwinbox</td>\n",
       "      <td>https://www.cbinsights.com/company/darwinbox</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1/25/2022</td>\n",
       "      <td>2022</td>\n",
       "      <td>Hyderabad</td>\n",
       "      <td>India</td>\n",
       "      <td>Internet software &amp; services</td>\n",
       "      <td>[\"Lightspeed India Partners\",\"Sequoia Capital ...</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>10/31/2022, 2:37:04 AM</td>\n",
       "      <td>Pentera</td>\n",
       "      <td>https://www.cbinsights.com/company/pcysys</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1/11/2022</td>\n",
       "      <td>2022</td>\n",
       "      <td>Petah Tikva</td>\n",
       "      <td>Israel</td>\n",
       "      <td>Cybersecurity</td>\n",
       "      <td>[\"AWZ Ventures\",\"Blackstone\",\"Insight Partners\"]</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "               Updated at     Company  \\\n",
       "0  10/31/2022, 2:37:05 AM       Esusu   \n",
       "1  10/31/2022, 2:37:05 AM  Fever Labs   \n",
       "2  10/31/2022, 2:37:04 AM       Minio   \n",
       "3  10/31/2022, 2:37:04 AM   Darwinbox   \n",
       "4  10/31/2022, 2:37:04 AM     Pentera   \n",
       "\n",
       "                                  Crunchbase Url  Last Valuation (Billion $)  \\\n",
       "0       https://www.cbinsights.com/company/esusu                         1.0   \n",
       "1  https://www.cbinsights.com/company/fever-labs                         1.0   \n",
       "2       https://www.cbinsights.com/company/minio                         1.0   \n",
       "3   https://www.cbinsights.com/company/darwinbox                         1.0   \n",
       "4      https://www.cbinsights.com/company/pcysys                         1.0   \n",
       "\n",
       "  Date Joined  Year Joined         City        Country  \\\n",
       "0   1/27/2022         2022     New York  United States   \n",
       "1   1/26/2022         2022     New York  United States   \n",
       "2   1/26/2022         2022    Palo Alto  United States   \n",
       "3   1/25/2022         2022    Hyderabad          India   \n",
       "4   1/11/2022         2022  Petah Tikva         Israel   \n",
       "\n",
       "                       Industry  \\\n",
       "0                       Fintech   \n",
       "1  Internet software & services   \n",
       "2   Data management & analytics   \n",
       "3  Internet software & services   \n",
       "4                Cybersecurity    \n",
       "\n",
       "                                           Investors  Company Website  \n",
       "0  [\"Next Play Ventures\",\"Zeal Capital Partners\",...              NaN  \n",
       "1                        [\"Accel\",\"14W\",\"GS Growth\"]              NaN  \n",
       "2  [\"General Catalyst\",\"Nexus Venture Partners\",\"...              NaN  \n",
       "3  [\"Lightspeed India Partners\",\"Sequoia Capital ...              NaN  \n",
       "4   [\"AWZ Ventures\",\"Blackstone\",\"Insight Partners\"]              NaN  "
      ]
     },
     "execution_count": 200,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's create a new column that summarizes each company with the information from the other columns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 215,
   "metadata": {},
   "outputs": [],
   "source": [
    "import ast \n",
    "def summary(company,crunchbase_url,city,country,industry,investor_list):\n",
    "    investors = 'The investors in the company are'\n",
    "     \n",
    "    for investor in ast.literal_eval(investor_list):\n",
    "        investors += f\" {investor}, \"\n",
    "\n",
    "    text = f\"{company} has headquarters in {city} in {country} and is in the field of {industry}. {investors}. You can find more information at {crunchbase_url}\"\n",
    "\n",
    "    return text "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 216,
   "metadata": {},
   "outputs": [],
   "source": [
    "df['summary'] = df.apply(lambda df: summary(df['Company'],df['Crunchbase Url'],df['City'],df['Country'],df['Industry'],df['Investors']),axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 217,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'Esusu has headquarters in New York in United States and is in the field of Fintech. The investors in the company are Next Play Ventures,  Zeal Capital Partners,  SoftBank Group, . You can find more information at https://www.cbinsights.com/company/esusu'"
      ]
     },
     "execution_count": 217,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df['summary'][0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 218,
   "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>Updated at</th>\n",
       "      <th>Company</th>\n",
       "      <th>Crunchbase Url</th>\n",
       "      <th>Last Valuation (Billion $)</th>\n",
       "      <th>Date Joined</th>\n",
       "      <th>Year Joined</th>\n",
       "      <th>City</th>\n",
       "      <th>Country</th>\n",
       "      <th>Industry</th>\n",
       "      <th>Investors</th>\n",
       "      <th>Company Website</th>\n",
       "      <th>summary</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>10/31/2022, 2:37:05 AM</td>\n",
       "      <td>Esusu</td>\n",
       "      <td>https://www.cbinsights.com/company/esusu</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1/27/2022</td>\n",
       "      <td>2022</td>\n",
       "      <td>New York</td>\n",
       "      <td>United States</td>\n",
       "      <td>Fintech</td>\n",
       "      <td>[\"Next Play Ventures\",\"Zeal Capital Partners\",...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Esusu has headquarters in New York in United S...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>10/31/2022, 2:37:05 AM</td>\n",
       "      <td>Fever Labs</td>\n",
       "      <td>https://www.cbinsights.com/company/fever-labs</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1/26/2022</td>\n",
       "      <td>2022</td>\n",
       "      <td>New York</td>\n",
       "      <td>United States</td>\n",
       "      <td>Internet software &amp; services</td>\n",
       "      <td>[\"Accel\",\"14W\",\"GS Growth\"]</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Fever Labs has headquarters in New York in Uni...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>10/31/2022, 2:37:04 AM</td>\n",
       "      <td>Minio</td>\n",
       "      <td>https://www.cbinsights.com/company/minio</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1/26/2022</td>\n",
       "      <td>2022</td>\n",
       "      <td>Palo Alto</td>\n",
       "      <td>United States</td>\n",
       "      <td>Data management &amp; analytics</td>\n",
       "      <td>[\"General Catalyst\",\"Nexus Venture Partners\",\"...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Minio has headquarters in Palo Alto in United ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>10/31/2022, 2:37:04 AM</td>\n",
       "      <td>Darwinbox</td>\n",
       "      <td>https://www.cbinsights.com/company/darwinbox</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1/25/2022</td>\n",
       "      <td>2022</td>\n",
       "      <td>Hyderabad</td>\n",
       "      <td>India</td>\n",
       "      <td>Internet software &amp; services</td>\n",
       "      <td>[\"Lightspeed India Partners\",\"Sequoia Capital ...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Darwinbox has headquarters in Hyderabad in Ind...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>10/31/2022, 2:37:04 AM</td>\n",
       "      <td>Pentera</td>\n",
       "      <td>https://www.cbinsights.com/company/pcysys</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1/11/2022</td>\n",
       "      <td>2022</td>\n",
       "      <td>Petah Tikva</td>\n",
       "      <td>Israel</td>\n",
       "      <td>Cybersecurity</td>\n",
       "      <td>[\"AWZ Ventures\",\"Blackstone\",\"Insight Partners\"]</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Pentera has headquarters in Petah Tikva in Isr...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "               Updated at     Company  \\\n",
       "0  10/31/2022, 2:37:05 AM       Esusu   \n",
       "1  10/31/2022, 2:37:05 AM  Fever Labs   \n",
       "2  10/31/2022, 2:37:04 AM       Minio   \n",
       "3  10/31/2022, 2:37:04 AM   Darwinbox   \n",
       "4  10/31/2022, 2:37:04 AM     Pentera   \n",
       "\n",
       "                                  Crunchbase Url  Last Valuation (Billion $)  \\\n",
       "0       https://www.cbinsights.com/company/esusu                         1.0   \n",
       "1  https://www.cbinsights.com/company/fever-labs                         1.0   \n",
       "2       https://www.cbinsights.com/company/minio                         1.0   \n",
       "3   https://www.cbinsights.com/company/darwinbox                         1.0   \n",
       "4      https://www.cbinsights.com/company/pcysys                         1.0   \n",
       "\n",
       "  Date Joined  Year Joined         City        Country  \\\n",
       "0   1/27/2022         2022     New York  United States   \n",
       "1   1/26/2022         2022     New York  United States   \n",
       "2   1/26/2022         2022    Palo Alto  United States   \n",
       "3   1/25/2022         2022    Hyderabad          India   \n",
       "4   1/11/2022         2022  Petah Tikva         Israel   \n",
       "\n",
       "                       Industry  \\\n",
       "0                       Fintech   \n",
       "1  Internet software & services   \n",
       "2   Data management & analytics   \n",
       "3  Internet software & services   \n",
       "4                Cybersecurity    \n",
       "\n",
       "                                           Investors  Company Website  \\\n",
       "0  [\"Next Play Ventures\",\"Zeal Capital Partners\",...              NaN   \n",
       "1                        [\"Accel\",\"14W\",\"GS Growth\"]              NaN   \n",
       "2  [\"General Catalyst\",\"Nexus Venture Partners\",\"...              NaN   \n",
       "3  [\"Lightspeed India Partners\",\"Sequoia Capital ...              NaN   \n",
       "4   [\"AWZ Ventures\",\"Blackstone\",\"Insight Partners\"]              NaN   \n",
       "\n",
       "                                             summary  \n",
       "0  Esusu has headquarters in New York in United S...  \n",
       "1  Fever Labs has headquarters in New York in Uni...  \n",
       "2  Minio has headquarters in Palo Alto in United ...  \n",
       "3  Darwinbox has headquarters in Hyderabad in Ind...  \n",
       "4  Pentera has headquarters in Petah Tikva in Isr...  "
      ]
     },
     "execution_count": 218,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Token Count"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In case you are ever worried about how many tokens your text actually has (to get an estimate of your costs) OpenAI has a library called \"tiktoken\", which allows you to estimate a cost based on token counts.\n",
    "\n",
    "Splitting text strings into tokens is useful because models like GPT-3 see text in the form of tokens. Knowing how many tokens are in a text string can tell you (a) whether the string is too long for a text model to process and (b) how much an OpenAI API call costs (as usage is priced by token). Different models use different encodings.\n",
    "\n",
    "**tiktoken** supports 3 different encodings for OpenAI models:\n",
    "\n",
    "* \"gpt2\" for most gpt-3 models\n",
    "* \"p50k_base\" for code models, and Davinci models, like \"text-davinci-003\"\n",
    "* \"cl100k_base\" for text-embedding-ada-002"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 166,
   "metadata": {},
   "outputs": [],
   "source": [
    "import tiktoken\n",
    "\n",
    "def num_tokens_from_string(string, encoding_name):\n",
    "    \"\"\"Returns the number of tokens in a text string.\"\"\"\n",
    "    encoding = tiktoken.get_encoding(encoding_name)\n",
    "    num_tokens = len(encoding.encode(string))\n",
    "    return num_tokens"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's run a quick example on some text:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 219,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "58"
      ]
     },
     "execution_count": 219,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "num_tokens_from_string(df['summary'][0],encoding_name='cl100k_base')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Note how this is higher than the actual word count, this is because OpenAI tokens are not the same as words, remember things like punctuation and word length come into play, as a rough estimate, 1000 tokens is about 750 words. But with the tool above you can check your real token count before sending text over to OpenAI. Let's get a cost estimate of vectorizing our entire data set:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 220,
   "metadata": {},
   "outputs": [],
   "source": [
    "df['token_count'] = df['summary'].apply(lambda text: num_tokens_from_string(text,'cl100k_base'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 221,
   "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>Updated at</th>\n",
       "      <th>Company</th>\n",
       "      <th>Crunchbase Url</th>\n",
       "      <th>Last Valuation (Billion $)</th>\n",
       "      <th>Date Joined</th>\n",
       "      <th>Year Joined</th>\n",
       "      <th>City</th>\n",
       "      <th>Country</th>\n",
       "      <th>Industry</th>\n",
       "      <th>Investors</th>\n",
       "      <th>Company Website</th>\n",
       "      <th>summary</th>\n",
       "      <th>token_count</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>10/31/2022, 2:37:05 AM</td>\n",
       "      <td>Esusu</td>\n",
       "      <td>https://www.cbinsights.com/company/esusu</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1/27/2022</td>\n",
       "      <td>2022</td>\n",
       "      <td>New York</td>\n",
       "      <td>United States</td>\n",
       "      <td>Fintech</td>\n",
       "      <td>[\"Next Play Ventures\",\"Zeal Capital Partners\",...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Esusu has headquarters in New York in United S...</td>\n",
       "      <td>58</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>10/31/2022, 2:37:05 AM</td>\n",
       "      <td>Fever Labs</td>\n",
       "      <td>https://www.cbinsights.com/company/fever-labs</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1/26/2022</td>\n",
       "      <td>2022</td>\n",
       "      <td>New York</td>\n",
       "      <td>United States</td>\n",
       "      <td>Internet software &amp; services</td>\n",
       "      <td>[\"Accel\",\"14W\",\"GS Growth\"]</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Fever Labs has headquarters in New York in Uni...</td>\n",
       "      <td>60</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>10/31/2022, 2:37:04 AM</td>\n",
       "      <td>Minio</td>\n",
       "      <td>https://www.cbinsights.com/company/minio</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1/26/2022</td>\n",
       "      <td>2022</td>\n",
       "      <td>Palo Alto</td>\n",
       "      <td>United States</td>\n",
       "      <td>Data management &amp; analytics</td>\n",
       "      <td>[\"General Catalyst\",\"Nexus Venture Partners\",\"...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Minio has headquarters in Palo Alto in United ...</td>\n",
       "      <td>57</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>10/31/2022, 2:37:04 AM</td>\n",
       "      <td>Darwinbox</td>\n",
       "      <td>https://www.cbinsights.com/company/darwinbox</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1/25/2022</td>\n",
       "      <td>2022</td>\n",
       "      <td>Hyderabad</td>\n",
       "      <td>India</td>\n",
       "      <td>Internet software &amp; services</td>\n",
       "      <td>[\"Lightspeed India Partners\",\"Sequoia Capital ...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Darwinbox has headquarters in Hyderabad in Ind...</td>\n",
       "      <td>62</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>10/31/2022, 2:37:04 AM</td>\n",
       "      <td>Pentera</td>\n",
       "      <td>https://www.cbinsights.com/company/pcysys</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1/11/2022</td>\n",
       "      <td>2022</td>\n",
       "      <td>Petah Tikva</td>\n",
       "      <td>Israel</td>\n",
       "      <td>Cybersecurity</td>\n",
       "      <td>[\"AWZ Ventures\",\"Blackstone\",\"Insight Partners\"]</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Pentera has headquarters in Petah Tikva in Isr...</td>\n",
       "      <td>58</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "               Updated at     Company  \\\n",
       "0  10/31/2022, 2:37:05 AM       Esusu   \n",
       "1  10/31/2022, 2:37:05 AM  Fever Labs   \n",
       "2  10/31/2022, 2:37:04 AM       Minio   \n",
       "3  10/31/2022, 2:37:04 AM   Darwinbox   \n",
       "4  10/31/2022, 2:37:04 AM     Pentera   \n",
       "\n",
       "                                  Crunchbase Url  Last Valuation (Billion $)  \\\n",
       "0       https://www.cbinsights.com/company/esusu                         1.0   \n",
       "1  https://www.cbinsights.com/company/fever-labs                         1.0   \n",
       "2       https://www.cbinsights.com/company/minio                         1.0   \n",
       "3   https://www.cbinsights.com/company/darwinbox                         1.0   \n",
       "4      https://www.cbinsights.com/company/pcysys                         1.0   \n",
       "\n",
       "  Date Joined  Year Joined         City        Country  \\\n",
       "0   1/27/2022         2022     New York  United States   \n",
       "1   1/26/2022         2022     New York  United States   \n",
       "2   1/26/2022         2022    Palo Alto  United States   \n",
       "3   1/25/2022         2022    Hyderabad          India   \n",
       "4   1/11/2022         2022  Petah Tikva         Israel   \n",
       "\n",
       "                       Industry  \\\n",
       "0                       Fintech   \n",
       "1  Internet software & services   \n",
       "2   Data management & analytics   \n",
       "3  Internet software & services   \n",
       "4                Cybersecurity    \n",
       "\n",
       "                                           Investors  Company Website  \\\n",
       "0  [\"Next Play Ventures\",\"Zeal Capital Partners\",...              NaN   \n",
       "1                        [\"Accel\",\"14W\",\"GS Growth\"]              NaN   \n",
       "2  [\"General Catalyst\",\"Nexus Venture Partners\",\"...              NaN   \n",
       "3  [\"Lightspeed India Partners\",\"Sequoia Capital ...              NaN   \n",
       "4   [\"AWZ Ventures\",\"Blackstone\",\"Insight Partners\"]              NaN   \n",
       "\n",
       "                                             summary  token_count  \n",
       "0  Esusu has headquarters in New York in United S...           58  \n",
       "1  Fever Labs has headquarters in New York in Uni...           60  \n",
       "2  Minio has headquarters in Palo Alto in United ...           57  \n",
       "3  Darwinbox has headquarters in Hyderabad in Ind...           62  \n",
       "4  Pentera has headquarters in Petah Tikva in Isr...           58  "
      ]
     },
     "execution_count": 221,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Estimating Embedding Costs\n",
    "\n",
    "Let's now do a quick monetary estimate of how much this will all cost, currently ADA-002 embedding model costs $0.0004 / 1K tokens\n",
    "\n",
    "Pay careful attention, that isn't 4 cents per 1000 tokens, that would be $0.04, this is 1/100 of that cost, so quite \"inexpensive\" depending on your document workload."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "So, let's estimate the cost:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 222,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.028168400000000003"
      ]
     },
     "execution_count": 222,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df['token_count'].sum() * 0.0004 / 1000"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Another thing to keep in mind is the size limit for embeddings, currently the ADA 002 model max token limit is 8191 tokens, let's quickly check against this limit:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 223,
   "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>Updated at</th>\n",
       "      <th>Company</th>\n",
       "      <th>Crunchbase Url</th>\n",
       "      <th>Last Valuation (Billion $)</th>\n",
       "      <th>Date Joined</th>\n",
       "      <th>Year Joined</th>\n",
       "      <th>City</th>\n",
       "      <th>Country</th>\n",
       "      <th>Industry</th>\n",
       "      <th>Investors</th>\n",
       "      <th>Company Website</th>\n",
       "      <th>summary</th>\n",
       "      <th>token_count</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "Empty DataFrame\n",
       "Columns: [Updated at, Company, Crunchbase Url, Last Valuation (Billion $), Date Joined, Year Joined, City, Country, Industry, Investors, Company Website, summary, token_count]\n",
       "Index: []"
      ]
     },
     "execution_count": 223,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df[df['token_count'] > 8191]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Looks like we're okay! It also looks like this will only cost us abou 3 cents to embed, not too bad!"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Text Embedding\n",
    "\n",
    "To begin, we'll create a simple function to grab the embedding, in our case, we'll specify the ADA 002 model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 227,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_embedding(text):\n",
    "  # Note how this function assumes you already set your Open AI key!\n",
    "    result = openai.Embedding.create(\n",
    "      model='text-embedding-ada-002',\n",
    "      input=text\n",
    "    )\n",
    "    return result[\"data\"][0][\"embedding\"]\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Create Embeddings"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now to create the embeddings, we can simply call these functions, for example:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 228,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[0.012057947926223278,\n",
       " -0.017802061513066292,\n",
       " -0.022373223677277565,\n",
       " -0.034451279789209366,\n",
       " -0.013807321898639202,\n",
       " 0.01033538393676281,\n",
       " -0.016340898349881172,\n",
       " 0.025375980883836746,\n",
       " 0.0037467442452907562,\n",
       " -0.02198447287082672,\n",
       " 0.01943749189376831,\n",
       " -0.013076740317046642,\n",
       " 0.005415687337517738,\n",
       " -0.016877105459570885,\n",
       " 0.01076434925198555,\n",
       " -0.023780766874551773,\n",
       " -0.0008374039898626506,\n",
       " -0.023686930537223816,\n",
       " 0.0170647781342268,\n",
       " 0.0026877359487116337,\n",
       " -0.023123912513256073,\n",
       " -0.015724260360002518,\n",
       " -0.009805879555642605,\n",
       " 0.021314214915037155,\n",
       " -0.02210512012243271,\n",
       " -0.0009475777624174953,\n",
       " -0.011776438914239407,\n",
       " -0.0014787574764341116,\n",
       " -0.011628982611000538,\n",
       " -0.011246935464441776,\n",
       " 0.011977517046034336,\n",
       " -0.0036830694880336523,\n",
       " -0.018968310207128525,\n",
       " -0.004118737298995256,\n",
       " -0.012687990441918373,\n",
       " 0.006142917554825544,\n",
       " -0.02238662913441658,\n",
       " 0.014263097196817398,\n",
       " 0.019370466470718384,\n",
       " -0.011816654354333878,\n",
       " 0.03174343332648277,\n",
       " 0.008733466267585754,\n",
       " -0.014075425453484058,\n",
       " 0.0021012600045651197,\n",
       " -0.028177659958600998,\n",
       " 0.013834131881594658,\n",
       " -0.005261527840048075,\n",
       " 0.0023425528779625893,\n",
       " 0.005512875039130449,\n",
       " 0.03635481372475624,\n",
       " -0.01022143941372633,\n",
       " 0.027561023831367493,\n",
       " -0.019477708265185356,\n",
       " 0.011595468968153,\n",
       " -0.007191872224211693,\n",
       " -0.01012090127915144,\n",
       " -0.005043693818151951,\n",
       " -0.022547490894794464,\n",
       " 0.00854579359292984,\n",
       " -0.00943053513765335,\n",
       " 0.005985407158732414,\n",
       " 0.006829932332038879,\n",
       " -0.022292792797088623,\n",
       " -0.018311457708477974,\n",
       " -0.006102702114731073,\n",
       " 0.007795104756951332,\n",
       " -0.025322360917925835,\n",
       " -0.0038070674054324627,\n",
       " 0.004031604155898094,\n",
       " 0.006183132994920015,\n",
       " 0.019196199253201485,\n",
       " 0.021877232939004898,\n",
       " -0.0019420735770836473,\n",
       " 0.009008272551000118,\n",
       " 0.005030288826674223,\n",
       " -0.001354759675450623,\n",
       " -0.02467891201376915,\n",
       " -0.01012090127915144,\n",
       " 0.014289908111095428,\n",
       " -0.0005056261434219778,\n",
       " -0.0040383064188063145,\n",
       " -0.010858185589313507,\n",
       " -0.03429041802883148,\n",
       " -0.007051117718219757,\n",
       " 0.01338505931198597,\n",
       " 0.02044287882745266,\n",
       " -0.0028888133820146322,\n",
       " 0.023686930537223816,\n",
       " -0.023606499657034874,\n",
       " -0.0004947344423271716,\n",
       " 0.028418954461812973,\n",
       " 0.025148093700408936,\n",
       " -0.004219276364892721,\n",
       " -0.008914436213672161,\n",
       " -0.045792050659656525,\n",
       " 0.036113519221544266,\n",
       " -0.01978602632880211,\n",
       " 0.010241547599434853,\n",
       " -0.01786908693611622,\n",
       " -0.006253510247915983,\n",
       " 0.024008654057979584,\n",
       " 0.009557884186506271,\n",
       " -0.00914902612566948,\n",
       " -0.007922453805804253,\n",
       " -0.020885249599814415,\n",
       " 0.004453866742551327,\n",
       " -0.009524371474981308,\n",
       " -0.01248021051287651,\n",
       " 0.031019555404782295,\n",
       " -0.0006032325327396393,\n",
       " -0.011830059811472893,\n",
       " 0.015469562262296677,\n",
       " -0.01969218999147415,\n",
       " 0.008304500952363014,\n",
       " 0.02906240150332451,\n",
       " -0.015147838741540909,\n",
       " 0.01577788218855858,\n",
       " 0.010771051980555058,\n",
       " -0.016300683841109276,\n",
       " -0.00874687172472477,\n",
       " 0.026904169470071793,\n",
       " 0.02467891201376915,\n",
       " 0.02510787732899189,\n",
       " -0.0096919359639287,\n",
       " 0.018995121121406555,\n",
       " 0.006052433047443628,\n",
       " -0.01394137367606163,\n",
       " -0.018740423023700714,\n",
       " -0.002851949306204915,\n",
       " -0.022748569026589394,\n",
       " 0.024182921275496483,\n",
       " -0.02107292227447033,\n",
       " -0.007366138976067305,\n",
       " 0.018110379576683044,\n",
       " -0.00478899572044611,\n",
       " 0.01054986659437418,\n",
       " -0.017306070774793625,\n",
       " 0.01797632873058319,\n",
       " -0.000850390293635428,\n",
       " -0.0185527503490448,\n",
       " 0.03340567648410797,\n",
       " 0.00043608684791252017,\n",
       " 0.00126511265989393,\n",
       " 0.02006753534078598,\n",
       " -0.0029709201771765947,\n",
       " 0.002344228560104966,\n",
       " 0.013291222974658012,\n",
       " 0.01989326812326908,\n",
       " 0.0077079711481928825,\n",
       " -0.002396173542365432,\n",
       " 0.01479930430650711,\n",
       " -0.03203834965825081,\n",
       " -0.0052380687557160854,\n",
       " -0.004406948573887348,\n",
       " 0.00667912419885397,\n",
       " -0.0006912039243616164,\n",
       " 0.033137571066617966,\n",
       " 0.019772620871663094,\n",
       " -0.013807321898639202,\n",
       " -0.01063700020313263,\n",
       " -0.031850676983594894,\n",
       " 0.009866203181445599,\n",
       " -0.004691808484494686,\n",
       " -0.01290917582809925,\n",
       " 0.009309888817369938,\n",
       " 0.02714546211063862,\n",
       " 0.009872904978692532,\n",
       " -0.014088829979300499,\n",
       " 0.00031544032390229404,\n",
       " -0.003900903509929776,\n",
       " -0.027359945699572563,\n",
       " 0.024638697504997253,\n",
       " -0.013820727355778217,\n",
       " 0.0016748079797253013,\n",
       " 0.00010294749517925084,\n",
       " 0.010938616469502449,\n",
       " 0.004567810799926519,\n",
       " 0.022467060014605522,\n",
       " -0.02705162763595581,\n",
       " -0.021193569526076317,\n",
       " 0.0450681708753109,\n",
       " 0.013606244698166847,\n",
       " 0.025657489895820618,\n",
       " 0.017399907112121582,\n",
       " 0.008646332658827305,\n",
       " -0.01329792570322752,\n",
       " 0.014222881756722927,\n",
       " 0.0016907267272472382,\n",
       " 0.026488609611988068,\n",
       " -0.01663581281900406,\n",
       " 0.020389258861541748,\n",
       " 0.011099478229880333,\n",
       " 0.002387795364484191,\n",
       " -0.015925338491797447,\n",
       " -0.6310080289840698,\n",
       " -0.01646154560148716,\n",
       " -0.0007712160004302859,\n",
       " 0.012761718593537807,\n",
       " -0.00018128393276128918,\n",
       " 0.022668136283755302,\n",
       " 0.02697119489312172,\n",
       " 0.008606117218732834,\n",
       " 0.003622746327891946,\n",
       " -0.002753086155280471,\n",
       " -0.014678657986223698,\n",
       " -0.015670640394091606,\n",
       " -0.015054002404212952,\n",
       " -0.013706782832741737,\n",
       " -0.01958494819700718,\n",
       " -0.014343529008328915,\n",
       " -0.008907733485102654,\n",
       " -0.051261357963085175,\n",
       " 0.002575467573478818,\n",
       " -0.010938616469502449,\n",
       " -0.003941119182854891,\n",
       " 0.018391888588666916,\n",
       " 0.004648241680115461,\n",
       " -0.016099605709314346,\n",
       " 0.029303694143891335,\n",
       " 0.011602171696722507,\n",
       " 0.002536927815526724,\n",
       " -0.021622534841299057,\n",
       " -0.018164001405239105,\n",
       " 0.014155856333673,\n",
       " -0.026300936937332153,\n",
       " 0.042440757155418396,\n",
       " 0.02687736041843891,\n",
       " 0.025014040991663933,\n",
       " 0.04101981222629547,\n",
       " 0.011729520745575428,\n",
       " -0.01809697598218918,\n",
       " 0.01394137367606163,\n",
       " -0.0060088662430644035,\n",
       " 0.023700334131717682,\n",
       " -0.0008839031797833741,\n",
       " -0.024169515818357468,\n",
       " 0.012037839740514755,\n",
       " -0.013740296475589275,\n",
       " 0.01125363726168871,\n",
       " 0.03884817287325859,\n",
       " 0.02702481672167778,\n",
       " -0.04152920842170715,\n",
       " 0.00918924156576395,\n",
       " 0.009370211511850357,\n",
       " 0.004031604155898094,\n",
       " -0.02790955826640129,\n",
       " -0.011789844371378422,\n",
       " 0.009611505083739758,\n",
       " 0.027333134785294533,\n",
       " 0.0006312995683401823,\n",
       " -0.0062367538921535015,\n",
       " -0.014062019996345043,\n",
       " 0.013076740317046642,\n",
       " -0.014490985311567783,\n",
       " -0.014692062512040138,\n",
       " -0.005935137625783682,\n",
       " -0.009484155103564262,\n",
       " -0.02447783388197422,\n",
       " -0.0367569662630558,\n",
       " 0.01168260257691145,\n",
       " -0.018204215914011,\n",
       " 0.003108323086053133,\n",
       " 0.030965935438871384,\n",
       " -0.00914902612566948,\n",
       " 0.0034585329703986645,\n",
       " 0.018874475732445717,\n",
       " 0.004427056293934584,\n",
       " 0.01900852657854557,\n",
       " 0.006809824611991644,\n",
       " 0.020456284284591675,\n",
       " 0.02717227302491665,\n",
       " -0.011200017295777798,\n",
       " -0.01179654709994793,\n",
       " 0.005358715541660786,\n",
       " -0.011850167065858841,\n",
       " -0.013163873925805092,\n",
       " 0.0009777393424883485,\n",
       " -0.027212489396333694,\n",
       " 0.0019102363148704171,\n",
       " -0.005700547248125076,\n",
       " -0.01223891694098711,\n",
       " -0.01202443428337574,\n",
       " -0.0021532049868255854,\n",
       " -0.010228142142295837,\n",
       " 0.015067407861351967,\n",
       " 0.006990794558078051,\n",
       " -0.011481525376439095,\n",
       " 0.000924118678085506,\n",
       " -0.0025788189377635717,\n",
       " 0.02075119875371456,\n",
       " 0.002419632626697421,\n",
       " 0.01125363726168871,\n",
       " 0.021408051252365112,\n",
       " -0.00601221714168787,\n",
       " 0.0021934204269200563,\n",
       " -0.003833877621218562,\n",
       " 0.003073134459555149,\n",
       " 0.0012383023276925087,\n",
       " 0.0183784831315279,\n",
       " 0.022976456210017204,\n",
       " 0.00043692465987987816,\n",
       " -0.01219870150089264,\n",
       " 0.032762229442596436,\n",
       " -0.060644976794719696,\n",
       " -0.011159801855683327,\n",
       " -0.02959860861301422,\n",
       " -0.020737793296575546,\n",
       " -0.0019705595914274454,\n",
       " -0.004859372973442078,\n",
       " -0.04426386207342148,\n",
       " 0.001105088391341269,\n",
       " -0.008210664615035057,\n",
       " 0.0055765495635569096,\n",
       " -0.0034266957081854343,\n",
       " -0.001913587562739849,\n",
       " -0.016126416623592377,\n",
       " 0.005851355381309986,\n",
       " -0.0114212017506361,\n",
       " 0.015121027827262878,\n",
       " 0.004772239364683628,\n",
       " -0.00957799144089222,\n",
       " -0.005516225937753916,\n",
       " -0.012976201251149178,\n",
       " 0.0009668476413935423,\n",
       " -0.012071352452039719,\n",
       " -0.04654273763298988,\n",
       " 0.002057693200185895,\n",
       " -0.002282229717820883,\n",
       " 0.004219276364892721,\n",
       " -0.00463148532435298,\n",
       " 0.008766978979110718,\n",
       " -0.007144954055547714,\n",
       " 0.031153608113527298,\n",
       " -0.021367836743593216,\n",
       " -0.02490679919719696,\n",
       " -0.009571288712322712,\n",
       " 0.015348915942013264,\n",
       " -0.010368896648287773,\n",
       " 0.0007297437987290323,\n",
       " -0.020147966220974922,\n",
       " 0.019102362915873528,\n",
       " -0.012466805055737495,\n",
       " -0.008438552729785442,\n",
       " -0.013981589116156101,\n",
       " -0.002751410473138094,\n",
       " -0.00594854261726141,\n",
       " 0.00028004232444800436,\n",
       " 0.015844907611608505,\n",
       " 0.013230899348855019,\n",
       " -0.012553938664495945,\n",
       " -0.015268485061824322,\n",
       " -0.017627794295549393,\n",
       " -0.014129046350717545,\n",
       " -0.014852925203740597,\n",
       " 0.006699231918901205,\n",
       " 0.018941501155495644,\n",
       " -0.033941883593797684,\n",
       " -0.015496373176574707,\n",
       " -0.0213544312864542,\n",
       " 0.009678530506789684,\n",
       " 0.001918614492751658,\n",
       " 0.002310715615749359,\n",
       " -0.02210512012243271,\n",
       " 0.010268357582390308,\n",
       " -0.015322105959057808,\n",
       " 0.010368896648287773,\n",
       " -0.003297670977190137,\n",
       " 0.016595596447587013,\n",
       " 0.03581860661506653,\n",
       " 0.015791285783052444,\n",
       " -0.009035082533955574,\n",
       " -0.00034434522967785597,\n",
       " -0.003954524174332619,\n",
       " -0.0011687630321830511,\n",
       " 0.018110379576683044,\n",
       " 0.0033848045859485865,\n",
       " -0.01565723493695259,\n",
       " 0.01232605054974556,\n",
       " 0.01943749189376831,\n",
       " -0.0010749268112704158,\n",
       " 0.011843465268611908,\n",
       " 0.018995121121406555,\n",
       " 0.005271581932902336,\n",
       " -0.0074934884905815125,\n",
       " -0.016300683841109276,\n",
       " 0.013043227605521679,\n",
       " -0.02298986166715622,\n",
       " 0.015335510484874249,\n",
       " 0.010087387636303902,\n",
       " 0.021233784034848213,\n",
       " 0.010020362213253975,\n",
       " 0.004946506582200527,\n",
       " 0.00454435171559453,\n",
       " 0.007346031256020069,\n",
       " 0.019544733688235283,\n",
       " 0.0014242989709600806,\n",
       " 0.026032833382487297,\n",
       " -0.018070165067911148,\n",
       " -0.008297798223793507,\n",
       " -0.027413565665483475,\n",
       " -0.0003805810702033341,\n",
       " -0.015509777702391148,\n",
       " 0.016019174829125404,\n",
       " 0.026300936937332153,\n",
       " 0.030188435688614845,\n",
       " -0.023378610610961914,\n",
       " -0.0008947948808781803,\n",
       " 0.0044572181068360806,\n",
       " 0.020429475232958794,\n",
       " 0.00863292720168829,\n",
       " 0.03557731211185455,\n",
       " -0.011568658985197544,\n",
       " -0.012976201251149178,\n",
       " 0.010851482860744,\n",
       " -0.00564022408798337,\n",
       " -0.012922581285238266,\n",
       " 0.01769481971859932,\n",
       " -0.015523183159530163,\n",
       " -0.011240232735872269,\n",
       " 0.021769991144537926,\n",
       " 0.04525584354996681,\n",
       " 0.026046238839626312,\n",
       " 0.008941246196627617,\n",
       " -0.001881750300526619,\n",
       " -0.00548941595479846,\n",
       " 0.02790955826640129,\n",
       " -0.017024561762809753,\n",
       " 0.003961226902902126,\n",
       " 0.00469515984877944,\n",
       " -0.02264132723212242,\n",
       " 0.037829380482435226,\n",
       " -0.03927714005112648,\n",
       " 0.010596784763038158,\n",
       " -0.013519111089408398,\n",
       " 0.027695074677467346,\n",
       " 0.003974631894379854,\n",
       " 0.04225308448076248,\n",
       " -0.01339846383780241,\n",
       " 0.037373606115579605,\n",
       " 0.006072540767490864,\n",
       " 0.025751326233148575,\n",
       " 0.037293173372745514,\n",
       " -0.03345929831266403,\n",
       " 0.004477325826883316,\n",
       " -0.018365079537034035,\n",
       " 0.008780384436249733,\n",
       " 0.0032658337149769068,\n",
       " -0.012748314067721367,\n",
       " 0.016622407361865044,\n",
       " -0.018633181229233742,\n",
       " -7.205277506727725e-05,\n",
       " 0.017279259860515594,\n",
       " 0.019571542739868164,\n",
       " 0.00564692635089159,\n",
       " 0.0021196920424699783,\n",
       " -0.007795104756951332,\n",
       " 0.014558011665940285,\n",
       " 0.02642158418893814,\n",
       " 0.0076409452594816685,\n",
       " -0.018190812319517136,\n",
       " -0.01860637217760086,\n",
       " -0.02916964329779148,\n",
       " -0.016783269122242928,\n",
       " -0.030483350157737732,\n",
       " -0.014222881756722927,\n",
       " -0.01437033899128437,\n",
       " 0.00043441119487397373,\n",
       " 0.007801807019859552,\n",
       " 0.020362447947263718,\n",
       " -0.015080812387168407,\n",
       " 0.014665252529084682,\n",
       " -0.022520679980516434,\n",
       " 0.015603614039719105,\n",
       " -0.028365332633256912,\n",
       " -0.0009492533863522112,\n",
       " -0.04045679420232773,\n",
       " 0.013431977480649948,\n",
       " 0.01618003658950329,\n",
       " 0.023888006806373596,\n",
       " -0.009450642392039299,\n",
       " 0.015764476731419563,\n",
       " 0.0023291476536542177,\n",
       " -0.01752055250108242,\n",
       " 0.039491619914770126,\n",
       " 0.009383616968989372,\n",
       " 0.001152844401076436,\n",
       " -0.02911602333188057,\n",
       " -0.005445849150419235,\n",
       " -0.009986849501729012,\n",
       " 0.011917193420231342,\n",
       " 0.04000101611018181,\n",
       " -0.02193085290491581,\n",
       " -0.0003608922124840319,\n",
       " -0.018391888588666916,\n",
       " 0.024115895852446556,\n",
       " -0.0160593893378973,\n",
       " -0.020710982382297516,\n",
       " 0.0005848003784194589,\n",
       " 0.05667704716324806,\n",
       " 0.015161244198679924,\n",
       " 0.005747465416789055,\n",
       " 0.006548423785716295,\n",
       " 0.006910363677889109,\n",
       " -0.021890638396143913,\n",
       " 0.00041199941188097,\n",
       " -0.017882492393255234,\n",
       " -0.04606015235185623,\n",
       " 0.026126669719815254,\n",
       " -0.003639502916485071,\n",
       " 0.007312518544495106,\n",
       " -0.014062019996345043,\n",
       " -0.007366138976067305,\n",
       " 0.04147558659315109,\n",
       " 0.008753573521971703,\n",
       " 0.00031041339389048517,\n",
       " -0.03418317437171936,\n",
       " -0.028821108862757683,\n",
       " 0.019598353654146194,\n",
       " 0.11367581784725189,\n",
       " 0.006354049313813448,\n",
       " -0.008706656284630299,\n",
       " 0.017735036090016365,\n",
       " -0.0035020997747778893,\n",
       " 0.009055189788341522,\n",
       " -0.018029948696494102,\n",
       " -0.0556582510471344,\n",
       " -0.01219870150089264,\n",
       " 0.007372841704636812,\n",
       " -0.030965935438871384,\n",
       " 0.0024732532911002636,\n",
       " 0.00433657132089138,\n",
       " -0.010965426452457905,\n",
       " 0.0023693633265793324,\n",
       " 0.0027212488930672407,\n",
       " 9.792055789148435e-05,\n",
       " -0.017533957958221436,\n",
       " 3.746639413293451e-05,\n",
       " -0.006783014629036188,\n",
       " -0.011300555430352688,\n",
       " 0.002020828891545534,\n",
       " 0.026502015069127083,\n",
       " 0.03656929358839989,\n",
       " 0.03329843282699585,\n",
       " 0.01774844154715538,\n",
       " 0.01883425936102867,\n",
       " 0.033432487398386,\n",
       " 0.005053747911006212,\n",
       " -0.014960166066884995,\n",
       " 0.012124973349273205,\n",
       " -0.002020828891545534,\n",
       " 0.011783141642808914,\n",
       " 0.020456284284591675,\n",
       " -0.045792050659656525,\n",
       " -0.0002888394519686699,\n",
       " 0.017453527078032494,\n",
       " 0.01454460620880127,\n",
       " 0.007305815815925598,\n",
       " 0.007205277215689421,\n",
       " -0.002642493462190032,\n",
       " 0.022547490894794464,\n",
       " -0.0013405167264863849,\n",
       " -0.013190683908760548,\n",
       " 0.00015960527525749058,\n",
       " -0.027534212917089462,\n",
       " -0.003897552378475666,\n",
       " 0.015603614039719105,\n",
       " 0.00802969466894865,\n",
       " 0.0020442879758775234,\n",
       " 0.01918279379606247,\n",
       " -0.043298687785863876,\n",
       " -0.03174343332648277,\n",
       " -0.017788656055927277,\n",
       " 0.0011143045267090201,\n",
       " -0.017614388838410378,\n",
       " 0.004112035036087036,\n",
       " -0.016783269122242928,\n",
       " -0.023888006806373596,\n",
       " -0.0427088625729084,\n",
       " -0.02327136881649494,\n",
       " -0.012084757909178734,\n",
       " -0.007332626264542341,\n",
       " -0.017413312569260597,\n",
       " 0.012446697801351547,\n",
       " -0.03383463993668556,\n",
       " -0.013445382006466389,\n",
       " -0.0005370444850996137,\n",
       " -0.022896025329828262,\n",
       " -0.02064395695924759,\n",
       " 0.02636796236038208,\n",
       " -0.012801934033632278,\n",
       " -0.015335510484874249,\n",
       " 0.01843210496008396,\n",
       " 0.052977219223976135,\n",
       " -0.00448067719116807,\n",
       " -0.003284265985712409,\n",
       " 0.0017678063595667481,\n",
       " 0.005556441843509674,\n",
       " 0.0354432612657547,\n",
       " -0.009155728854238987,\n",
       " 0.005871463101357222,\n",
       " -0.005522928666323423,\n",
       " -0.0205233097076416,\n",
       " -0.0019085606327280402,\n",
       " 0.002702816855162382,\n",
       " 0.024866584688425064,\n",
       " -0.0012885716278105974,\n",
       " -0.007453272584825754,\n",
       " 0.009450642392039299,\n",
       " -0.007010902278125286,\n",
       " -0.014437364414334297,\n",
       " 0.0043901922181248665,\n",
       " -0.005080558359622955,\n",
       " 0.007124846335500479,\n",
       " -0.0083581218495965,\n",
       " 0.019960293546319008,\n",
       " -0.007607432082295418,\n",
       " -0.008391634561121464,\n",
       " -0.013706782832741737,\n",
       " 0.0006015568505972624,\n",
       " -0.025443006306886673,\n",
       " 0.012875663116574287,\n",
       " 0.004195817280560732,\n",
       " 0.02084503509104252,\n",
       " 0.03512153774499893,\n",
       " 0.003249077359214425,\n",
       " -0.009624909609556198,\n",
       " -0.008606117218732834,\n",
       " 0.027963178232312202,\n",
       " -0.001916938810609281,\n",
       " -0.021005896851420403,\n",
       " 0.009142323397099972,\n",
       " 0.0003261225647293031,\n",
       " -0.008123531006276608,\n",
       " 0.011588767170906067,\n",
       " 0.006719340104609728,\n",
       " 0.02112654224038124,\n",
       " 0.003255779854953289,\n",
       " 0.012419886887073517,\n",
       " 0.013847537338733673,\n",
       " -0.0371055006980896,\n",
       " 0.024290163069963455,\n",
       " 0.016474949195981026,\n",
       " 0.018633181229233742,\n",
       " -0.014638442546129227,\n",
       " -0.0011377634946256876,\n",
       " -0.015281890518963337,\n",
       " -0.005221312399953604,\n",
       " 0.02407567948102951,\n",
       " -0.03493386507034302,\n",
       " 0.024209732189774513,\n",
       " -0.004326517693698406,\n",
       " 0.007265600375831127,\n",
       " -0.0234992578625679,\n",
       " -0.013619649223983288,\n",
       " 0.004229329992085695,\n",
       " -0.004014847334474325,\n",
       " 0.013143765740096569,\n",
       " -0.011427904479205608,\n",
       " -0.004246086813509464,\n",
       " -0.017560768872499466,\n",
       " -0.004993424750864506,\n",
       " -0.01883425936102867,\n",
       " -0.012446697801351547,\n",
       " -0.03168981522321701,\n",
       " 0.0004109521396458149,\n",
       " 0.009866203181445599,\n",
       " 0.016796674579381943,\n",
       " 0.00528163556009531,\n",
       " -0.007607432082295418,\n",
       " 0.01809697598218918,\n",
       " -0.005476010497659445,\n",
       " 0.0018415347440168262,\n",
       " -0.007279005832970142,\n",
       " -0.029786281287670135,\n",
       " -0.019290035590529442,\n",
       " -0.009651720523834229,\n",
       " 0.028418954461812973,\n",
       " 0.023083698004484177,\n",
       " 0.037293173372745514,\n",
       " 0.016247062012553215,\n",
       " 0.0011821681400761008,\n",
       " -0.0026592500507831573,\n",
       " 0.010871590115129948,\n",
       " -0.004259491804987192,\n",
       " -0.01494676060974598,\n",
       " 0.01737309619784355,\n",
       " 0.00973215140402317,\n",
       " 0.016394518315792084,\n",
       " 0.02235981822013855,\n",
       " 0.0025905484799295664,\n",
       " 0.02296305075287819,\n",
       " -0.03171662613749504,\n",
       " -0.02951817773282528,\n",
       " 0.021568913012742996,\n",
       " -0.0118367625400424,\n",
       " -0.01351240836083889,\n",
       " -0.03426360711455345,\n",
       " -0.04477325826883316,\n",
       " -0.022882619872689247,\n",
       " -0.0069639841094613075,\n",
       " -0.01479930430650711,\n",
       " 0.008994867093861103,\n",
       " -0.050912823528051376,\n",
       " -0.005874814465641975,\n",
       " 0.002719573210924864,\n",
       " 0.002964217448607087,\n",
       " 0.010020362213253975,\n",
       " -0.010100793093442917,\n",
       " 0.019531328231096268,\n",
       " -0.01646154560148716,\n",
       " 0.030510159209370613,\n",
       " 0.007017605006694794,\n",
       " 0.006246807985007763,\n",
       " 0.002461523748934269,\n",
       " -0.004333220422267914,\n",
       " -0.02089865505695343,\n",
       " -0.0069237686693668365,\n",
       " 0.02714546211063862,\n",
       " 0.01832486316561699,\n",
       " 0.0068969582207500935,\n",
       " 0.015308700501918793,\n",
       " 0.006555126514285803,\n",
       " -0.0009836041135713458,\n",
       " 0.02825809270143509,\n",
       " -0.023298179730772972,\n",
       " -0.003009459935128689,\n",
       " 0.03600627928972244,\n",
       " -0.017533957958221436,\n",
       " -0.011246935464441776,\n",
       " 0.0002509279584046453,\n",
       " -0.017533957958221436,\n",
       " -0.02992033213376999,\n",
       " 0.012473507784307003,\n",
       " 0.00594854261726141,\n",
       " -0.015710854902863503,\n",
       " 0.018418699502944946,\n",
       " -0.04938463494181633,\n",
       " -4.731081207864918e-05,\n",
       " -0.00772137613967061,\n",
       " 0.0046851057559251785,\n",
       " 0.021394645795226097,\n",
       " 0.008257582783699036,\n",
       " 0.02914283238351345,\n",
       " 0.03155576437711716,\n",
       " 0.018338268622756004,\n",
       " -0.053352564573287964,\n",
       " -0.018043354153633118,\n",
       " -0.000957631622441113,\n",
       " -0.0008231610408984125,\n",
       " 0.015308700501918793,\n",
       " 0.011119585484266281,\n",
       " -0.013793916441500187,\n",
       " -0.00411538640037179,\n",
       " 0.015134433284401894,\n",
       " 0.014102235436439514,\n",
       " -0.01817740686237812,\n",
       " 0.0030044331215322018,\n",
       " 0.005375471897423267,\n",
       " -0.009182539768517017,\n",
       " 0.018499130383133888,\n",
       " -0.011742926202714443,\n",
       " -0.01883425936102867,\n",
       " -0.01597895845770836,\n",
       " 0.0063574002124369144,\n",
       " -0.021716371178627014,\n",
       " 6.613564619328827e-05,\n",
       " -0.010683918371796608,\n",
       " -0.013847537338733673,\n",
       " -0.006256861612200737,\n",
       " 0.01327111478894949,\n",
       " -0.005848004017025232,\n",
       " 0.02092546597123146,\n",
       " 0.00500347837805748,\n",
       " -0.004668349400162697,\n",
       " -0.016783269122242928,\n",
       " -0.0261668860912323,\n",
       " -0.000997847062535584,\n",
       " -0.02528214454650879,\n",
       " -0.013927968218922615,\n",
       " -0.006209943443536758,\n",
       " -0.0015834852820262313,\n",
       " 0.009624909609556198,\n",
       " 0.006809824611991644,\n",
       " 0.007982776500284672,\n",
       " -0.01475908886641264,\n",
       " 0.0038740932941436768,\n",
       " 0.004004793707281351,\n",
       " 0.02697119489312172,\n",
       " -0.0035925847478210926,\n",
       " -0.03244050219655037,\n",
       " 0.021662749350070953,\n",
       " -0.008170449174940586,\n",
       " 0.009698638692498207,\n",
       " -0.02390141226351261,\n",
       " -0.021515293046832085,\n",
       " -0.01321079209446907,\n",
       " -0.004510838538408279,\n",
       " -0.013807321898639202,\n",
       " 0.021113138645887375,\n",
       " 0.015482967719435692,\n",
       " -0.0183784831315279,\n",
       " -0.0010874941945075989,\n",
       " -0.0009132269769906998,\n",
       " -0.02018818072974682,\n",
       " -0.01923641376197338,\n",
       " 0.005385525990277529,\n",
       " 0.01878063939511776,\n",
       " 0.0063741570338606834,\n",
       " -0.011240232735872269,\n",
       " -0.011179909110069275,\n",
       " 0.006608747411519289,\n",
       " 0.017989734187722206,\n",
       " -0.0029725958593189716,\n",
       " 0.02530895546078682,\n",
       " 0.007922453805804253,\n",
       " -0.001229924033395946,\n",
       " -0.004453866742551327,\n",
       " 0.011642387136816978,\n",
       " -0.013981589116156101,\n",
       " 0.019450897350907326,\n",
       " -0.010013659484684467,\n",
       " 0.02508106641471386,\n",
       " -0.007761591579765081,\n",
       " -0.015911933034658432,\n",
       " 0.014450769871473312,\n",
       " -0.01786908693611622,\n",
       " 0.005754167679697275,\n",
       " 0.007714673411101103,\n",
       " 0.008559199050068855,\n",
       " 0.0009735502535477281,\n",
       " 0.011863572522997856,\n",
       " 0.003175348974764347,\n",
       " -0.0014913248596712947,\n",
       " -0.003099944908171892,\n",
       " 0.005566495470702648,\n",
       " -0.0030245408415794373,\n",
       " -0.0033831289038062096,\n",
       " 0.015536588616669178,\n",
       " -0.0031937810126692057,\n",
       " -0.022453654557466507,\n",
       " 0.0038439317140728235,\n",
       " -0.005928434897214174,\n",
       " 0.016769863665103912,\n",
       " 0.004990073386579752,\n",
       " 0.014826114289462566,\n",
       " 0.0025218469090759754,\n",
       " -0.012714800424873829,\n",
       " -0.0328158475458622,\n",
       " -0.0061965384520590305,\n",
       " -0.008813897147774696,\n",
       " -0.021649343892931938,\n",
       " -0.005750816781073809,\n",
       " -0.00601221714168787,\n",
       " -0.021139947697520256,\n",
       " -0.006997497286647558,\n",
       " -0.003441776614636183,\n",
       " 0.03219921141862869,\n",
       " -0.014879735186696053,\n",
       " -0.004212573636323214,\n",
       " 0.016568785533308983,\n",
       " 0.017761847004294395,\n",
       " 0.0025922241620719433,\n",
       " 0.004031604155898094,\n",
       " -0.02006753534078598,\n",
       " -0.033941883593797684,\n",
       " -0.02264132723212242,\n",
       " -0.029250074177980423,\n",
       " -0.0018130488460883498,\n",
       " -9.420271817361936e-05,\n",
       " 0.05018894374370575,\n",
       " 0.006819878704845905,\n",
       " -0.023030076175928116,\n",
       " -0.03466576337814331,\n",
       " 0.004075170960277319,\n",
       " -0.06026963144540787,\n",
       " 0.007661052979528904,\n",
       " -0.003864039434120059,\n",
       " 0.037775758653879166,\n",
       " 0.031421709805727005,\n",
       " 0.00844525545835495,\n",
       " -0.009169134311378002,\n",
       " -0.009470750577747822,\n",
       " 0.007386247161775827,\n",
       " 0.02009434439241886,\n",
       " -0.04423705115914345,\n",
       " 0.003327832790091634,\n",
       " 0.02172977477312088,\n",
       " -0.03286946937441826,\n",
       " 0.005268230568617582,\n",
       " -0.018914690241217613,\n",
       " -0.03284265846014023,\n",
       " -0.0023693633265793324,\n",
       " 0.0183784831315279,\n",
       " 0.00011687629739753902,\n",
       " 0.011649089865386486,\n",
       " 0.022922836244106293,\n",
       " -0.012078055180609226,\n",
       " 0.0062903743237257,\n",
       " 0.009008272551000118,\n",
       " -0.000804728944785893,\n",
       " -0.027721885591745377,\n",
       " 0.002992703579366207,\n",
       " 0.01966537907719612,\n",
       " -0.04348636046051979,\n",
       " 0.0035188563633710146,\n",
       " 0.030215246602892876,\n",
       " -0.020282017067074776,\n",
       " -0.027989989146590233,\n",
       " 0.01660900190472603,\n",
       " -0.013123658485710621,\n",
       " -0.0014033534098416567,\n",
       " 0.005693844519555569,\n",
       " -0.00792915653437376,\n",
       " 0.005593305919319391,\n",
       " -0.03458533063530922,\n",
       " -0.005462605506181717,\n",
       " 0.011696008034050465,\n",
       " -0.018740423023700714,\n",
       " 0.009296483360230923,\n",
       " 0.023190937936306,\n",
       " 0.006786365527659655,\n",
       " -0.0160593893378973,\n",
       " -0.032333262264728546,\n",
       " 0.043727654963731766,\n",
       " -0.021367836743593216,\n",
       " 0.00643448019400239,\n",
       " -0.0019236414227634668,\n",
       " -0.011568658985197544,\n",
       " 0.00433657132089138,\n",
       " -0.010643702931702137,\n",
       " -0.003612692467868328,\n",
       " 0.00820396188646555,\n",
       " -0.018499130383133888,\n",
       " -0.01561701949685812,\n",
       " 0.0019940186757594347,\n",
       " -0.016340898349881172,\n",
       " 0.008619522675871849,\n",
       " 0.00317367329262197,\n",
       " -0.01789589785039425,\n",
       " -0.02674330770969391,\n",
       " 0.024866584688425064,\n",
       " 0.011260339990258217,\n",
       " -0.019477708265185356,\n",
       " -0.012393076904118061,\n",
       " 0.010389003902673721,\n",
       " -0.02044287882745266,\n",
       " -0.0043499767780303955,\n",
       " -0.010415814816951752,\n",
       " 0.00780850974842906,\n",
       " -0.029786281287670135,\n",
       " -0.004313112236559391,\n",
       " 0.010610189288854599,\n",
       " -0.02642158418893814,\n",
       " 0.03383463993668556,\n",
       " 0.19485750794410706,\n",
       " 0.014477579854428768,\n",
       " 0.007104738615453243,\n",
       " 0.036462053656578064,\n",
       " -0.0023760658223181963,\n",
       " 0.008096721023321152,\n",
       " 0.019920077174901962,\n",
       " 0.005784329492598772,\n",
       " -0.01978602632880211,\n",
       " -0.004728672560304403,\n",
       " -0.03721274435520172,\n",
       " -0.012178594246506691,\n",
       " 0.0044974335469305515,\n",
       " -0.0006597855826839805,\n",
       " 0.006360751576721668,\n",
       " -0.024410808458924294,\n",
       " -0.028311712667346,\n",
       " -0.015724260360002518,\n",
       " -0.009933228604495525,\n",
       " 0.028847919777035713,\n",
       " 0.002392822178080678,\n",
       " -0.0047588343732059,\n",
       " 0.008565901778638363,\n",
       " -0.023137317970395088,\n",
       " 0.02473253197968006,\n",
       " 0.004249437712132931,\n",
       " 0.0048225088976323605,\n",
       " 0.002793301595374942,\n",
       " 0.036542486399412155,\n",
       " 0.019651975482702255,\n",
       " -0.01089169830083847,\n",
       " 0.02175658568739891,\n",
       " 0.016823483631014824,\n",
       " 0.016676027327775955,\n",
       " -0.0026944384444504976,\n",
       " -0.028579816222190857,\n",
       " 0.00987960770726204,\n",
       " -0.014678657986223698,\n",
       " 0.023123912513256073,\n",
       " 0.008968056179583073,\n",
       " 0.0014226232888177037,\n",
       " -0.013391762040555477,\n",
       " -0.02788274735212326,\n",
       " 0.007654350250959396,\n",
       " -0.015362321399152279,\n",
       " 0.014142450876533985,\n",
       " ...]"
      ]
     },
     "execution_count": 228,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "get_embedding(df['summary'][0])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's do the rest via our 2nd function:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 229,
   "metadata": {},
   "outputs": [],
   "source": [
    "# this will take awhile due to the amount of calls to the API.\n",
    "# it will take about 0.5 seconds per row\n",
    "df['embedding'] = df['summary'].apply(get_embedding)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 230,
   "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>Updated at</th>\n",
       "      <th>Company</th>\n",
       "      <th>Crunchbase Url</th>\n",
       "      <th>Last Valuation (Billion $)</th>\n",
       "      <th>Date Joined</th>\n",
       "      <th>Year Joined</th>\n",
       "      <th>City</th>\n",
       "      <th>Country</th>\n",
       "      <th>Industry</th>\n",
       "      <th>Investors</th>\n",
       "      <th>Company Website</th>\n",
       "      <th>summary</th>\n",
       "      <th>token_count</th>\n",
       "      <th>embedding</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>10/31/2022, 2:37:05 AM</td>\n",
       "      <td>Esusu</td>\n",
       "      <td>https://www.cbinsights.com/company/esusu</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1/27/2022</td>\n",
       "      <td>2022</td>\n",
       "      <td>New York</td>\n",
       "      <td>United States</td>\n",
       "      <td>Fintech</td>\n",
       "      <td>[\"Next Play Ventures\",\"Zeal Capital Partners\",...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Esusu has headquarters in New York in United S...</td>\n",
       "      <td>58</td>\n",
       "      <td>[0.01195491198450327, -0.017717931419610977, -...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>10/31/2022, 2:37:05 AM</td>\n",
       "      <td>Fever Labs</td>\n",
       "      <td>https://www.cbinsights.com/company/fever-labs</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1/26/2022</td>\n",
       "      <td>2022</td>\n",
       "      <td>New York</td>\n",
       "      <td>United States</td>\n",
       "      <td>Internet software &amp; services</td>\n",
       "      <td>[\"Accel\",\"14W\",\"GS Growth\"]</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Fever Labs has headquarters in New York in Uni...</td>\n",
       "      <td>60</td>\n",
       "      <td>[0.009171437472105026, 0.01314949057996273, -0...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>10/31/2022, 2:37:04 AM</td>\n",
       "      <td>Minio</td>\n",
       "      <td>https://www.cbinsights.com/company/minio</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1/26/2022</td>\n",
       "      <td>2022</td>\n",
       "      <td>Palo Alto</td>\n",
       "      <td>United States</td>\n",
       "      <td>Data management &amp; analytics</td>\n",
       "      <td>[\"General Catalyst\",\"Nexus Venture Partners\",\"...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Minio has headquarters in Palo Alto in United ...</td>\n",
       "      <td>57</td>\n",
       "      <td>[0.002730059437453747, -0.03737899661064148, 0...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>10/31/2022, 2:37:04 AM</td>\n",
       "      <td>Darwinbox</td>\n",
       "      <td>https://www.cbinsights.com/company/darwinbox</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1/25/2022</td>\n",
       "      <td>2022</td>\n",
       "      <td>Hyderabad</td>\n",
       "      <td>India</td>\n",
       "      <td>Internet software &amp; services</td>\n",
       "      <td>[\"Lightspeed India Partners\",\"Sequoia Capital ...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Darwinbox has headquarters in Hyderabad in Ind...</td>\n",
       "      <td>62</td>\n",
       "      <td>[-0.0024771858006715775, -0.024587858468294144...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>10/31/2022, 2:37:04 AM</td>\n",
       "      <td>Pentera</td>\n",
       "      <td>https://www.cbinsights.com/company/pcysys</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1/11/2022</td>\n",
       "      <td>2022</td>\n",
       "      <td>Petah Tikva</td>\n",
       "      <td>Israel</td>\n",
       "      <td>Cybersecurity</td>\n",
       "      <td>[\"AWZ Ventures\",\"Blackstone\",\"Insight Partners\"]</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Pentera has headquarters in Petah Tikva in Isr...</td>\n",
       "      <td>58</td>\n",
       "      <td>[0.011331121437251568, -0.011193273589015007, ...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "               Updated at     Company  \\\n",
       "0  10/31/2022, 2:37:05 AM       Esusu   \n",
       "1  10/31/2022, 2:37:05 AM  Fever Labs   \n",
       "2  10/31/2022, 2:37:04 AM       Minio   \n",
       "3  10/31/2022, 2:37:04 AM   Darwinbox   \n",
       "4  10/31/2022, 2:37:04 AM     Pentera   \n",
       "\n",
       "                                  Crunchbase Url  Last Valuation (Billion $)  \\\n",
       "0       https://www.cbinsights.com/company/esusu                         1.0   \n",
       "1  https://www.cbinsights.com/company/fever-labs                         1.0   \n",
       "2       https://www.cbinsights.com/company/minio                         1.0   \n",
       "3   https://www.cbinsights.com/company/darwinbox                         1.0   \n",
       "4      https://www.cbinsights.com/company/pcysys                         1.0   \n",
       "\n",
       "  Date Joined  Year Joined         City        Country  \\\n",
       "0   1/27/2022         2022     New York  United States   \n",
       "1   1/26/2022         2022     New York  United States   \n",
       "2   1/26/2022         2022    Palo Alto  United States   \n",
       "3   1/25/2022         2022    Hyderabad          India   \n",
       "4   1/11/2022         2022  Petah Tikva         Israel   \n",
       "\n",
       "                       Industry  \\\n",
       "0                       Fintech   \n",
       "1  Internet software & services   \n",
       "2   Data management & analytics   \n",
       "3  Internet software & services   \n",
       "4                Cybersecurity    \n",
       "\n",
       "                                           Investors  Company Website  \\\n",
       "0  [\"Next Play Ventures\",\"Zeal Capital Partners\",...              NaN   \n",
       "1                        [\"Accel\",\"14W\",\"GS Growth\"]              NaN   \n",
       "2  [\"General Catalyst\",\"Nexus Venture Partners\",\"...              NaN   \n",
       "3  [\"Lightspeed India Partners\",\"Sequoia Capital ...              NaN   \n",
       "4   [\"AWZ Ventures\",\"Blackstone\",\"Insight Partners\"]              NaN   \n",
       "\n",
       "                                             summary  token_count  \\\n",
       "0  Esusu has headquarters in New York in United S...           58   \n",
       "1  Fever Labs has headquarters in New York in Uni...           60   \n",
       "2  Minio has headquarters in Palo Alto in United ...           57   \n",
       "3  Darwinbox has headquarters in Hyderabad in Ind...           62   \n",
       "4  Pentera has headquarters in Petah Tikva in Isr...           58   \n",
       "\n",
       "                                           embedding  \n",
       "0  [0.01195491198450327, -0.017717931419610977, -...  \n",
       "1  [0.009171437472105026, 0.01314949057996273, -0...  \n",
       "2  [0.002730059437453747, -0.03737899661064148, 0...  \n",
       "3  [-0.0024771858006715775, -0.024587858468294144...  \n",
       "4  [0.011331121437251568, -0.011193273589015007, ...  "
      ]
     },
     "execution_count": 230,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 231,
   "metadata": {},
   "outputs": [],
   "source": [
    "df.to_csv('unicorns_with_embeddings.csv',index=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Document Similarity \n",
    "\n",
    "We can now take a new string, embed it into a vector, and perform a cosine similarity search against all the vector embeddings in our DataFrame:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 232,
   "metadata": {},
   "outputs": [],
   "source": [
    "prompt = \"What does the company Pentera do and who invested in it?\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 233,
   "metadata": {},
   "outputs": [],
   "source": [
    "prompt_embedding = get_embedding(prompt)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 235,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "# There are other services/programs for larger amount of vectors\n",
    "# Take a look at vector search engines like Pinecone or Weaviate\n",
    "def vector_similarity(vec1,vec2):\n",
    "    \"\"\"\n",
    "    Returns the similarity between two vectors.\n",
    "    \n",
    "    Because OpenAI Embeddings are normalized to length 1, the cosine similarity is the same as the dot product.\n",
    "    \"\"\"\n",
    "    return np.dot(np.array(vec1), np.array(vec2))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 237,
   "metadata": {},
   "outputs": [],
   "source": [
    "df[\"prompt_similarity\"] = df['embedding'].apply(lambda vector: vector_similarity(vector, prompt_embedding))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 239,
   "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>Updated at</th>\n",
       "      <th>Company</th>\n",
       "      <th>Crunchbase Url</th>\n",
       "      <th>Last Valuation (Billion $)</th>\n",
       "      <th>Date Joined</th>\n",
       "      <th>Year Joined</th>\n",
       "      <th>City</th>\n",
       "      <th>Country</th>\n",
       "      <th>Industry</th>\n",
       "      <th>Investors</th>\n",
       "      <th>Company Website</th>\n",
       "      <th>summary</th>\n",
       "      <th>token_count</th>\n",
       "      <th>embedding</th>\n",
       "      <th>prompt_similarity</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>10/31/2022, 2:37:04 AM</td>\n",
       "      <td>Pentera</td>\n",
       "      <td>https://www.cbinsights.com/company/pcysys</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1/11/2022</td>\n",
       "      <td>2022</td>\n",
       "      <td>Petah Tikva</td>\n",
       "      <td>Israel</td>\n",
       "      <td>Cybersecurity</td>\n",
       "      <td>[\"AWZ Ventures\",\"Blackstone\",\"Insight Partners\"]</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Pentera has headquarters in Petah Tikva in Isr...</td>\n",
       "      <td>58</td>\n",
       "      <td>[0.011331121437251568, -0.011193273589015007, ...</td>\n",
       "      <td>0.883279</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>933</th>\n",
       "      <td>10/31/2022, 2:34:02 AM</td>\n",
       "      <td>Pendo</td>\n",
       "      <td>https://www.cbinsights.com/company/pendoio</td>\n",
       "      <td>2.6</td>\n",
       "      <td>10/17/2019</td>\n",
       "      <td>2019</td>\n",
       "      <td>Raleigh</td>\n",
       "      <td>United States</td>\n",
       "      <td>Internet software &amp; services</td>\n",
       "      <td>[\"Contour Venture Partners\",\"Battery Ventures\"...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Pendo has headquarters in Raleigh in United St...</td>\n",
       "      <td>59</td>\n",
       "      <td>[0.01703517511487007, -0.0028536927420645952, ...</td>\n",
       "      <td>0.826227</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>61</th>\n",
       "      <td>10/31/2022, 2:36:13 AM</td>\n",
       "      <td>Perimeter 81</td>\n",
       "      <td>https://www.cbinsights.com/company/perimeter-81</td>\n",
       "      <td>1.0</td>\n",
       "      <td>6/6/2022</td>\n",
       "      <td>2022</td>\n",
       "      <td>Tel Aviv</td>\n",
       "      <td>Israel</td>\n",
       "      <td>Cybersecurity</td>\n",
       "      <td>[\"Insight Partners\",\"Toba Capital\",\"Spring Ven...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Perimeter 81 has headquarters in Tel Aviv in I...</td>\n",
       "      <td>57</td>\n",
       "      <td>[0.006611596792936325, -0.0017099825199693441,...</td>\n",
       "      <td>0.819634</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1183</th>\n",
       "      <td>10/31/2022, 2:33:34 AM</td>\n",
       "      <td>Intarcia Therapeutics</td>\n",
       "      <td>https://www.cbinsights.com/company/intarcia-th...</td>\n",
       "      <td>3.8</td>\n",
       "      <td>4/1/2014</td>\n",
       "      <td>2014</td>\n",
       "      <td>Boston</td>\n",
       "      <td>United States</td>\n",
       "      <td>Health</td>\n",
       "      <td>[\"New Enterprise Associates\",\"New Leaf Venture...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Intarcia Therapeutics has headquarters in Bost...</td>\n",
       "      <td>62</td>\n",
       "      <td>[0.016609707847237587, -0.002032819902524352, ...</td>\n",
       "      <td>0.804336</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>988</th>\n",
       "      <td>10/31/2022, 2:36:23 AM</td>\n",
       "      <td>Momenta</td>\n",
       "      <td>https://www.cbinsights.com/company/momenta</td>\n",
       "      <td>1.0</td>\n",
       "      <td>10/17/2018</td>\n",
       "      <td>2018</td>\n",
       "      <td>Beijing</td>\n",
       "      <td>China</td>\n",
       "      <td>Artificial intelligence</td>\n",
       "      <td>[\"Sinovation Ventures\",\"Tencent Holdings\",\"Seq...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Momenta has headquarters in Beijing in China a...</td>\n",
       "      <td>55</td>\n",
       "      <td>[0.0063818651251494884, -0.03462127968668938, ...</td>\n",
       "      <td>0.803542</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                  Updated at                Company  \\\n",
       "4     10/31/2022, 2:37:04 AM                Pentera   \n",
       "933   10/31/2022, 2:34:02 AM                  Pendo   \n",
       "61    10/31/2022, 2:36:13 AM           Perimeter 81   \n",
       "1183  10/31/2022, 2:33:34 AM  Intarcia Therapeutics   \n",
       "988   10/31/2022, 2:36:23 AM                Momenta   \n",
       "\n",
       "                                         Crunchbase Url  \\\n",
       "4             https://www.cbinsights.com/company/pcysys   \n",
       "933          https://www.cbinsights.com/company/pendoio   \n",
       "61      https://www.cbinsights.com/company/perimeter-81   \n",
       "1183  https://www.cbinsights.com/company/intarcia-th...   \n",
       "988          https://www.cbinsights.com/company/momenta   \n",
       "\n",
       "      Last Valuation (Billion $) Date Joined  Year Joined         City  \\\n",
       "4                            1.0   1/11/2022         2022  Petah Tikva   \n",
       "933                          2.6  10/17/2019         2019      Raleigh   \n",
       "61                           1.0    6/6/2022         2022     Tel Aviv   \n",
       "1183                         3.8    4/1/2014         2014       Boston   \n",
       "988                          1.0  10/17/2018         2018      Beijing   \n",
       "\n",
       "            Country                      Industry  \\\n",
       "4            Israel                Cybersecurity    \n",
       "933   United States  Internet software & services   \n",
       "61           Israel                 Cybersecurity   \n",
       "1183  United States                        Health   \n",
       "988           China       Artificial intelligence   \n",
       "\n",
       "                                              Investors  Company Website  \\\n",
       "4      [\"AWZ Ventures\",\"Blackstone\",\"Insight Partners\"]              NaN   \n",
       "933   [\"Contour Venture Partners\",\"Battery Ventures\"...              NaN   \n",
       "61    [\"Insight Partners\",\"Toba Capital\",\"Spring Ven...              NaN   \n",
       "1183  [\"New Enterprise Associates\",\"New Leaf Venture...              NaN   \n",
       "988   [\"Sinovation Ventures\",\"Tencent Holdings\",\"Seq...              NaN   \n",
       "\n",
       "                                                summary  token_count  \\\n",
       "4     Pentera has headquarters in Petah Tikva in Isr...           58   \n",
       "933   Pendo has headquarters in Raleigh in United St...           59   \n",
       "61    Perimeter 81 has headquarters in Tel Aviv in I...           57   \n",
       "1183  Intarcia Therapeutics has headquarters in Bost...           62   \n",
       "988   Momenta has headquarters in Beijing in China a...           55   \n",
       "\n",
       "                                              embedding  prompt_similarity  \n",
       "4     [0.011331121437251568, -0.011193273589015007, ...           0.883279  \n",
       "933   [0.01703517511487007, -0.0028536927420645952, ...           0.826227  \n",
       "61    [0.006611596792936325, -0.0017099825199693441,...           0.819634  \n",
       "1183  [0.016609707847237587, -0.002032819902524352, ...           0.804336  \n",
       "988   [0.0063818651251494884, -0.03462127968668938, ...           0.803542  "
      ]
     },
     "execution_count": 239,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.sort_values(\"prompt_similarity\", ascending=False).head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now we can easily grab the summary for the most similar embedding, then insert that as context to our GPT request!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 261,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'Pentera has headquarters in Petah Tikva in Israel and is in the field of Cybersecurity . The investors in the company are AWZ Ventures,  Blackstone,  Insight Partners, . You can find more information at https://www.cbinsights.com/company/pcysys'"
      ]
     },
     "execution_count": 261,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Could also use sort_values() with ascending=False, but nlargest should be more performant\n",
    "df.nlargest(1,'prompt_similarity').iloc[0]['summary'] "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Question Answering with Embeddings"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's try inserting the summary to help the model to see if it actually helps:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 262,
   "metadata": {},
   "outputs": [],
   "source": [
    "summary = df.nlargest(1,'prompt_similarity').iloc[0]['summary'] "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 263,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Pentera is a start-up company in the field of Cybersecurity with headquarters in Petah Tikva, Israel. The investors in the company are AWZ Ventures, Blackstone, and Insight Partners.\n"
     ]
    }
   ],
   "source": [
    "prompt = f\"\"\"Only answer the question below if you have 100% certainty of the facts, use the context below to answer.\n",
    "Here is some context:\n",
    "{summary}\n",
    "Q: What does the start-up company Pentera do and who invested in it?\n",
    "A:\"\"\"\n",
    "\n",
    "\n",
    "response = openai.Completion.create(\n",
    "    prompt=prompt,\n",
    "    temperature=0,\n",
    "    max_tokens=500,\n",
    "    model=\"text-davinci-003\"\n",
    ")\n",
    "print(response[\"choices\"][0][\"text\"].strip(\" \\n\"))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Nice! Let's clean this all up by creating a function that wraps all the functions together. Note how this is pretty limited due to our data, but hopefully you can see how this is generalizable to your own data sets and prompts, each situation will be different!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 265,
   "metadata": {},
   "outputs": [],
   "source": [
    "def embed_prompt_lookup():\n",
    "    # initial question\n",
    "    question = input(\"What question do you have about a Unicorn company? \")\n",
    "    # Get embedding\n",
    "    prompt_embedding = get_embedding(question)\n",
    "    # Get prompt similarity with embeddings\n",
    "    # Note how this will overwrite the prompt similarity column each time!\n",
    "    df[\"prompt_similarity\"] = df['embedding'].apply(lambda vector: vector_similarity(vector, prompt_embedding))\n",
    "\n",
    "    # get most similar summary\n",
    "    summary = df.nlargest(1,'prompt_similarity').iloc[0]['summary'] \n",
    "\n",
    "    prompt = f\"\"\"Only answer the question below if you have 100% certainty of the facts, use the context below to answer.\n",
    "            Here is some context:\n",
    "            {summary}\n",
    "            Q: {question}\n",
    "            A:\"\"\"\n",
    "\n",
    "\n",
    "    response = openai.Completion.create(\n",
    "        prompt=prompt,\n",
    "        temperature=0,\n",
    "        max_tokens=500,\n",
    "        model=\"text-davinci-003\"\n",
    "    )\n",
    "    print(response[\"choices\"][0][\"text\"].strip(\" \\n\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 266,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Momenta is a company in the field of Artificial Intelligence with headquarters in Beijing, China.\n"
     ]
    }
   ],
   "source": [
    "embed_prompt_lookup()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.9.12"
  },
  "vscode": {
   "interpreter": {
    "hash": "5455cba45dca8dec86c35d949ecf676492206ef67d9fdf2560e140a882ab099f"
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}