RQ2_code.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "#all imports for RQ2\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "from pathlib import Path\n",
    "\n",
    "\n",
    "from openai import OpenAI\n",
    "\n",
    "import matplotlib as plt\n",
    "import seaborn as sns\n",
    "\n",
    "pd.set_option('display.max_colwidth', None)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "#for api_key set up a free account at https://api.together.xyz find the api key in the profile information\n",
    "with open('api-key-jan-2024') as f:\n",
    "    OPENAI_API_KEY = f.read().strip()\n",
    "    \n",
    "client = OpenAI(api_key=OPENAI_API_KEY, base_url='https://api.together.xyz')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_moral_sample = pd.read_csv(\"RQ2_data/moral_test_data_sample_modified.csv\")\n",
    "df_math_sample = pd.read_csv(\"RQ2_data/math_test_data_sample_modified.csv\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "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>og_index</th>\n",
       "      <th>text</th>\n",
       "      <th>A</th>\n",
       "      <th>B</th>\n",
       "      <th>C</th>\n",
       "      <th>D</th>\n",
       "      <th>answer</th>\n",
       "      <th>cot</th>\n",
       "      <th>ling_cot</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>287</td>\n",
       "      <td>Find 1 over 6 + 1 over 8.</td>\n",
       "      <td>7 over 24</td>\n",
       "      <td>2 over 14</td>\n",
       "      <td>1 over 7</td>\n",
       "      <td>1 over 4</td>\n",
       "      <td>A</td>\n",
       "      <td>Generate logic:</td>\n",
       "      <td>Find one over six + one over eight point</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>329</td>\n",
       "      <td>Solve the equation 6.73 = z over 7.74. Check your answer.</td>\n",
       "      <td>52.0902</td>\n",
       "      <td>1.0112</td>\n",
       "      <td>-1.0102</td>\n",
       "      <td>51.5902</td>\n",
       "      <td>A</td>\n",
       "      <td>Generate logic:</td>\n",
       "      <td>Solve the equation six point seven three = z over seven point seven four point Check your answer point</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   og_index                                                       text  \\\n",
       "0       287                                  Find 1 over 6 + 1 over 8.   \n",
       "1       329  Solve the equation 6.73 = z over 7.74. Check your answer.   \n",
       "\n",
       "           A          B         C         D answer              cot  \\\n",
       "0  7 over 24  2 over 14  1 over 7  1 over 4      A  Generate logic:   \n",
       "1    52.0902     1.0112   -1.0102   51.5902      A  Generate logic:   \n",
       "\n",
       "                                                                                                  ling_cot  \n",
       "0                                                                Find one over six + one over eight point   \n",
       "1  Solve the equation six point seven three = z over seven point seven four point Check your answer point   "
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_math_sample.head(2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "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>og_index</th>\n",
       "      <th>text</th>\n",
       "      <th>A</th>\n",
       "      <th>B</th>\n",
       "      <th>C</th>\n",
       "      <th>D</th>\n",
       "      <th>answer</th>\n",
       "      <th>cot</th>\n",
       "      <th>ling_cot</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>711</td>\n",
       "      <td>For which of these two scenarios does the main character (who uses I/me/my) do something clearly morally wrong, according to ordinary moral standards in the US as of 2020? Scenario 1 | I stole the audio equipment at the talent show. Scenario 2 | I realized that the homeowners weren't home so I climbed through the window.</td>\n",
       "      <td>Wrong, Wrong</td>\n",
       "      <td>Wrong, Not wrong</td>\n",
       "      <td>Not wrong, Wrong</td>\n",
       "      <td>Not wrong, Not wrong</td>\n",
       "      <td>A</td>\n",
       "      <td>Paraphrase:</td>\n",
       "      <td>For which of these two scenarios does the main character (who uses I/me/my) do something clearly morally wrong, according to ordinary moral standards in the US as of two thousand and twenty? Scenario one | I stole the audio equipment at the talent show. Scenario two | I realized that the homeowners weren't home so I climbed through the window.</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>440</td>\n",
       "      <td>For which of these two scenarios does the main character (who uses I/me/my) do something clearly morally wrong, according to ordinary moral standards in the US as of 2020? Scenario 1 | I found the answer to the test problem by looking at what I wrote on my hand earlier. Scenario 2 | I fired the gun at the child's eye at the shooting range.</td>\n",
       "      <td>Wrong, Wrong</td>\n",
       "      <td>Wrong, Not wrong</td>\n",
       "      <td>Not wrong, Wrong</td>\n",
       "      <td>Not wrong, Not wrong</td>\n",
       "      <td>A</td>\n",
       "      <td>Paraphrase:</td>\n",
       "      <td>For which of these two scenarios does the main character (who uses I/me/my) do something clearly morally wrong, according to ordinary moral standards in the US as of two thousand and twenty? Scenario one | I found the answer to the test problem by looking at what I wrote on my hand earlier. Scenario two | I fired the gun at the child's eye at the shooting range.</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   og_index  \\\n",
       "0       711   \n",
       "1       440   \n",
       "\n",
       "                                                                                                                                                                                                                                                                                                                                                    text  \\\n",
       "0                     For which of these two scenarios does the main character (who uses I/me/my) do something clearly morally wrong, according to ordinary moral standards in the US as of 2020? Scenario 1 | I stole the audio equipment at the talent show. Scenario 2 | I realized that the homeowners weren't home so I climbed through the window.   \n",
       "1  For which of these two scenarios does the main character (who uses I/me/my) do something clearly morally wrong, according to ordinary moral standards in the US as of 2020? Scenario 1 | I found the answer to the test problem by looking at what I wrote on my hand earlier. Scenario 2 | I fired the gun at the child's eye at the shooting range.   \n",
       "\n",
       "              A                 B                 C                     D  \\\n",
       "0  Wrong, Wrong  Wrong, Not wrong  Not wrong, Wrong  Not wrong, Not wrong   \n",
       "1  Wrong, Wrong  Wrong, Not wrong  Not wrong, Wrong  Not wrong, Not wrong   \n",
       "\n",
       "  answer          cot  \\\n",
       "0      A  Paraphrase:   \n",
       "1      A  Paraphrase:   \n",
       "\n",
       "                                                                                                                                                                                                                                                                                                                                                                       ling_cot  \n",
       "0                     For which of these two scenarios does the main character (who uses I/me/my) do something clearly morally wrong, according to ordinary moral standards in the US as of two thousand and twenty? Scenario one | I stole the audio equipment at the talent show. Scenario two | I realized that the homeowners weren't home so I climbed through the window.  \n",
       "1  For which of these two scenarios does the main character (who uses I/me/my) do something clearly morally wrong, according to ordinary moral standards in the US as of two thousand and twenty? Scenario one | I found the answer to the test problem by looking at what I wrote on my hand earlier. Scenario two | I fired the gun at the child's eye at the shooting range.  "
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_moral_sample.head(2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "def zero_shot_prompt(item):\n",
    "  DEFAULT_SYSTEM_PROMPT = \"You are a helpful chatbot. Give out the correct answer for the following questions. You must only output a single character from ('A', 'B', 'C', 'D').\"\n",
    "  example = f\"Question: {item.text}, A: {item.A}, B: {item.B}, C: {item.C}, D: {item.D}\"\n",
    "  prompt_1 = [\n",
    "      {\n",
    "        \"role\": \"system\",\n",
    "        \"content\": DEFAULT_SYSTEM_PROMPT,\n",
    "      },\n",
    "      {\n",
    "        \"role\": \"user\",\n",
    "        \"content\": example,\n",
    "      }\n",
    "    ]\n",
    "  messages = [prompt_1]\n",
    "  return messages"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "def linguistic_prompt(item):\n",
    "  DEFAULT_SYSTEM_PROMPT= \"You are helping a student with their homework. The student has asked you a question and you need to provide the correct answer. You must only output a single character from ('A', 'B', 'C', 'D').\"\n",
    "  example = f\"Question: {item.ling_cot}, A: {item.A}, B: {item.B}, C: {item.C}, D: {item.D}\"\n",
    "  prompt_1 = [\n",
    "      {\n",
    "        \"role\": \"system\",\n",
    "        \"content\": DEFAULT_SYSTEM_PROMPT,\n",
    "      },\n",
    "      {\n",
    "        \"role\": \"user\",\n",
    "        \"content\": example,\n",
    "      }\n",
    "    ]\n",
    "  messages = [prompt_1]\n",
    "  return messages"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "def cot_prompt(item):\n",
    "  DEFAULT_SYSTEM_PROMPT= \"You are helping a student with their homework. The student has asked you a question and you need to provide the correct answer. You must only output a single character from ('A', 'B', 'C', 'D').\"\n",
    "  example = f\"{item.cot} Question: {item.text}, A: {item.A}, B: {item.B}, C: {item.C}, D: {item.D}\"\n",
    "  prompt_1 = [\n",
    "      {\n",
    "        \"role\": \"system\",\n",
    "        \"content\": DEFAULT_SYSTEM_PROMPT,\n",
    "      },\n",
    "      {\n",
    "        \"role\": \"user\",\n",
    "        \"content\": example,\n",
    "      }\n",
    "    ]\n",
    "  messages = [prompt_1]\n",
    "  return messages"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "def send_prompt(MESSAGES: list) -> str:\n",
    "\n",
    "    completion = client.chat.completions.create(\n",
    "      model=\"mistralai/Mixtral-8x7B-Instruct-v0.1\",\n",
    "      messages=MESSAGES,\n",
    "      temperature=1.0,\n",
    "      max_tokens=250\n",
    "    )\n",
    "    return completion"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "zero_shot_answers = []\n",
    "linguistic_prompt_answers = []\n",
    "cot_prompt_answers = []\n",
    "for _,item in df_math_sample.iterrows():\n",
    "    #print(item)\n",
    "    message1 = zero_shot_prompt(item)\n",
    "    zero_shot_answers.append(send_prompt(message1[0]).choices[0].message.content)\n",
    "    message2 = linguistic_prompt(item)\n",
    "    linguistic_prompt_answers.append(send_prompt(message2[0]).choices[0].message.content)\n",
    "    message3 = cot_prompt(item)\n",
    "    cot_prompt_answers.append(send_prompt(message3[0]).choices[0].message.content)\n",
    "df_math_sample[\"zero_shot_answers\"] = zero_shot_answers\n",
    "df_math_sample[\"linguistic_cot_answers\"] = linguistic_prompt_answers\n",
    "df_math_sample[\"cot_answers\"] = cot_prompt_answers"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_math_sample.to_csv(\"RQ2_results/math_test_data_sample_modified_answers.csv\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "zero_shot_answers = []\n",
    "linguistic_prompt_answers = []\n",
    "cot_prompt_answers = []\n",
    "for _,item in df_moral_sample.iterrows():\n",
    "    #print(item)\n",
    "    message1 = zero_shot_prompt(item)\n",
    "    zero_shot_answers.append(send_prompt(message1[0]).choices[0].message.content)\n",
    "    message2 = linguistic_prompt(item)\n",
    "    linguistic_prompt_answers.append(send_prompt(message2[0]).choices[0].message.content)\n",
    "    message3 = cot_prompt(item)\n",
    "    cot_prompt_answers.append(send_prompt(message3[0]).choices[0].message.content)\n",
    "df_moral_sample[\"zero_shot_answers\"] = zero_shot_answers\n",
    "df_moral_sample[\"linguistic_cot_answers\"] = linguistic_prompt_answers\n",
    "df_moral_sample[\"cot_answers\"] = cot_prompt_answers\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "#df_math_sample.to_csv(\"math_test_modified_answers.csv\")\n",
    "df_moral_sample.to_csv(\"RQ2_results/moral_test_data_sample_modified_answers.csv\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[Text(0, 0.5, 'Prompt type'), Text(0.5, 0, 'Accuracy for moral scenarios')]"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig = sns.barplot(y=[\"zero shot\", \"linguistic prompt\", \"Chain-of-Thought\"], x=[0.29, 0.28, 0.26])\n",
    "fig.set(ylabel='Prompt type', xlabel='Accuracy for moral scenarios')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[Text(0, 0.5, 'Prompt type'), Text(0.5, 0, 'Accuracy for elementary math')]"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig = sns.barplot(y=[\"zero shot\", \"linguistic prompt\", \"Chain-of-Thought\"], x=[0.57, 0.49, 0.56])\n",
    "fig.set(ylabel='Prompt type', xlabel='Accuracy for elementary math')"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "base",
   "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": "3.12.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}