sentiment_via_functions.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "631536ea",
   "metadata": {},
   "outputs": [],
   "source": [
    "import json\n",
    "import openai\n",
    "\n",
    "openai.api_key = \"YOUR_API_KEY\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "f6819b52",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'sentiment': 'negative'}\n"
     ]
    }
   ],
   "source": [
    "#classifier by https://twitter.com/_ScottCondron/status/1670827747684364288\n",
    "def classify(input_string: str) -> str:\n",
    "    functions = [\n",
    "        {\n",
    "            \"name\": \"print_sentiment\",\n",
    "            \"description\": \"A function that prints the given sentiment\",\n",
    "            \"parameters\": {\n",
    "                \"type\": \"object\",\n",
    "                \"properties\": {\n",
    "                    \"sentiment\": {\n",
    "                        \"type\": \"string\",\n",
    "                        \"enum\": [\"positive\", \"negative\", \"neutral\"],\n",
    "                        \"description\": \"The sentiment.\"\n",
    "                    }\n",
    "                },\n",
    "                \"required\": [\"sentiment\"]\n",
    "            }\n",
    "        }\n",
    "    ]\n",
    "\n",
    "    messages = [\n",
    "        {\"role\": \"user\", \"content\": input_string}\n",
    "    ]\n",
    "\n",
    "    response = openai.ChatCompletion.create(\n",
    "        model=\"gpt-3.5-turbo-0613\",\n",
    "        messages=messages,\n",
    "        functions=functions,\n",
    "       # log_level=\"info\"\n",
    "    )\n",
    "\n",
    "    choices = response['choices']\n",
    "    if len(choices) > 0 and 'message' in choices[0]:\n",
    "        function_call = choices[0]['message']['function_call']\n",
    "        argument = json.loads(function_call['arguments'])\n",
    "        return argument\n",
    "\n",
    "    return None\n",
    "\n",
    "\n",
    "\n",
    "input_string = \"I can't stand Twitter!\"\n",
    "result = classify(input_string)\n",
    "print(result)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "3174cff2",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'sentiment_polarity': 'negative', 'aspect_category': 'jury'}\n"
     ]
    }
   ],
   "source": [
    "#added aspect categories, attempt at multiple sentiments per input does not work atm\n",
    "def classify(input_string: str) -> list:\n",
    "    functions = [\n",
    "        {\n",
    "            \"name\": \"print_sentiment\",\n",
    "            \"description\": \"Print all the given sentiments and aspect categories, may have multiple sentiments per input string\",\n",
    "            \"parameters\": {\n",
    "                \"type\": \"object\",\n",
    "                \"properties\": {\n",
    "                    \"sentiment_polarity\": {\n",
    "                        \"type\": \"string\",\n",
    "                        \"enum\": [\"positive\", \"negative\", \"neutral\"],\n",
    "                        \"description\": \"The sentiment.\"\n",
    "                    },\n",
    "                    \"aspect_category\": {\n",
    "                        \"type\": \"string\",\n",
    "                        \"enum\": [\"contender\", \"jury\", \"event\"],\n",
    "                        \"description\": \"The aspect category.\"\n",
    "                    }\n",
    "                },\n",
    "                \"required\": [\"sentiment_polarity\", \"aspect_category\"]\n",
    "            }\n",
    "        }\n",
    "    ]\n",
    "\n",
    "    messages = [\n",
    "        {\"role\": \"user\", \"content\": input_string}\n",
    "    ]\n",
    "\n",
    "    response = openai.ChatCompletion.create(\n",
    "        model=\"gpt-3.5-turbo-0613\",\n",
    "        messages=messages,\n",
    "        functions=functions,\n",
    "        # log_level=\"info\"\n",
    "    )\n",
    "\n",
    "    results = []\n",
    "    choices = response['choices']\n",
    "    for choice in choices:\n",
    "        if 'message' in choice:\n",
    "            function_call = choice['message']['function_call']\n",
    "            argument = json.loads(function_call['arguments'])\n",
    "            results.append(argument)\n",
    "\n",
    "    return results\n",
    "\n",
    "\n",
    "input_string = \"I can't stand this jury member, but I really like this contender!\"\n",
    "results = classify(input_string)\n",
    "for result in results:\n",
    "    print(result)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "c1d2d5b2",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{\n",
      "  \"id\": \"chatcmpl-7TSlZ4CXgJtNfyvHHzlIFjeUxNANk\",\n",
      "  \"object\": \"chat.completion\",\n",
      "  \"created\": 1687256445,\n",
      "  \"model\": \"gpt-3.5-turbo-0613\",\n",
      "  \"choices\": [\n",
      "    {\n",
      "      \"index\": 0,\n",
      "      \"message\": {\n",
      "        \"role\": \"assistant\",\n",
      "        \"content\": null,\n",
      "        \"function_call\": {\n",
      "          \"name\": \"print_sentiments\",\n",
      "          \"arguments\": \"{\\n  \\\"sentiment_polarity\\\": \\\"positive\\\",\\n  \\\"aspect_category\\\": \\\"contender\\\"\\n}\"\n",
      "        }\n",
      "      },\n",
      "      \"finish_reason\": \"function_call\"\n",
      "    }\n",
      "  ],\n",
      "  \"usage\": {\n",
      "    \"prompt_tokens\": 105,\n",
      "    \"completion_tokens\": 28,\n",
      "    \"total_tokens\": 133\n",
      "  }\n",
      "}\n",
      "{'sentiment_polarity': 'positive', 'aspect_category': 'contender'}\n"
     ]
    }
   ],
   "source": [
    "#verbose version to see what it does\n",
    "def classify(input_string: str) -> list:\n",
    "    functions = [\n",
    "        {\n",
    "            \"name\": \"print_sentiments\",\n",
    "            \"description\": \"A function that prints all the given sentiment polarities and aspect categories\",\n",
    "            \"parameters\": {\n",
    "                \"type\": \"object\",\n",
    "                \"properties\": {\n",
    "                    \"sentiment_polarity\": {\n",
    "                        \"type\": \"string\",\n",
    "                        \"enum\": [\"positive\", \"negative\", \"neutral\"],\n",
    "                        \"description\": \"The sentiment.\"\n",
    "                    },\n",
    "                    \"aspect_category\": {\n",
    "                        \"type\": \"string\",\n",
    "                        \"enum\": [\"contender\", \"jury\", \"event\"],\n",
    "                        \"description\": \"The aspect category.\"\n",
    "                    }\n",
    "                },\n",
    "                \"required\": [\"sentiment_polarity\", \"aspect_category\"]\n",
    "            }\n",
    "        }\n",
    "    ]\n",
    "\n",
    "    messages = [\n",
    "        {\"role\": \"user\", \"content\": input_string}\n",
    "    ]\n",
    "\n",
    "    response = openai.ChatCompletion.create(\n",
    "        model=\"gpt-3.5-turbo-0613\",\n",
    "        messages=messages,\n",
    "        functions=functions,\n",
    " #       log_level=\"info\"\n",
    "    )\n",
    "\n",
    "    results = []\n",
    "    choices = response['choices']\n",
    "    for choice in choices:\n",
    "        if 'message' in choice:\n",
    "            function_call = choice['message']['function_call']\n",
    "            argument = json.loads(function_call['arguments'])\n",
    "            results.append(argument)\n",
    "    \n",
    "    print(response)\n",
    "    \n",
    "    return results\n",
    "\n",
    "\n",
    "input_string = \"I can't stand this jury member, but I really like this contender!\"\n",
    "results = classify(input_string)\n",
    "for result in results:\n",
    "    print(result)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "f01149ac",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'mad'"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#via logits in MarvinAI https://twitter.com/AAAzzam/status/1669753722828730378\n",
    "import tiktoken\n",
    "import openai\n",
    "\n",
    "openai.ChatCompletion.create(\n",
    "    model='gpt-3.5-turbo',\n",
    "    messages=[\n",
    "        {'role': 'system', 'content': 'You respond with whether the user input is happy, sad, or mad'},\n",
    "        {'role': 'user', 'content': \"I can't stand Twitter!\"}\n",
    "    ],\n",
    "    logit_bias={\n",
    "        '57621': 100,  # 57621 is the token for 'happy'\n",
    "        '83214': 100,  # 83214 is the token for 'sad'\n",
    "        '20920': 100   # 20920 is the token for 'mad'\n",
    "    },\n",
    "    max_tokens=1,  # You can only respond with a single token\n",
    "    temperature=0\n",
    ").choices[0].message.content\n"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}