examples/notebooks/OpenAIChatExperiment.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "0a13ddc8",
   "metadata": {},
   "source": [
    "# OpenAI Chat Experiment Example"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "623f0cfe",
   "metadata": {},
   "source": [
    "## Installations"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "885dabeb",
   "metadata": {},
   "outputs": [],
   "source": [
    "# !pip install --quiet --force-reinstall prompttools"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2eac35f8",
   "metadata": {},
   "source": [
    "## Setup imports and API keys"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5edba05a",
   "metadata": {},
   "source": [
    "First, we'll need to set our API keys. If we are in DEBUG mode, we don't need to use a real OpenAI key, so for now we'll set them to empty strings."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "ed4e635e",
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "\n",
    "os.environ[\"DEBUG\"] = \"1\"  # Set this to \"\" to call OpenAI's API\n",
    "os.environ[\"OPENAI_API_KEY\"] = \"\"  # Insert your key here"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "842f1e47",
   "metadata": {},
   "source": [
    "Then we'll import the relevant `prompttools` modules to setup our experiment."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "beaa70a1",
   "metadata": {},
   "outputs": [],
   "source": [
    "from typing import Dict, List\n",
    "from prompttools.experiment import OpenAIChatExperiment"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "622dea9a",
   "metadata": {},
   "source": [
    "## Run an experiment"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3babfe5a",
   "metadata": {},
   "source": [
    "Next, we create our test inputs. We can iterate over models, inputs, and configurations like temperature."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "9114cfbf",
   "metadata": {},
   "outputs": [],
   "source": [
    "models = [\"gpt-3.5-turbo\", \"gpt-3.5-turbo-0613\"]  # You can also use a fine-tuned model here, e.g. [\"ft:gpt-3.5-turbo:org_id\"]\n",
    "messages = [\n",
    "    [\n",
    "        {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n",
    "        {\"role\": \"user\", \"content\": \"Who was the first president?\"},\n",
    "    ]\n",
    "]\n",
    "temperatures = [0.0, 1.0]\n",
    "# You can add more parameters that you'd like to test here.\n",
    "\n",
    "experiment = OpenAIChatExperiment(models, messages, temperature=temperatures)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f3fa5450",
   "metadata": {},
   "source": [
    "We can then run the experiment to get results."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "83b33130",
   "metadata": {},
   "outputs": [],
   "source": [
    "experiment.run()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "7598332b",
   "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>model</th>\n",
       "      <th>messages</th>\n",
       "      <th>temperature</th>\n",
       "      <th>response</th>\n",
       "      <th>response_usage</th>\n",
       "      <th>latency</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>gpt-3.5-turbo</td>\n",
       "      <td>[{'role': 'system', 'content': 'You are a helpful assistant.'}, {'role': 'user', 'content': 'Who was the first president?'}]</td>\n",
       "      <td>0.0</td>\n",
       "      <td>George Washington</td>\n",
       "      <td>{'completion_tokens': 18, 'prompt_tokens': 57, 'total_tokens': 75}</td>\n",
       "      <td>0.000006</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>gpt-3.5-turbo</td>\n",
       "      <td>[{'role': 'system', 'content': 'You are a helpful assistant.'}, {'role': 'user', 'content': 'Who was the first president?'}]</td>\n",
       "      <td>1.0</td>\n",
       "      <td>George Washington</td>\n",
       "      <td>{'completion_tokens': 18, 'prompt_tokens': 57, 'total_tokens': 75}</td>\n",
       "      <td>0.000005</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>gpt-3.5-turbo-0613</td>\n",
       "      <td>[{'role': 'system', 'content': 'You are a helpful assistant.'}, {'role': 'user', 'content': 'Who was the first president?'}]</td>\n",
       "      <td>0.0</td>\n",
       "      <td>George Washington</td>\n",
       "      <td>{'completion_tokens': 18, 'prompt_tokens': 57, 'total_tokens': 75}</td>\n",
       "      <td>0.000003</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>gpt-3.5-turbo-0613</td>\n",
       "      <td>[{'role': 'system', 'content': 'You are a helpful assistant.'}, {'role': 'user', 'content': 'Who was the first president?'}]</td>\n",
       "      <td>1.0</td>\n",
       "      <td>George Washington</td>\n",
       "      <td>{'completion_tokens': 18, 'prompt_tokens': 57, 'total_tokens': 75}</td>\n",
       "      <td>0.000002</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                model  \\\n",
       "0  gpt-3.5-turbo        \n",
       "1  gpt-3.5-turbo        \n",
       "2  gpt-3.5-turbo-0613   \n",
       "3  gpt-3.5-turbo-0613   \n",
       "\n",
       "                                                                                                                       messages  \\\n",
       "0  [{'role': 'system', 'content': 'You are a helpful assistant.'}, {'role': 'user', 'content': 'Who was the first president?'}]   \n",
       "1  [{'role': 'system', 'content': 'You are a helpful assistant.'}, {'role': 'user', 'content': 'Who was the first president?'}]   \n",
       "2  [{'role': 'system', 'content': 'You are a helpful assistant.'}, {'role': 'user', 'content': 'Who was the first president?'}]   \n",
       "3  [{'role': 'system', 'content': 'You are a helpful assistant.'}, {'role': 'user', 'content': 'Who was the first president?'}]   \n",
       "\n",
       "   temperature           response  \\\n",
       "0  0.0          George Washington   \n",
       "1  1.0          George Washington   \n",
       "2  0.0          George Washington   \n",
       "3  1.0          George Washington   \n",
       "\n",
       "                                                       response_usage  \\\n",
       "0  {'completion_tokens': 18, 'prompt_tokens': 57, 'total_tokens': 75}   \n",
       "1  {'completion_tokens': 18, 'prompt_tokens': 57, 'total_tokens': 75}   \n",
       "2  {'completion_tokens': 18, 'prompt_tokens': 57, 'total_tokens': 75}   \n",
       "3  {'completion_tokens': 18, 'prompt_tokens': 57, 'total_tokens': 75}   \n",
       "\n",
       "    latency  \n",
       "0  0.000006  \n",
       "1  0.000005  \n",
       "2  0.000003  \n",
       "3  0.000002  "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "experiment.visualize()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "266c13eb",
   "metadata": {},
   "source": [
    "## Evaluate the model response"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "bebb8023",
   "metadata": {},
   "source": [
    "To evaluate the results, we'll define an eval function. We can use semantic distance to check if the model's response is similar to our expected output.\n",
    "\n",
    "Since we are using semantic similarity, you need to have the library `sentence_transformers` installed."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "d861ab10",
   "metadata": {},
   "outputs": [],
   "source": [
    "# !pip install --quiet sentence_transformers"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "8ddbb951",
   "metadata": {},
   "outputs": [],
   "source": [
    "from prompttools.utils import similarity"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "e80dfeec",
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "experiment.evaluate(\"similar_to_expected\", similarity.semantic_similarity, expected=[\"George Washington\"] * 4)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "974d6065",
   "metadata": {},
   "source": [
    "Finally, we can visualize the results."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "4d09c18e",
   "metadata": {
    "scrolled": true
   },
   "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>model</th>\n",
       "      <th>messages</th>\n",
       "      <th>temperature</th>\n",
       "      <th>response</th>\n",
       "      <th>response_usage</th>\n",
       "      <th>latency</th>\n",
       "      <th>similar_to_expected</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>gpt-3.5-turbo</td>\n",
       "      <td>[{'role': 'system', 'content': 'You are a helpful assistant.'}, {'role': 'user', 'content': 'Who was the first president?'}]</td>\n",
       "      <td>0.0</td>\n",
       "      <td>George Washington</td>\n",
       "      <td>{'completion_tokens': 18, 'prompt_tokens': 57, 'total_tokens': 75}</td>\n",
       "      <td>0.000006</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>gpt-3.5-turbo</td>\n",
       "      <td>[{'role': 'system', 'content': 'You are a helpful assistant.'}, {'role': 'user', 'content': 'Who was the first president?'}]</td>\n",
       "      <td>1.0</td>\n",
       "      <td>George Washington</td>\n",
       "      <td>{'completion_tokens': 18, 'prompt_tokens': 57, 'total_tokens': 75}</td>\n",
       "      <td>0.000005</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>gpt-3.5-turbo-0613</td>\n",
       "      <td>[{'role': 'system', 'content': 'You are a helpful assistant.'}, {'role': 'user', 'content': 'Who was the first president?'}]</td>\n",
       "      <td>0.0</td>\n",
       "      <td>George Washington</td>\n",
       "      <td>{'completion_tokens': 18, 'prompt_tokens': 57, 'total_tokens': 75}</td>\n",
       "      <td>0.000003</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>gpt-3.5-turbo-0613</td>\n",
       "      <td>[{'role': 'system', 'content': 'You are a helpful assistant.'}, {'role': 'user', 'content': 'Who was the first president?'}]</td>\n",
       "      <td>1.0</td>\n",
       "      <td>George Washington</td>\n",
       "      <td>{'completion_tokens': 18, 'prompt_tokens': 57, 'total_tokens': 75}</td>\n",
       "      <td>0.000002</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                model  \\\n",
       "0  gpt-3.5-turbo        \n",
       "1  gpt-3.5-turbo        \n",
       "2  gpt-3.5-turbo-0613   \n",
       "3  gpt-3.5-turbo-0613   \n",
       "\n",
       "                                                                                                                       messages  \\\n",
       "0  [{'role': 'system', 'content': 'You are a helpful assistant.'}, {'role': 'user', 'content': 'Who was the first president?'}]   \n",
       "1  [{'role': 'system', 'content': 'You are a helpful assistant.'}, {'role': 'user', 'content': 'Who was the first president?'}]   \n",
       "2  [{'role': 'system', 'content': 'You are a helpful assistant.'}, {'role': 'user', 'content': 'Who was the first president?'}]   \n",
       "3  [{'role': 'system', 'content': 'You are a helpful assistant.'}, {'role': 'user', 'content': 'Who was the first president?'}]   \n",
       "\n",
       "   temperature           response  \\\n",
       "0  0.0          George Washington   \n",
       "1  1.0          George Washington   \n",
       "2  0.0          George Washington   \n",
       "3  1.0          George Washington   \n",
       "\n",
       "                                                       response_usage  \\\n",
       "0  {'completion_tokens': 18, 'prompt_tokens': 57, 'total_tokens': 75}   \n",
       "1  {'completion_tokens': 18, 'prompt_tokens': 57, 'total_tokens': 75}   \n",
       "2  {'completion_tokens': 18, 'prompt_tokens': 57, 'total_tokens': 75}   \n",
       "3  {'completion_tokens': 18, 'prompt_tokens': 57, 'total_tokens': 75}   \n",
       "\n",
       "    latency  similar_to_expected  \n",
       "0  0.000006  1.0                  \n",
       "1  0.000005  1.0                  \n",
       "2  0.000003  1.0                  \n",
       "3  0.000002  1.0                  "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "experiment.visualize()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b6802b34",
   "metadata": {},
   "source": [
    "## Saving to Remote"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "fb74a9a2",
   "metadata": {},
   "source": [
    "For users interested to manage their experiments in a hosted environment. We provide APIs to securely store your experiments in our backend, from which you can load experiments for re-run or share with your team. \n",
    "\n",
    "If you are interested in our hosted solution, please reach out to us."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "d30cd8ad",
   "metadata": {},
   "outputs": [],
   "source": [
    "os.environ[\"HEGELAI_API_KEY\"] = \"\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "c084bb43",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "id": "19269a04",
   "metadata": {},
   "source": [
    "## Aggregate Results"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e5626394",
   "metadata": {},
   "source": [
    "You can optionally visualize the results with the following command"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "d0007a1f",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1200x800 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "experiment.aggregate(metric_name=\"latency\", column_name=\"model\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "78c0c9f6",
   "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": "3.11.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}