langchain/cookbook/multi_modal_output_agent.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "cd835d40",
   "metadata": {},
   "source": [
    "# Multi-modal outputs: Image & Text"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "fa88e03a",
   "metadata": {},
   "source": [
    "This notebook shows how non-text producing tools can be used to create multi-modal agents.\n",
    "\n",
    "This example is limited to text and image outputs and uses UUIDs to transfer content across tools and agents. \n",
    "\n",
    "This example uses Steamship to generate and store generated images. Generated are auth protected by default. \n",
    "\n",
    "You can get your Steamship api key here: https://steamship.com/account/api"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "0653da01",
   "metadata": {},
   "outputs": [],
   "source": [
    "import re\n",
    "\n",
    "from IPython.display import Image\n",
    "from steamship import Block, Steamship"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "f6933033",
   "metadata": {},
   "outputs": [],
   "source": [
    "from langchain.agents import AgentType, initialize_agent\n",
    "from langchain.llms import OpenAI\n",
    "from langchain.tools import SteamshipImageGenerationTool"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "71e51e53",
   "metadata": {},
   "outputs": [],
   "source": [
    "llm = OpenAI(temperature=0)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a9fc769d",
   "metadata": {},
   "source": [
    "## Dall-E "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "cd177dfe",
   "metadata": {},
   "outputs": [],
   "source": [
    "tools = [SteamshipImageGenerationTool(model_name=\"dall-e\")]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "c71b1e46",
   "metadata": {},
   "outputs": [],
   "source": [
    "mrkl = initialize_agent(\n",
    "    tools, llm, agent=AgentType.ZERO_SHOT_REACT_DESCRIPTION, verbose=True\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "603aeb9a",
   "metadata": {},
   "outputs": [],
   "source": [
    "output = mrkl.run(\"How would you visualize a parot playing soccer?\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "25eb4efe",
   "metadata": {},
   "outputs": [],
   "source": [
    "def show_output(output):\n",
    "    \"\"\"Display the multi-modal output from the agent.\"\"\"\n",
    "    UUID_PATTERN = re.compile(\n",
    "        r\"([0-9A-Za-z]{8}-[0-9A-Za-z]{4}-[0-9A-Za-z]{4}-[0-9A-Za-z]{4}-[0-9A-Za-z]{12})\"\n",
    "    )\n",
    "\n",
    "    outputs = UUID_PATTERN.split(output)\n",
    "    outputs = [\n",
    "        re.sub(r\"^\\W+\", \"\", el) for el in outputs\n",
    "    ]  # Clean trailing and leading non-word characters\n",
    "\n",
    "    for output in outputs:\n",
    "        maybe_block_id = UUID_PATTERN.search(output)\n",
    "        if maybe_block_id:\n",
    "            display(Image(Block.get(Steamship(), _id=maybe_block_id.group()).raw()))\n",
    "        else:\n",
    "            print(output, end=\"\\n\\n\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e247b2c4",
   "metadata": {},
   "source": [
    "## StableDiffusion "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "315025e7",
   "metadata": {},
   "outputs": [],
   "source": [
    "tools = [SteamshipImageGenerationTool(model_name=\"stable-diffusion\")]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "7930064a",
   "metadata": {},
   "outputs": [],
   "source": [
    "mrkl = initialize_agent(\n",
    "    tools, llm, agent=AgentType.ZERO_SHOT_REACT_DESCRIPTION, verbose=True\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "611a833d",
   "metadata": {},
   "outputs": [],
   "source": [
    "output = mrkl.run(\"How would you visualize a parot playing soccer?\")"
   ]
  }
 ],
 "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": "3.11.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}