langchain/cookbook/autogpt/autogpt.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "14f8b67b",
   "metadata": {},
   "source": [
    "# AutoGPT\n",
    "\n",
    "Implementation of https://github.com/Significant-Gravitas/Auto-GPT but with LangChain primitives (LLMs, PromptTemplates, VectorStores, Embeddings, Tools)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "192496a7",
   "metadata": {},
   "source": [
    "## Set up tools\n",
    "\n",
    "We'll set up an AutoGPT with a search tool, and write-file tool, and a read-file tool"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "7c2c9b54",
   "metadata": {},
   "outputs": [],
   "source": [
    "from langchain.utilities import SerpAPIWrapper\n",
    "from langchain.agents import Tool\n",
    "from langchain.tools.file_management.write import WriteFileTool\n",
    "from langchain.tools.file_management.read import ReadFileTool\n",
    "\n",
    "search = SerpAPIWrapper()\n",
    "tools = [\n",
    "    Tool(\n",
    "        name=\"search\",\n",
    "        func=search.run,\n",
    "        description=\"useful for when you need to answer questions about current events. You should ask targeted questions\",\n",
    "    ),\n",
    "    WriteFileTool(),\n",
    "    ReadFileTool(),\n",
    "]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8e39ee28",
   "metadata": {},
   "source": [
    "## Set up memory\n",
    "\n",
    "The memory here is used for the agents intermediate steps"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "72bc204d",
   "metadata": {},
   "outputs": [],
   "source": [
    "from langchain.vectorstores import FAISS\n",
    "from langchain.docstore import InMemoryDocstore\n",
    "from langchain.embeddings import OpenAIEmbeddings"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "1df7b724",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Define your embedding model\n",
    "embeddings_model = OpenAIEmbeddings()\n",
    "# Initialize the vectorstore as empty\n",
    "import faiss\n",
    "\n",
    "embedding_size = 1536\n",
    "index = faiss.IndexFlatL2(embedding_size)\n",
    "vectorstore = FAISS(embeddings_model.embed_query, index, InMemoryDocstore({}), {})"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e40fd657",
   "metadata": {},
   "source": [
    "## Setup model and AutoGPT\n",
    "\n",
    "Initialize everything! We will use ChatOpenAI model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "3393bc23",
   "metadata": {},
   "outputs": [],
   "source": [
    "from langchain_experimental.autonomous_agents import AutoGPT\n",
    "from langchain.chat_models import ChatOpenAI"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "709c08c2",
   "metadata": {},
   "outputs": [],
   "source": [
    "agent = AutoGPT.from_llm_and_tools(\n",
    "    ai_name=\"Tom\",\n",
    "    ai_role=\"Assistant\",\n",
    "    tools=tools,\n",
    "    llm=ChatOpenAI(temperature=0),\n",
    "    memory=vectorstore.as_retriever(),\n",
    ")\n",
    "# Set verbose to be true\n",
    "agent.chain.verbose = True"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f0f208d9",
   "metadata": {
    "collapsed": false
   },
   "source": [
    "## Run an example\n",
    "\n",
    "Here we will make it write a weather report for SF"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "d119d788",
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "agent.run([\"write a weather report for SF today\"])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f13f8322",
   "metadata": {
    "collapsed": false
   },
   "source": [
    "## Chat History Memory\n",
    "\n",
    "In addition to the memory that holds the agent immediate steps, we also have a chat history memory. By default, the agent will use 'ChatMessageHistory' and it can be changed. This is useful when you want to use a different type of memory for example 'FileChatHistoryMemory'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "2a81f5ad",
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "from langchain.memory.chat_message_histories import FileChatMessageHistory\n",
    "\n",
    "agent = AutoGPT.from_llm_and_tools(\n",
    "    ai_name=\"Tom\",\n",
    "    ai_role=\"Assistant\",\n",
    "    tools=tools,\n",
    "    llm=ChatOpenAI(temperature=0),\n",
    "    memory=vectorstore.as_retriever(),\n",
    "    chat_history_memory=FileChatMessageHistory(\"chat_history.txt\"),\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b1403008",
   "metadata": {
    "collapsed": false
   },
   "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": "3.9.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}