langchain_rag.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Interactive Introduction to ML and AI with a RAG-System\n",
    "\n",
    "Based on a PDF containing a starter set of DND 5e character [sheets](https://dnd5echaractersheet.com/)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## sys admin\n",
    "\n",
    "Create a .env file with the following content:\n",
    "\n",
    "`OPENAI_API_KEY = \"^<API_KEY>\"`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "# required libraries for the tutorial\n",
    "import openai\n",
    "import os\n",
    "from dotenv import load_dotenv\n",
    "from langchain.document_loaders import PyPDFLoader\n",
    "from langchain.chains import RetrievalQA\n",
    "from langchain_openai import ChatOpenAI\n",
    "from langchain_openai import OpenAIEmbeddings\n",
    "from langchain.text_splitter import RecursiveCharacterTextSplitter\n",
    "from langchain.vectorstores import Chroma"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Load PDF data\n",
    "Loads the data and splits it into chunks.\n",
    "Each chunk contains 1000 characters max with a max overlap of 100 characters."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "text_splitter = RecursiveCharacterTextSplitter(chunk_size=1000, chunk_overlap=100)\n",
    "loader = PyPDFLoader(\"data/StarterSet_Charactersv2.pdf\")\n",
    "chunks = loader.load_and_split(text_splitter)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Check the chunks\n",
    "get chunk content with: chunks[index].page_content"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "print(chunks[0])\n",
    "print(\"The chunk contains \" +  + \" characters\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Setup models\n",
    "\n",
    "We need to prepare an embedding model to vectorise our chunks before storing them into our ChromaDB and a language model to generate answers to our questions."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "# Load environment variables from .env file\n",
    "load_dotenv()\n",
    "\n",
    "# Access the API key using the variable name defined in the .env file\n",
    "openai.api_key = os.getenv(\"OPENAI_API_KEY\")\n",
    "\n",
    "# Initialize the OpenAI chat model\n",
    "llm = ChatOpenAI(model_name=\"gpt-3.5-turbo\", temperature=0.8)\n",
    "\n",
    "# initialize the OpenAI embeddings model\n",
    "embeddings = OpenAIEmbeddings()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Load / Create Chroma DB\n",
    "\n",
    "We check for the existence of the directory for 2 reasons:\n",
    "1) We use Openai Embeddings and pay for the embedding generation\n",
    "2) Chroma does not overwrite an existing database, but allows to upate it"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "if os.path.exists(\"chroma\"):\n",
    "    print(\"Loading Chroma from disk...\")\n",
    "    Chroma(persist_directory=\"chroma\", embedding_function=embeddings)\n",
    "else:\n",
    "    chroma_db = Chroma.from_documents(documents=chunks,\n",
    "                                    embedding=embeddings,\n",
    "                                    persist_directory=\"chroma\",\n",
    "                                    collection_name=\"lc_chroma_demo\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Test Your Database"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "query = \"What is this document about?\""
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Simple Similarity Search"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "result = chroma_db.similarity_search(query)\n",
    "print(result)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Similarity Search with Scores"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "result_with_scores = chroma_db.similarity_search_with_score(query)\n",
    "print(result_with_scores)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "chain = RetrievalQA.from_chain_type(llm=llm, chain_type=\"stuff\", retriever=chroma_db.as_retriever())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "response = chain.invoke(query)\n",
    "print(response)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Test some queries Yourself"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_response(query:str):\n",
    "    ## add the functionality to combine the functionalities above."
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "venv",
   "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": 2
}