wehub-resource-sync
772 lines
27 KiB
Plaintext
772 lines
27 KiB
Plaintext
{
|
|
"cells": [
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": null,
|
|
"metadata": {
|
|
"id": "iyi0u1inBcr1"
|
|
},
|
|
"outputs": [],
|
|
"source": [
|
|
"# Copyright 2024 Google LLC\n",
|
|
"#\n",
|
|
"# Licensed under the Apache License, Version 2.0 (the \"License\");\n",
|
|
"# you may not use this file except in compliance with the License.\n",
|
|
"# You may obtain a copy of the License at\n",
|
|
"#\n",
|
|
"# https://www.apache.org/licenses/LICENSE-2.0\n",
|
|
"#\n",
|
|
"# Unless required by applicable law or agreed to in writing, software\n",
|
|
"# distributed under the License is distributed on an \"AS IS\" BASIS,\n",
|
|
"# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.\n",
|
|
"# See the License for the specific language governing permissions and\n",
|
|
"# limitations under the License."
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {
|
|
"id": "BLIlkXRfCNyQ"
|
|
},
|
|
"source": [
|
|
"# Intra Knowledge QnA\n",
|
|
"\n",
|
|
"<table align=\"left\">\n",
|
|
" <td style=\"text-align: center\">\n",
|
|
" <a href=\"https://colab.research.google.com/github/GoogleCloudPlatform/generative-ai/blob/main/gemini/use-cases/retrieval-augmented-generation/intra_knowledge_qna.ipynb\">\n",
|
|
" <img width=\"32px\" src=\"https://www.gstatic.com/pantheon/images/bigquery/welcome_page/colab-logo.svg\" alt=\"Google Colaboratory logo\"><br> Open in Colab\n",
|
|
" </a>\n",
|
|
" </td>\n",
|
|
" <td style=\"text-align: center\">\n",
|
|
" <a href=\"https://console.cloud.google.com/vertex-ai/colab/import/https:%2F%2Fraw.githubusercontent.com%2FGoogleCloudPlatform%2Fgenerative-ai%2Fmain%2Fgemini%2Fuse-cases%2Fretrieval-augmented-generation%2Fintra_knowledge_qna.ipynb\">\n",
|
|
" <img width=\"32px\" src=\"https://lh3.googleusercontent.com/JmcxdQi-qOpctIvWKgPtrzZdJJK-J3sWE1RsfjZNwshCFgE_9fULcNpuXYTilIR2hjwN\" alt=\"Google Cloud Colab Enterprise logo\"><br> Open in Colab Enterprise\n",
|
|
" </a>\n",
|
|
" </td> \n",
|
|
" <td style=\"text-align: center\">\n",
|
|
" <a href=\"https://console.cloud.google.com/vertex-ai/workbench/deploy-notebook?download_url=https://raw.githubusercontent.com/GoogleCloudPlatform/generative-ai/main/gemini/use-cases/retrieval-augmented-generation/intra_knowledge_qna.ipynb\">\n",
|
|
" <img src=\"https://lh3.googleusercontent.com/UiNooY4LUgW_oTvpsNhPpQzsstV5W8F7rYgxgGBD85cWJoLmrOzhVs_ksK_vgx40SHs7jCqkTkCk=e14-rj-sc0xffffff-h130-w32\" alt=\"Vertex AI logo\"><br> Open in Workbench\n",
|
|
" </a>\n",
|
|
" </td>\n",
|
|
" <td style=\"text-align: center\">\n",
|
|
" <a href=\"https://github.com/GoogleCloudPlatform/generative-ai/blob/main/gemini/use-cases/retrieval-augmented-generation/intra_knowledge_qna.ipynb\">\n",
|
|
" <img width=\"32px\" src=\"https://raw.githubusercontent.com/primer/octicons/refs/heads/main/icons/mark-github-24.svg\" alt=\"GitHub logo\"><br> View on GitHub\n",
|
|
" </a>\n",
|
|
" </td>\n",
|
|
"</table>\n",
|
|
"\n",
|
|
"<div style=\"clear: both;\"></div>\n",
|
|
"\n",
|
|
"<b>Share to:</b>\n",
|
|
"\n",
|
|
"<a href=\"https://www.linkedin.com/sharing/share-offsite/?url=https%3A//github.com/GoogleCloudPlatform/generative-ai/blob/main/gemini/use-cases/retrieval-augmented-generation/intra_knowledge_qna.ipynb\" target=\"_blank\">\n",
|
|
" <img width=\"20px\" src=\"https://upload.wikimedia.org/wikipedia/commons/8/81/LinkedIn_icon.svg\" alt=\"LinkedIn logo\">\n",
|
|
"</a>\n",
|
|
"\n",
|
|
"<a href=\"https://bsky.app/intent/compose?text=https%3A//github.com/GoogleCloudPlatform/generative-ai/blob/main/gemini/use-cases/retrieval-augmented-generation/intra_knowledge_qna.ipynb\" target=\"_blank\">\n",
|
|
" <img width=\"20px\" src=\"https://upload.wikimedia.org/wikipedia/commons/7/7a/Bluesky_Logo.svg\" alt=\"Bluesky logo\">\n",
|
|
"</a>\n",
|
|
"\n",
|
|
"<a href=\"https://twitter.com/intent/tweet?url=https%3A//github.com/GoogleCloudPlatform/generative-ai/blob/main/gemini/use-cases/retrieval-augmented-generation/intra_knowledge_qna.ipynb\" target=\"_blank\">\n",
|
|
" <img width=\"20px\" src=\"https://upload.wikimedia.org/wikipedia/commons/5/5a/X_icon_2.svg\" alt=\"X logo\">\n",
|
|
"</a>\n",
|
|
"\n",
|
|
"<a href=\"https://reddit.com/submit?url=https%3A//github.com/GoogleCloudPlatform/generative-ai/blob/main/gemini/use-cases/retrieval-augmented-generation/intra_knowledge_qna.ipynb\" target=\"_blank\">\n",
|
|
" <img width=\"20px\" src=\"https://redditinc.com/hubfs/Reddit%20Inc/Brand/Reddit_Logo.png\" alt=\"Reddit logo\">\n",
|
|
"</a>\n",
|
|
"\n",
|
|
"<a href=\"https://www.facebook.com/sharer/sharer.php?u=https%3A//github.com/GoogleCloudPlatform/generative-ai/blob/main/gemini/use-cases/retrieval-augmented-generation/intra_knowledge_qna.ipynb\" target=\"_blank\">\n",
|
|
" <img width=\"20px\" src=\"https://upload.wikimedia.org/wikipedia/commons/5/51/Facebook_f_logo_%282019%29.svg\" alt=\"Facebook logo\">\n",
|
|
"</a> "
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {
|
|
"id": "WBpYfJP8Ce4s"
|
|
},
|
|
"source": [
|
|
"| | |\n",
|
|
"|--------------------------------------------|---------------------------------------------------------|\n",
|
|
"|Author(s) | |\n",
|
|
"|[Tanya Warrier](https://github.com/tanyarw) |[Samriddhi Mishra](https://github.com/samriddhimishra07) |\n",
|
|
"|[Neelay Shah](https://github.com/neelay21) | [Kumar Saurabh](https://github.com/kusaurabh24) |"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {
|
|
"id": "WZPabwVQCh1y"
|
|
},
|
|
"source": [
|
|
"## Overview\n",
|
|
"\n",
|
|
"- **Gemini:** [Gemini](https://ai.google.dev/models/gemini) is a family of generative AI models that lets developers generate content and solve problems. These models are designed and trained to handle both text and images as input.\n",
|
|
" - **Gemini 2.0 model (gemini-2.0-flash):** Designed to handle multimodal inputs, including multi-turn text and code chat, and code generation.\n",
|
|
"\n",
|
|
"- **LangChain:** [LangChain](https://www.langchain.com/) is a framework designed to make integration of Large Language Models (LLM) like Gemini easier for applications.\n",
|
|
"\n",
|
|
"- **Chroma DB:** [Chroma](https://python.langchain.com/docs/integrations/vectorstores/chroma) is the open-source embedding database. Chroma makes it easy to build LLM apps by making knowledge, facts, and skills pluggable for LLMs.\n",
|
|
"\n",
|
|
"- **Vertex AI Embeddings for Text:** With [text-embedding-005](https://cloud.google.com/vertex-ai/generative-ai/docs/model-reference/text-embeddings) models we can easily create a text embedding with LLM. *text-embedding-005* is the newest stable embedding model.\n",
|
|
"\n",
|
|
"\n",
|
|
"For more information, see the [Generative AI](https://cloud.google.com/vertex-ai/docs/generative-ai/learn/overview) on Vertex AI documentation.\n"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {
|
|
"id": "ZPJulXlX0oSd"
|
|
},
|
|
"source": [
|
|
"### Objectives\n",
|
|
"\n",
|
|
"This notebook leverages Google Vertex AI's Generative capabilities to parse multiple PDF documents, offering Question & Answering functionality by cross-referencing the provided documents. Additionally, it assists in identification of relevant answer sections within the provided documents.\n",
|
|
"\n",
|
|
"This notebook uses [LangChain](https://python.langchain.com/docs/get_started/introduction.html) to operationalize workflows for Question & Answer capabilities, and it utilizes [Chroma DB](https://python.langchain.com/docs/integrations/vectorstores/chroma) to persist document embeddings for similarity searches.\n"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {
|
|
"id": "K6-K0sBh4Ape"
|
|
},
|
|
"source": [
|
|
"### Architecture outlining the workflow approach"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {
|
|
"id": "0zvl1NbGDKdR"
|
|
},
|
|
"source": [
|
|
""
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {
|
|
"id": "fdU7KNcCCkOA"
|
|
},
|
|
"source": [
|
|
"## Getting Started\n"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {
|
|
"id": "6XczCHZYCmi1"
|
|
},
|
|
"source": [
|
|
"### Install Vertex AI SDK and other required packages\n",
|
|
"\n",
|
|
"Install LangChain's Python library, `langchain`, LangChain's integration package for Google Vertex AI, `langchain_google_vertexai`, Chroma for persisting embeddings, `chromadb` and other required packages."
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": null,
|
|
"metadata": {
|
|
"id": "nzZqdQN4_GDz"
|
|
},
|
|
"outputs": [],
|
|
"source": [
|
|
"!apt-get install poppler-utils \\\n",
|
|
"tesseract-ocr -y\n",
|
|
"\n",
|
|
"%pip install google-cloud-aiplatform==1.46.0 \\\n",
|
|
"'bigframes<1.0.0' \\\n",
|
|
"langchain==0.1.14 \\\n",
|
|
"langchain_google_vertexai==0.1.2 \\\n",
|
|
"chromadb==0.4.24 \\\n",
|
|
"unstructured==0.12.6 \\\n",
|
|
"pillow-heif==0.15.0 \\\n",
|
|
"unstructured-inference==0.7.25 \\\n",
|
|
"pypdf==4.1.0 \\\n",
|
|
"pdf2image==1.17.0 \\\n",
|
|
"unstructured_pytesseract==0.3.12 \\\n",
|
|
"pikepdf==8.14.0 \\\n",
|
|
"--upgrade \\\n",
|
|
"--user --quiet"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {
|
|
"id": "bHV6vh55Cqln"
|
|
},
|
|
"source": [
|
|
"### Restart runtime (Colab only)\n",
|
|
"\n",
|
|
"To use the newly installed packages in this Jupyter runtime, you must restart the runtime. You can do this by running the cell below, which will restart the current kernel."
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": null,
|
|
"metadata": {
|
|
"id": "mB85g9FpDu0e"
|
|
},
|
|
"outputs": [],
|
|
"source": [
|
|
"import sys\n",
|
|
"\n",
|
|
"if \"google.colab\" in sys.modules:\n",
|
|
" import IPython\n",
|
|
"\n",
|
|
" app = IPython.Application.instance()\n",
|
|
" app.kernel.do_shutdown(True)"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {
|
|
"id": "WcesVUMO6sDg"
|
|
},
|
|
"source": [
|
|
"<div class=\"alert alert-block alert-warning\">\n",
|
|
"<b>⚠️ The kernel is going to restart. Please wait until it is finished before continuing to the next step. ⚠️</b>\n",
|
|
"</div>"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {
|
|
"id": "K810s6DICwe1"
|
|
},
|
|
"source": [
|
|
"### Authenticate your notebook environment (Colab only)\n",
|
|
"\n",
|
|
"\n",
|
|
"If you are running this notebook on Google Colab, run the following cell to authenticate your environment."
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": null,
|
|
"metadata": {
|
|
"id": "K0xFZPSpDcog"
|
|
},
|
|
"outputs": [],
|
|
"source": [
|
|
"import sys\n",
|
|
"\n",
|
|
"if \"google.colab\" in sys.modules:\n",
|
|
" from google.colab import auth\n",
|
|
"\n",
|
|
" auth.authenticate_user()"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {
|
|
"id": "xyDpSIRWC0g2"
|
|
},
|
|
"source": [
|
|
"### Set Google Cloud project information and initialize Vertex AI SDK\n",
|
|
"\n",
|
|
"To get started using Vertex AI, you must have an existing Google Cloud project and [enable the Vertex AI API](https://console.cloud.google.com/flows/enableapi?apiid=aiplatform.googleapis.com).\n",
|
|
"\n",
|
|
"Learn more about [setting up a project and a development environment](https://cloud.google.com/vertex-ai/docs/start/cloud-environment)."
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": null,
|
|
"metadata": {
|
|
"id": "mv3ho2AwDkID"
|
|
},
|
|
"outputs": [],
|
|
"source": [
|
|
"PROJECT_ID = \"your-project-id\" # @param {type:\"string\"}\n",
|
|
"LOCATION = \"us-central1\" # @param {type:\"string\"}\n",
|
|
"\n",
|
|
"import vertexai\n",
|
|
"\n",
|
|
"vertexai.init(project=PROJECT_ID, location=LOCATION)"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {
|
|
"id": "z2IAV3YtEFQ4"
|
|
},
|
|
"source": [
|
|
"### Variables\n",
|
|
"\n",
|
|
"Create variables to store the folder location of all the pdf documents needed to be indexed. Also, the location to store the embedding database for the documents."
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": null,
|
|
"metadata": {
|
|
"id": "c-cbO2o9F0Yx"
|
|
},
|
|
"outputs": [],
|
|
"source": [
|
|
"INDEX_PATH = \"./Dataset/\"\n",
|
|
"PERSIST_PATH = \"./PersistentDB/\"\n",
|
|
"\n",
|
|
"MULTIMODAL_MODEL = \"gemini-2.0-flash\"\n",
|
|
"EMBEDDING_MODEL = \"text-embedding-005\""
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {
|
|
"id": "ooauhzOWPvC1"
|
|
},
|
|
"source": [
|
|
"### Import Libraries\n"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": null,
|
|
"metadata": {
|
|
"id": "Hj6K6SFIEnRk"
|
|
},
|
|
"outputs": [],
|
|
"source": [
|
|
"# Utils\n",
|
|
"import os\n",
|
|
"\n",
|
|
"# LangChain\n",
|
|
"import langchain\n",
|
|
"from langchain.chains import create_retrieval_chain\n",
|
|
"from langchain.chains.combine_documents import create_stuff_documents_chain\n",
|
|
"from langchain.document_loaders import TextLoader, UnstructuredPDFLoader\n",
|
|
"from langchain.prompts import PromptTemplate\n",
|
|
"from langchain.text_splitter import CharacterTextSplitter\n",
|
|
"from langchain.vectorstores.chroma import Chroma\n",
|
|
"from langchain_google_vertexai import VertexAI, VertexAIEmbeddings\n",
|
|
"\n",
|
|
"print(f\"LangChain version: {langchain.__version__}\")\n",
|
|
"\n",
|
|
"# Vertex AI\n",
|
|
"from google.cloud import aiplatform\n",
|
|
"\n",
|
|
"print(f\"Vertex AI SDK version: {aiplatform.__version__}\")\n",
|
|
"\n",
|
|
"from IPython.display import clear_output\n",
|
|
"\n",
|
|
"# HTML Widgets\n",
|
|
"import ipywidgets as widgets"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {
|
|
"id": "_D5yf9ILC3ZM"
|
|
},
|
|
"source": [
|
|
"### Data Preparation\n",
|
|
"\n",
|
|
"We will be using the public [Internal Revenue Service(IRS) document](https://www.irs.gov/pub/irs-pdf/p554.pdf) which states the details for each section of tax eligibility for seniors in the USA. It consists of 37 pages.\n",
|
|
"\n",
|
|
"This document serves as the input PDF for generating and indexing embeddings, querying the model, and facilitating Question and Answer scenarios based on the data corpus."
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": null,
|
|
"metadata": {
|
|
"id": "QyFyNdqvD08b"
|
|
},
|
|
"outputs": [],
|
|
"source": [
|
|
"# Create the folder for input files\n",
|
|
"!mkdir -p $INDEX_PATH\n",
|
|
"\n",
|
|
"# Download the files\n",
|
|
"!wget https://www.irs.gov/pub/irs-pdf/p554.pdf -P $INDEX_PATH"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {
|
|
"id": "hy1MspayJCaO"
|
|
},
|
|
"source": [
|
|
"### Define utility functions"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {
|
|
"id": "hul0HgRpUVCC"
|
|
},
|
|
"source": [
|
|
"**Document Loading:** Read in the content of the documents from the provided directory.\n",
|
|
"\n",
|
|
"**Document Splitting:**\n",
|
|
"Creates a CharacterTextSplitter object with parameters:\n",
|
|
"- `chunk_size=8192`: Aiming for text chunks of around 8192 characters.\n",
|
|
"- `chunk_overlap=128:` A small overlap between chunks, likely to preserve context when documents are split.\n",
|
|
"\n",
|
|
"Applies the *text_splitter* to the loaded documents, breaking them into smaller, more manageable text chunks.\n",
|
|
"\n",
|
|
"Gathers all the split text chunks into a single list and return for further processing."
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": null,
|
|
"metadata": {
|
|
"id": "l3RcBOL3T544"
|
|
},
|
|
"outputs": [],
|
|
"source": [
|
|
"def get_split_documents(index_path: str) -> list[str]:\n",
|
|
" \"\"\"\n",
|
|
" This function is used to chunk documents and convert them into a list.\n",
|
|
"\n",
|
|
" Args:\n",
|
|
" index_path: Path of the dataset folder containing the documents.\n",
|
|
"\n",
|
|
" Returns:\n",
|
|
" List of chunked, or split documents.\n",
|
|
" \"\"\"\n",
|
|
"\n",
|
|
" split_docs = []\n",
|
|
"\n",
|
|
" for file_name in os.listdir(index_path):\n",
|
|
" print(f\"file_name : {file_name}\")\n",
|
|
" if file_name.endswith(\".pdf\"):\n",
|
|
" loader = UnstructuredPDFLoader(index_path + file_name)\n",
|
|
" else:\n",
|
|
" loader = TextLoader(index_path + file_name)\n",
|
|
"\n",
|
|
" text_splitter = CharacterTextSplitter(chunk_size=8192, chunk_overlap=128)\n",
|
|
" split_docs.extend(text_splitter.split_documents(loader.load()))\n",
|
|
"\n",
|
|
" return split_docs"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {
|
|
"id": "QLZZqHlbS1QS"
|
|
},
|
|
"source": [
|
|
"### Create Vector Database\n",
|
|
"\n",
|
|
"Instantiate a VertexAIEmbeddings embedding object that will efficiently generate text embeddings using the specified `text-embedding-005` model."
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": null,
|
|
"metadata": {
|
|
"id": "Zgf_ZbKAFuIr"
|
|
},
|
|
"outputs": [],
|
|
"source": [
|
|
"# Custom Vertex AI Embeddings object\n",
|
|
"EMBEDDING_NUM_BATCH = 5\n",
|
|
"\n",
|
|
"embeddings = VertexAIEmbeddings(\n",
|
|
" model_name=EMBEDDING_MODEL, batch_size=EMBEDDING_NUM_BATCH\n",
|
|
")"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {
|
|
"id": "U7hZWC6_X26R"
|
|
},
|
|
"source": [
|
|
"**Load Documents:** `get_split_documents` function retrieves and preprocesses documents from the specified `INDEX_PATH`.\n",
|
|
"\n",
|
|
"**Generate Embeddings:** The code generates vector embeddings\n",
|
|
"\n",
|
|
"**Create Vector Database:** A Chroma vector database (db) is initialized. This specialized database is designed for storing and efficiently searching vector embeddings.\n",
|
|
"\n",
|
|
"**Persist Database:** The `db.persist()` command saves the newly created vector database to disk at the location defined by `PERSIST_PATH` environment variable."
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": null,
|
|
"metadata": {
|
|
"id": "CvvVjL2pGzCB"
|
|
},
|
|
"outputs": [],
|
|
"source": [
|
|
"# Load documents, generate vectors and store in Vector database\n",
|
|
"split_docs = get_split_documents(INDEX_PATH)\n",
|
|
"\n",
|
|
"db = Chroma.from_documents(\n",
|
|
" documents=split_docs, embedding=embeddings, persist_directory=PERSIST_PATH\n",
|
|
")\n",
|
|
"db.persist() # Ensure DB persist"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {
|
|
"id": "9nqDSy0QYsgC"
|
|
},
|
|
"source": [
|
|
"### Create the Retriever"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {
|
|
"id": "nypubf9LRCdk"
|
|
},
|
|
"source": [
|
|
"Load the `gemini-2.0-flash` generative model with parameters."
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": null,
|
|
"metadata": {
|
|
"id": "zi7c-8sxRACS"
|
|
},
|
|
"outputs": [],
|
|
"source": [
|
|
"# Initializing the Vertex AI Language model with required parameters\n",
|
|
"llm = VertexAI(\n",
|
|
" model=MULTIMODAL_MODEL,\n",
|
|
" max_output_tokens=2048,\n",
|
|
" temperature=0.2,\n",
|
|
" top_p=0.8,\n",
|
|
" top_k=40,\n",
|
|
" verbose=True,\n",
|
|
")"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {
|
|
"id": "gwBobHo6jLO_"
|
|
},
|
|
"source": [
|
|
"Creates a retriever utilizing the Chroma vector store for similarity search.\n",
|
|
"\n",
|
|
"`search_kwargs={\"k\": 3}` - This parameter, specific to similarity search, dictates that the retriever should return the top 3 most relevant documents based on the calculated similarity scores.\n"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": null,
|
|
"metadata": {
|
|
"id": "XbFtaHWPMmlA"
|
|
},
|
|
"outputs": [],
|
|
"source": [
|
|
"# Expose index to the retriever\n",
|
|
"retriever = db.as_retriever(search_type=\"similarity\", search_kwargs={\"k\": 3})"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {
|
|
"id": "jdXK30NKa8cK"
|
|
},
|
|
"source": [
|
|
"Define a prompt template for a language model that consists of a string template. It accepts a set of parameters from the user that can be used to generate a prompt for a language model.\n",
|
|
"\n",
|
|
"More on [PromptTemplate](https://api.python.langchain.com/en/latest/prompts/langchain_core.prompts.prompt.PromptTemplate.html)."
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": null,
|
|
"metadata": {
|
|
"id": "-Mbb-3X_mFwl"
|
|
},
|
|
"outputs": [],
|
|
"source": [
|
|
"template = \"\"\"\n",
|
|
"You are a helpful AI assistant. You're tasked to answer the question given below, but only based on the context provided.\n",
|
|
"context:\n",
|
|
"<context>\n",
|
|
"{context}\n",
|
|
"</context>\n",
|
|
"\n",
|
|
"question:\n",
|
|
"<question>\n",
|
|
"{input}\n",
|
|
"</question>\n",
|
|
"\n",
|
|
"If you cannot find an answer ask the user to rephrase the question.\n",
|
|
"answer:\n",
|
|
"\n",
|
|
"\"\"\"\n",
|
|
"prompt = PromptTemplate.from_template(template)"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {
|
|
"id": "HOHivtVAmOtr"
|
|
},
|
|
"source": [
|
|
"Create the retrieval chain and invoke it by passing the question as an input."
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": null,
|
|
"metadata": {
|
|
"id": "NL6r_Yn5GzlS"
|
|
},
|
|
"outputs": [],
|
|
"source": [
|
|
"# Create the retrieval chain\n",
|
|
"combine_docs_chain = create_stuff_documents_chain(llm, prompt)\n",
|
|
"retrieval_chain = create_retrieval_chain(retriever, combine_docs_chain)"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": null,
|
|
"metadata": {
|
|
"id": "auqtXV1ijfQH"
|
|
},
|
|
"outputs": [],
|
|
"source": [
|
|
"# Invoke the retrieval chain\n",
|
|
"response = retrieval_chain.invoke({\"input\": \"Tell me about Figuring the EIC.\"})"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": null,
|
|
"metadata": {
|
|
"id": "Tbbks_OV6ce0"
|
|
},
|
|
"outputs": [],
|
|
"source": [
|
|
"print(response[\"answer\"])"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {
|
|
"id": "1raHWPNlrnff"
|
|
},
|
|
"source": [
|
|
"### Example Questions:\n",
|
|
"\n",
|
|
"- Special rules for joint returns.\n",
|
|
"- Tell about persons not eligible for the standard deduction.\n",
|
|
"- Tell me about Figuring the EIC.\n",
|
|
"- Tell about contributions to Kay Bailey Hutchison Spousal IRAs.\n",
|
|
"- Tell about standard deduction for dependents.\n"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {
|
|
"id": "ie7RLNMbHJnl"
|
|
},
|
|
"source": [
|
|
"## Interactive UI Widget for Question-Answering\n",
|
|
"\n",
|
|
"Enter your question in the input box and choose one of the options:\n",
|
|
"\n",
|
|
"**Ask Me!:** This option will generate an answer using similarity search on vector embeddings.\n",
|
|
"\n",
|
|
"**More Details:** Select this option to input the generated answer into a Large Language Model (LLM) for a more elaborate response."
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": null,
|
|
"metadata": {
|
|
"id": "-Hu94QetHM5v"
|
|
},
|
|
"outputs": [],
|
|
"source": [
|
|
"button = widgets.Button(description=\"Ask Me!\")\n",
|
|
"output = widgets.Output()\n",
|
|
"button_stp = widgets.Button(description=\"More details\")\n",
|
|
"output = widgets.Output()\n",
|
|
"text = widgets.Text(\n",
|
|
" description=\"Question:\", layout=widgets.Layout(width=\"80%\", height=\"50px\")\n",
|
|
")\n",
|
|
"display(text, button, button_stp, output)\n",
|
|
"\n",
|
|
"\n",
|
|
"@output.capture()\n",
|
|
"def on_button_clicked(b):\n",
|
|
" clear_output()\n",
|
|
" question = text.value\n",
|
|
"\n",
|
|
" result = retrieval_chain.invoke({\"input\": question})\n",
|
|
" source_documents = list({doc.metadata[\"source\"] for doc in result[\"context\"]})\n",
|
|
"\n",
|
|
" print(\"\\nAnswer-\", result[\"answer\"])\n",
|
|
" print(\"\\nSource-\", \"\\n\".join(source_documents))\n",
|
|
" print(\"\\n\")\n",
|
|
"\n",
|
|
"\n",
|
|
"@output.capture()\n",
|
|
"def on_stp_clicked(b):\n",
|
|
" clear_output()\n",
|
|
" question = text.value\n",
|
|
" query = question + \"Give detailed information as much as possible. \"\n",
|
|
" result = retrieval_chain.invoke({\"input\": query})\n",
|
|
" source_documents = list({doc.metadata[\"source\"] for doc in result[\"context\"]})\n",
|
|
"\n",
|
|
" print(\"\\nAnswer-\", result[\"answer\"])\n",
|
|
" print(\"\\nSource-\", \"\\n\".join(source_documents))\n",
|
|
" print(\"\\n\")\n",
|
|
"\n",
|
|
"\n",
|
|
"button.on_click(on_button_clicked)\n",
|
|
"button_stp.on_click(on_stp_clicked)"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {
|
|
"id": "X33AAFNiC55l"
|
|
},
|
|
"source": [
|
|
"## Cleaning up"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {
|
|
"id": "cJ2NCKA7RlkX"
|
|
},
|
|
"source": [
|
|
"Remove all the downloaded files and the vector database"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": null,
|
|
"metadata": {
|
|
"id": "_b56oorFHmS3"
|
|
},
|
|
"outputs": [],
|
|
"source": [
|
|
"!rm -r $INDEX_PATH\n",
|
|
"!rm -r $PERSIST_PATH"
|
|
]
|
|
}
|
|
],
|
|
"metadata": {
|
|
"colab": {
|
|
"name": "intra_knowledge_qna.ipynb",
|
|
"toc_visible": true
|
|
},
|
|
"kernelspec": {
|
|
"display_name": "Python 3",
|
|
"name": "python3"
|
|
}
|
|
},
|
|
"nbformat": 4,
|
|
"nbformat_minor": 0
|
|
}
|