06-LoadingDocumentsToVectoreStore.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "from typing import Optional, List, Tuple\n",
    "from dotenv import load_dotenv\n",
    "\n",
    "load_dotenv()\n",
    "pd.set_option(\"display.max_colwidth\", None)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "import datasets\n",
    "\n",
    "ds = datasets.load_dataset(\"clarin-pl/poquad\", split=\"validation\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "a0870394e8cf4fa3b141c7c5964725f0",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/1449 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from tqdm.notebook import tqdm\n",
    "from langchain.docstore.document import Document as LangchainDocument\n",
    "\n",
    "unique_contexts = set((row['context'], row['title']) for row in ds)\n",
    "\n",
    "RAW_KNOWLEDGE_BASE = [\n",
    "    LangchainDocument(page_content=doc[0], metadata={\"title\": doc[1]}) for doc in tqdm(unique_contexts)\n",
    "]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "from langchain.text_splitter import RecursiveCharacterTextSplitter\n",
    "\n",
    "text_splitter = RecursiveCharacterTextSplitter(\n",
    "    chunk_size=1000,\n",
    "    chunk_overlap=100, \n",
    "    add_start_index=True,\n",
    "    strip_whitespace=True,\n",
    ")\n",
    "\n",
    "docs_processed = []\n",
    "for doc in RAW_KNOWLEDGE_BASE:\n",
    "    docs_processed += text_splitter.split_documents([doc])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model's maximum sequence length: 512\n"
     ]
    }
   ],
   "source": [
    "from sentence_transformers import SentenceTransformer\n",
    "\n",
    "# To get the value of the max sequence_length, we will query the underlying `SentenceTransformer` object used in the RecursiveCharacterTextSplitter.\n",
    "print(f\"Model's maximum sequence length: {SentenceTransformer('sdadas/mmlw-retrieval-roberta-large').max_seq_length}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "e3ebd4faab244cf5a734e97be7e7dcf1",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/1964 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from transformers import AutoTokenizer\n",
    "\n",
    "tokenizer = AutoTokenizer.from_pretrained(\"sdadas/mmlw-retrieval-roberta-large\")\n",
    "lengths = [len(tokenizer.encode(doc.page_content)) for doc in tqdm(docs_processed)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "\n",
    "fig = pd.Series(lengths).hist()\n",
    "plt.title(\"Distribution of document lengths in the knowledge base (in count of tokens)\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "EMBEDDING_MODEL_NAME = \"sdadas/mmlw-retrieval-roberta-large\"\n",
    "\n",
    "def split_documents(\n",
    "    chunk_size: int,\n",
    "    knowledge_base: List[LangchainDocument],\n",
    "    tokenizer_name: Optional[str] = EMBEDDING_MODEL_NAME,\n",
    ") -> List[LangchainDocument]:\n",
    "    \"\"\"\n",
    "    Split documents into chunks of maximum size `chunk_size` tokens and return a list of documents.\n",
    "    \"\"\"\n",
    "    text_splitter = RecursiveCharacterTextSplitter.from_huggingface_tokenizer(\n",
    "        AutoTokenizer.from_pretrained(tokenizer_name),\n",
    "        chunk_size=chunk_size,\n",
    "        chunk_overlap=int(chunk_size / 10),\n",
    "        add_start_index=True,\n",
    "        strip_whitespace=True,\n",
    "    )\n",
    "\n",
    "    docs_processed = []\n",
    "    for doc in knowledge_base:\n",
    "        docs_processed += text_splitter.split_documents([doc])\n",
    "\n",
    "    # Remove duplicates\n",
    "    unique_texts = {}\n",
    "    docs_processed_unique = []\n",
    "    for doc in docs_processed:\n",
    "        if doc.page_content not in unique_texts:\n",
    "            unique_texts[doc.page_content] = True\n",
    "            docs_processed_unique.append(doc)\n",
    "\n",
    "    return docs_processed_unique\n",
    "\n",
    "\n",
    "docs_processed = split_documents(\n",
    "    512,  # We choose a chunk size adapted to our model\n",
    "    RAW_KNOWLEDGE_BASE,\n",
    "    tokenizer_name=EMBEDDING_MODEL_NAME,\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "b6fdfba40c414c9c9ef9eb81c674d648",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/2603 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "lengths = [len(tokenizer.encode(doc.page_content)) for doc in tqdm(docs_processed)]\n",
    "fig = pd.Series(lengths).hist()\n",
    "plt.title(\"Distribution of document lengths in the knowledge base (in count of tokens)\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "22295791922d42f2a20a8ab6c85c97c2",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Ingesting documents:   0%|          | 0/2603 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from langchain.vectorstores import Qdrant\n",
    "from langchain_community.embeddings import HuggingFaceEmbeddings\n",
    "from langchain_community.vectorstores.utils import DistanceStrategy\n",
    "\n",
    "embedding_model = HuggingFaceEmbeddings(\n",
    "    model_name=EMBEDDING_MODEL_NAME,\n",
    "    multi_process=True,\n",
    "    model_kwargs={\"device\": \"mps\"},\n",
    "    encode_kwargs={\"normalize_embeddings\": True},  # set True for cosine similarity\n",
    ")\n",
    "\n",
    "collection_name = f\"poquad_{EMBEDDING_MODEL_NAME.replace('/', '_')}\"\n",
    "\n",
    "KNOWLEDGE_VECTOR_DATABASE = None\n",
    "\n",
    "with tqdm(total=len(docs_processed), desc=\"Ingesting documents\") as pbar:\n",
    "    for d in docs_processed:\n",
    "        if KNOWLEDGE_VECTOR_DATABASE:\n",
    "            KNOWLEDGE_VECTOR_DATABASE.add_documents([d])\n",
    "        else:\n",
    "            KNOWLEDGE_VECTOR_DATABASE = Qdrant.from_documents(\n",
    "                [d], \n",
    "                embedding_model,    \n",
    "                url=\"http://localhost:6333\",\n",
    "                prefer_grpc=True, \n",
    "                collection_name=collection_name\n",
    "            )\n",
    "        pbar.update(1)  "
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "env",
   "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": "3.12.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}