🧠 PDF Q&A System with LangChain, ChromaDB, and Sentence Transformers

This project allows you to load a PDF file, split it into manageable chunks, create embeddings using Sentence Transformers, store them in ChromaDB, and ask questions based on the content.

---

🛠️ Technologies Used

• Python 🐍
• langchain
• langchain-chroma
• sentence-transformers
• chroma
• transformers==4.52.4
• formers==4.1.0
• torch==2.7.1

---

📦 Installation

Clone the repository:

git clone https://github.com/solanki505/RAG-Vector-embeddings.git
cd RAG-Vector-embeddings

Install dependencies:

pip install -r requirements.txt

Or manually:

pip install langchain langchain-chroma sentence-transformers chromadb transformers==4.52.4 torch==2.7.1 formers==4.1.0

---

📄 How It Works

1. Import Libraries

from langchain_chroma import Chroma
from langchain_community.document_loaders import PyPDFLoader
from langchain_community.embeddings.sentence_transformer import SentenceTransformerEmbeddings
from langchain_text_splitters import RecursiveCharacterTextSplitter

2. Load the PDF

loader = PyPDFLoader("paul-graham-ideas.pdf")
documents = loader.load()

3. Split into Chunks

text_splitter = RecursiveCharacterTextSplitter(chunk_size=100, chunk_overlap=30)
docs = text_splitter.split_documents(documents)

4. Create Embeddings

embedding_function = SentenceTransformerEmbeddings(model_name="all-MiniLM-L6-v2")

5. Load into ChromaDB

db = Chroma.from_documents(docs, embedding_function)

6. Ask a Question

query = "This essay is derived from where?"
docs = db.similarity_search(query)
print(docs[0].page_content)

---

🎯 Use Cases

• Query academic or research PDFs
• Build educational or legal Q&A bots
• Extract insights from any long document

---

📁 Project Structure

RAG-Vector-embeddings/
│
├── paul-graham-ideas.pdf
├── main.py
├── requirements.txt
└── README.md

---

---

✨

Made with ❤️ by Solanki Sarkar