🎥 YouTube RAG ChatBot

A production-style Retrieval-Augmented Generation (RAG) chatbot that answers questions about any YouTube video using its transcript.

Instead of relying only on an LLM's knowledge, the chatbot retrieves the most relevant transcript chunks using FAISS and supplies them as context to the LLM, resulting in more accurate and grounded answers.

---

🚀 Features

• 📺 Load transcripts directly from YouTube
• ✂️ Intelligent document chunking
• 🔍 Semantic search using FAISS
• 🧠 Hugging Face Embeddings
• 🤖 Qwen 2.5 Instruct LLM
• 💬 Interactive command-line chatbot
• 🏗️ Modular production-style architecture
• ⚡ Easily extensible to PDFs, websites, and documents

---

Tech Stack

Technology	Purpose
Python	Programming Language
LangChain	RAG Pipeline
Hugging Face	LLM & Embeddings
FAISS	Vector Database
Sentence Transformers	Embedding Model
YouTube Transcript API	Transcript Extraction
dotenv	Environment Variables

---

Project Architecture

User Question
      │
      ▼
Retriever
      │
      ▼
FAISS Vector Store
      ▲
      │
Transcript Chunks
      ▲
      │
Text Splitter
      ▲
      │
YouTube Transcript
      │
      ▼
Large Language Model
      │
      ▼
Final Answer

---

Project Structure

YouTube_RAG_ChatBot/

│
├── chunking/
│   └── chunker.py
│
├── embeddings/
│   └── embed.py
│
├── ingestion/
│   └── loader.py
│
├── llm/
│   └── model.py
│
├── prompt/
│   └── prompt_temp.py
│
├── retrieval/
│   └── retriever.py
│
├── vector_store/
│   └── faiss_store.py
│
├── tests/
│
├── .env
├── .gitignore
├── main.py
└── requirements.txt

---

Pipeline

YouTube Video
      │
      ▼
Load Transcript
      │
      ▼
Split into Chunks
      │
      ▼
Generate Embeddings
      │
      ▼
Create FAISS Index
      │
      ▼
Retriever
      │
      ▼
Prompt Template
      │
      ▼
Qwen LLM
      │
      ▼
Answer

---

Installation

Clone the repository

git clone https://github.com/alobroke/YouTube_RAG_ChatBot.git

cd YouTube_RAG_ChatBot

Create Virtual Environment

python -m venv .venv

Activate

Windows

.venv\Scripts\activate

Linux / Mac

source .venv/bin/activate

Install dependencies

pip install -r requirements.txt

---

Environment Variables

Create a .env file.

HUGGINGFACEHUB_API_TOKEN=your_huggingface_token

---

Run

python main.py

---

Example

Input

Question:
Explain RAG

Output

RAG stands for Retrieval-Augmented Generation.

It combines retrieval systems with Large Language Models
to generate context-aware and accurate responses.

This is intentionally focused on the fundamentals so that anyone starting their AI journey can understand how RAG works before diving into more advanced concepts. 🔗 Read the full article here: https://lnkd.in/dawAvDpZ