🧠 RAG Chatbot (Retrieval-Augmented Generation)

A simple Persian-language chatbot based on the Retrieval-Augmented Generation (RAG) architecture. It combines:

• A model to retrieve relevant information from a small knowledge base
• A language model to generate natural language responses

🚀 Features

• Supports Persian language queries and responses
• Uses Hugging Face Transformers for text generation
• Employs Sentence Transformers for semantic similarity
• Simple and educational structure for learning RAG concepts

🛠 Technologies Used

• transformers (for GPT-2 or compatible models)
• sentence-transformers (for sentence embeddings)
• scikit-learn (for cosine similarity)
• PyTorch (for model execution)

🧩 How It Works

1. Retrieve relevant knowledge from a simple knowledge base using semantic similarity.
2. Build a context prompt using the retrieved information and the user query.
3. Generate a response using a causal language model (e.g., GPT-2).

📦 Installation

pip install torch transformers sentence-transformers scikit-learn numpy