🤖 Chatbot with Memory – Python & Hugging Face Transformers

A beginner-friendly chatbot built with Python using the Hugging Face transformers library. This chatbot can hold simple memory of past conversation turns and responds using a pre-trained model. Perfect as a hands-on intro project into LLMs and conversational AI.

---

📂 Project Structure

chatbot/
├── chatbot\_with\_memory.py     # Main chatbot logic with contextual memory
├── app.py                       # Gradio-based web interface
├── requirements.txt             # Python dependencies
├── README.md                    # Project documentation
└── venv/                        # Virtual environment (excluded from version control)

---

🚀 Features

• Uses a pre-trained conversational model from Hugging Face (FLAN-T5 Base).
• Maintains simple conversational history for context-aware responses.
• Clean CLI interface and modern Gradio web UI.
• Easy to swap out models and extend with more capabilities.

---

🧠 Model Info

• Model Used: google/flan-t5-base
• Library: transformers by Hugging Face
• This model balances performance and speed, making it suitable for educational projects and light experimentation.

💡 Note: The previously used model (facebook/blenderbot_small-90M) has been replaced to improve response quality.

---

⚙️ Installation & Setup

1. Clone the repo

git clone https://github.com/mariam-khediri/HugBot.git
cd Hugbot

2. Set up a virtual environment

python -m venv venv
# Activate it:
# On Windows:
venv\Scripts\activate
# On macOS/Linux:
source venv/bin/activate

3. Install dependencies

pip install -r requirements.txt

4. Run the CLI chatbot

python chatbot_with_memory.py

5. Or run the Gradio web app

python app.py

Then open your browser and go to: http://127.0.0.1:7860

---

💬 Example Conversation

You: Hi, who are you?
Bot: I'm your friendly AI assistant. How can I help you today?

You: What's 2 + 2?
Bot: 2 + 2 is 4.

You: And the capital of France?
Bot: The capital of France is Paris.

---

🛠️ Requirements

Ensure you have Python 3.8+ and an internet connection (for initial model download).

transformers>=4.40.0
torch>=2.1.0
gradio>=4.0.0

To regenerate the requirements.txt:

pip freeze > requirements.txt

---

🧪 Future Improvements

• Add persistent memory (e.g., database or embeddings-based memory).
• Add voice input/output using gradio's audio features.
• Support model selection from the UI.
• Integrate with LangChain, OpenRouter, or other LLM APIs.

---

🧠 Learning Goals

This repo is ideal for:

• Beginners learning about LLMs
• Experimenting with Hugging Face Transformers
• Building your first memory-aware chatbot
• Exploring Gradio for quick UI prototyping

---

🙌 Acknowledgments

• Hugging Face Transformers
• Gradio by Hugging Face
• Google FLAN-T5