🧠 Sentence-Level Small Language Model (SLM) with GPT2

This project demonstrates a lightweight implementation of a small language model (SLM) using the GPT2 architecture via Hugging Face 🤗 Transformers.

It supports text generation in both Persian and English, making it ideal for educational use, rapid experimentation, and low-resource environments.

---

✨ Features

• ✅ Based on the compact GPT2 model: openai-community/gpt2
• 🧠 Suitable as a small language model (SLM) for quick testing and concept validation
• 🗣️ Supports bilingual prompts: Farsi (فارسی) & English
• ⚡ Fast execution with minimal dependencies
• 🔁 Easily extendable for fine-tuning or creative applications (e.g., poetry, Q&A, storytelling)

---

📦 Installation

Install required packages using pip:

pip install transformers accelerate torch

---

🚀 How It Works

1. Load the GPT2 model and tokenizer from Hugging Face Hub
2. Define a generation function with parameters like temperature and top-p
3. Provide a prompt in Farsi or English
4. Generate text continuation using the generate() method

---

📋 Example Prompts

# Persian example
prompt = "چرا آسمان آبی است؟"

# English example
prompt = "Write a short story about a brave knight and a dragon."

Sample output:

Prompt (فارسی): چرا آسمان آبی است؟
Generated: به دلیل پراکندگی نور خورشید توسط جو زمین، بیشتر نور آبی دیده می‌شود...

Prompt (English): Write a short story about a brave knight...
Generated: ...who ventured into the mountains to protect his village from the dragon...

---

📁 Project Structure

slm.py                # Main Python script for bilingual text generation
README.md             # Project description and instructions

---

💡 Use Cases

• Bilingual prompt testing
• Educational demos for language modeling
• Creative writing, poetry, or storytelling
• Exploring small LLM behavior in controlled setups

---

🛠️ Built With

• Transformers
• PyTorch
• Accelerate

---