Welcome to this beginner-friendly repository that demystifies how Large Language Models (LLMs) like GPT, Gemini, and others work – step by step! ✨
If you’ve ever wondered "How can a machine generate text so well?", you're in the right place. This repo breaks down the key concepts behind LLMs in simple code and explanations. No AI degree needed! 👩💻👨🏫
Each folder in this repo is a building block that takes you closer to understanding how LLMs operate:
Concept: Tokenization with Byte Pair Encoding (BPE)
📖 Read the README
🛠️ Code: bpe.py
Learn how words are split into subword units to make models efficient and flexible with language.
Concept: Basic Bi-gram Language Models
📖 Read the README
🛠️ Code: bi_gram.py
📝 Sample Data: data/sample_data.txt
Understand how simple statistical models predict the next word using word pairs.
Concept: The magic of Transformer models (like GPT!)
📖 Read the README
🛠️ Code: gpt_demo.py
See how the Transformer architecture powers state-of-the-art language models.
This project is perfect for:
- 🧑🎓 Students curious about AI
- 🛠️ Developers wanting to peek under the hood
- 🧠 Enthusiasts who enjoy learning how things work
No advanced math or machine learning knowledge required. Just basic Python and curiosity! 🐍💡
Understanding LLMs helps us:
- Make informed decisions about AI usage
- Appreciate the complexity of tools like ChatGPT
- Build our own simple models from scratch!
This project is licensed under the MIT License.