LLM from scratch + training (pre-training / fine-tuning) - View it on GitHub