ningg/happy-llm-colab

Converts the original happy-llm repository into executable Jupyter notebooks optimized for Google Colab, enabling GPU-accelerated LLM training and inference without local setup. Covers the complete pipeline from NLP fundamentals through Transformer architecture, pre-training, and fine-tuning techniques (LoRA/QLoRA), with hands-on implementations of models like LLaMA2 and tokenizers. Maintained on weekly cadence to sync with upstream happy-llm updates, with automated notebook generation via nbformat/nbconvert tooling.