llm-action

Why we included this project

llm-action is a Chinese-language, hands-on curriculum for large language models that gathers what would otherwise be scattered blog posts and framework docs into one structured path. The author works through the full arc of a model project, from data engineering and parameter-efficient fine-tuning (LoRA, QLoRA, P-Tuning v2) to distributed training parallelism, inference optimization, quantization, and evaluation, and most tutorials come with runnable notebooks or scripts, some of which fit on a single consumer GPU. The coverage also reaches areas that are easy to miss when you are just starting out, such as adapting models to domestic hardware, AI compilers, and cluster networking. For developers and small teams planning a first serious fine-tuning or deployment effort, this is a solid companion to keep open next to the official documentation.