self-llm

Why we included this project

Self-LLM is a hands-on, Chinese-language tutorial for getting open-source large models running on Linux. It walks beginners through the whole lifecycle: setting up the environment, deploying a model locally, calling it from the command line or a demo, and fine-tuning it with full-parameter, LoRA, or P-tuning methods. The real strength is the breadth of coverage. Step-by-step guides exist for more than fifty mainstream models, including Qwen, ChatGLM, InternLM, LLaMA, and MiniCPM, with dedicated sections for AMD GPUs and Ascend NPUs. Students, researchers, and small teams who want to run or adapt an open model without paying for APIs will find reproducible instructions rather than abstract theory. Worked examples, such as fine-tuning a model on a specific character's dialogue or building a math-tutoring assistant, show how the same pipeline transfers to a custom dataset. This is a curated learning resource and community-maintained reference, not a deployable application.