datawhalechina/self-llm

This project is a Chinese-language tutorial collection for learning how to set up, run, and fine-tune open-source large language models (LLMs), AI systems trained to understand and generate text, on a Linux system. The goal is to make it accessible to students and researchers who want to work with these models locally rather than relying on paid API services. The tutorials cover three main stages: configuring the hardware and software environment, deploying models so you can actually chat with them, and fine-tuning, the process of adapting a general-purpose model to a specific task or style using your own data. Fine-tuning techniques covered include LoRA (a lightweight method that adjusts only a small portion of the model's parameters) and full-parameter training. Over 50 open-source language models are supported, including models like LLaMA, ChatGLM, Qwen, InternLM, and DeepSeek. Each model has its own step-by-step guide. The tutorials are written primarily in Chinese and presented as Jupyter Notebooks, interactive documents that mix explanatory text and runnable code. You would use this project if you want hands-on experience running AI models on your own hardware without needing to pay for cloud AI API access.