Open-Source LLM Cookbook
A hands-on, Chinese-language tutorial for configuring, deploying and fine-tuning open-source large language models on Linux, covering more than 50 mainstream Chinese and international models.
The tutorial walks through Linux environment setup for different open-source LLMs, deployment and usage guidance for major models such as LLaMA, ChatGLM and InternLM, including command-line invocation, online demo deployment and LangChain integration, and full-parameter and parameter-efficient fine-tuning methods such as distributed full fine-tuning, LoRA and P-tuning. It recommends learning environment setup first, then deployment and usage, and fine-tuning last.
It documents deployment and fine-tuning steps for more than 50 mainstream models, including Qwen, GLM, MiniCPM, DeepSeek, InternLM, Llama, Gemma and Baichuan families, with a separate section for AMD GPUs, and includes example applications such as a chat model fine-tuned to imitate a television character's speaking style and a model trained on advanced-mathematics problem sets. It is aimed at students, researchers and other learners with some technical background who want to deploy and fine-tune open-source LLMs at low cost.
It is delivered mainly as Jupyter notebooks and licensed under Apache-2.0.