datawhalechina/llm-universe

★ 12,988Jupyter NotebookAudience · developerComplexity · 2/5Setup · easy

Mindmap

mindmap
  root((llm-universe))
    Course Parts
      LLM basics
      Advanced RAG
      Real projects
    Hands-on Project
      Knowledge base bot
      Document queries
    LLM Providers
      ChatGPT
      Baidu Wenxin
      Zhipu GLM
    Audience
      Python beginners
      AI newcomers

mindmap root((llm-universe)) Course Parts LLM basics Advanced RAG Real projects Hands-on Project Knowledge base bot Document queries LLM Providers ChatGPT Baidu Wenxin Zhipu GLM Audience Python beginners AI newcomers

Click or tap to explore — scroll the page freely

Things people build with this

USE CASE 1

Build a chatbot that answers questions from your own documents using the RAG technique covered in the course.

USE CASE 2

Learn to call ChatGPT and Chinese AI APIs like Baidu Wenxin or Zhipu GLM from Python code.

USE CASE 3

Follow a structured zero-to-app course to build a working LLM application without needing machine learning theory.

Tech stack

PythonJupyter NotebookRAG

Getting it running

Difficulty · easy Time to first run · 30min

Primary content is in Chinese, no GPU or specialized hardware needed, runs on a laptop or free cloud server.

In plain English

LLM Universe is a Chinese-language tutorial series from Datawhale, a Chinese open-source AI learning community, aimed at beginners who want to learn how to build applications using large language models. The project is written primarily in Chinese and targets developers who know basic Python but have no background in machine learning or AI theory. The course is structured in three parts. The first part covers the fundamentals: what large language models are, what RAG means (a technique for giving an AI model access to your own documents so it can answer questions about them), and how to call the APIs provided by various AI services. The second part covers more advanced RAG techniques, including how to process different types of documents, optimize search results, and fine-tune the retrieval pipeline. The third part analyzes real open-source AI application projects to show what kinds of things can actually be built. The hands-on project used throughout the course is a personal knowledge base assistant: an AI chatbot that can answer questions by searching through a collection of documents you provide. This gives learners a concrete, practical goal to work toward rather than just reading theory. The course supports multiple AI providers, including ChatGPT and several Chinese-market models such as Baidu Wenxin, iFlytek Spark, and Zhipu GLM. The unified wrapper code means you can switch between providers without rewriting your application logic. No GPU or specialized hardware is needed. The course can be completed on a personal laptop or on a free cloud server. All materials are available as Jupyter notebooks and as a readable website, and a PDF download of the first completed section is also provided.

Copy-paste prompts

Prompt 1

Using the llm-universe RAG approach, show me how to build a Python pipeline that answers questions from a PDF document.

Prompt 2

How do I switch between ChatGPT and Baidu Wenxin using the unified API wrapper from llm-universe?

Prompt 3

Walk me through setting up a personal knowledge base assistant in Python, following the llm-universe project structure.

Prompt 4

What is RAG and how does it work? Use the llm-universe knowledge base chatbot as a concrete example.

Open on GitHub → Explain another repo

← datawhalechina on gitmyhub — every repo by this author, as a profile.

Verify against the repo before relying on details.