# nndl/llm-beginner

LLM、Agent上手教程

Repository: https://github.com/nndl/llm-beginner
Canonical: https://ross.abutalabs.com/products/llm-beginner
Language: Python
License: MIT
License Family: permissive
Topics: fudannlp, step-by-step, agent, llm, openmoss
Last push: 2026-06-23T09:01:05+00:00

## Health v2 (maintenance only)
Score: 79/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 89, release rhythm 54, longevity 100
- inputs: {"age_days": 3444, "days_push": 71, "days_rel": 97, "gap_med": null, "n_releases_24m": 1}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 6681, forks 1356 (observed 2026-08-28T04:09:47.604074+00:00)

## What it is
A step-by-step beginner tutorial series (with a companion textbook) for learning large language models and agents through six progressive hands-on Python tasks, from implementing a Transformer and mini-GPT to fine-tuning, RAG, and building tool-calling and coding agents. Each task includes data download scripts, self-check evaluation harnesses, and LLM-assisted code review prompts.

## Use cases
- learn how LLMs work by implementing a mini-GPT from scratch
- hands-on tutorial for building AI agents
- practice instruction fine-tuning and DPO alignment
- build a RAG pipeline step by step
- learn transformer architecture with exercises
- beginner exercises for tool-calling agents
- self-study curriculum for large language models

## When to choose
- you have Python and basic deep learning knowledge and want a structured, hands-on LLM/agent curriculum
- you prefer learning by implementing from scratch before using frameworks
- you want self-checkable exercises with evaluation scripts
- you read Chinese and want a tutorial aligned with the NNDL textbook series

## When to avoid
- you need production-ready LLM or agent libraries rather than educational exercises
- you want an English-language tutorial
- you lack a GPU and cannot use quantized models
- you need a quick reference rather than a multi-week course

## Facets
- artifact type: learning-resource
- maturity: active
- function: machine-learning, deep-learning, llm-training, rag, agent-framework, prompt-engineering
- domain: large-language-models, deep-learning, tutorials
- platform: python, cross-platform
- tags: llm, agents, hands-on-tutorial, step-by-step, chinese, transformer, mini-gpt, sft, dpo, rag, tool-calling, coding-agent, nndl, ai-agents, natural-language-processing

## Member repositories
- nndl/llm-beginner (main) score 79

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:09:47.604074+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T17:43:00.279823+00:00, confidence not recorded.
  - readme: https://github.com/nndl/llm-beginner (fetched 2026-08-28T04:09:47.604074+00:00, sha 160921f6020b)
- Data as of 2026-08-30T08:39:29.467469+00:00.