# Tongyun1/from-minimind-to-more

🎓从0开始训练一个大模型Minimind项目的超详细解析，包括但不限于用到的架构，算法，以及大模型面试经验

Repository: https://github.com/Tongyun1/from-minimind-to-more
Canonical: https://ross.abutalabs.com/products/from-minimind-to-more
Language: Python
License Family: other
Last push: 2026-05-25T14:40:35+00:00

## Health v2 (maintenance only)
Score: 53/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 84, release rhythm 35, longevity 15
- inputs: {"age_days": 219, "days_push": 100, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases, no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1652, forks 89 (observed 2026-08-28T04:05:17.005049+00:00)

## What it is
A detailed Chinese-language study guide and annotated walkthrough of the Minimind project for training a large language model from scratch, covering tokenization, Transformer architecture, and alignment algorithms like SFT, DPO, PPO, GRPO, and SPO. It also includes LLM job interview preparation materials and is actively updated.

## Use cases
- learn how to train a large language model from scratch
- understand transformer architecture details like MoE, KV Cache, and Flash Attention
- study how DPO, PPO, and GRPO alignment algorithms work with source code
- prepare for LLM engineer interview questions
- read heavily annotated source code of a small LLM training project
- understand tokenizers and positional embeddings in depth

## When to choose
- you want a structured, in-depth tutorial companion to the Minimind codebase
- you are preparing for LLM-related job interviews
- you learn best from annotated source code and theory explanations in Chinese

## When to avoid
- you need production-ready LLM training code rather than educational notes
- you need English-language materials
- you need coverage of LoRA and distillation, which are not yet written

## Facets
- artifact type: learning-resource
- maturity: active
- function: llm-training, machine-learning, deep-learning, documentation
- domain: large-language-models, deep-learning, machine-learning, tutorials, education
- platform: python
- tags: llm-from-scratch, minimind, transformer-architecture, rlhf, dpo, ppo, grpo, sft, pretraining, interview-preparation, study-notes, chinese

## Member repositories
- Tongyun1/from-minimind-to-more (main) score 53

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:05:17.005049+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 2026-08-30T03:44:59.455711+00:00, confidence not recorded.
  - readme: https://github.com/Tongyun1/from-minimind-to-more (fetched 2026-08-28T04:05:17.005049+00:00, sha 318ef296cdee)
- Data as of 2026-08-30T08:39:29.467469+00:00.