# Enping-Hu/minimind-deep-dive

<https://repo.masters-masters.com/en/repos/Enping-Hu/minimind-deep-dive>

[Home](https://repo.masters-masters.com/en) › [Enping-Hu](https://repo.masters-masters.com/en/owners/Enping-Hu) › [minimind-deep-dive](https://repo.masters-masters.com/en/repos/Enping-Hu/minimind-deep-dive)

Enping-Hu/minimind-deep-dive has 217 stars on GitHub. Its RepoMasters card: Prototype tier. 从 MiniMind 源码读起,再延伸到现代大模型技术体系的中文学习笔记。主线逐行精读预训练 / SFT / DPO / PPO /…

| Field | Value |
| --- | --- |
| Owner | [Enping-Hu](https://repo.masters-masters.com/en/owners/Enping-Hu) |
| Stars | 217 |
| Forks | 14 |
| Language | [Python](https://repo.masters-masters.com/en/languages/python) |
| License | CC-BY-4.0 |
| Created | 2026 |
| Topics | deep-learning, dpo, fine-tuning, from-scratch, grpo, large-language-models, llm, llm-training, minimind, moe |
| Tier | [Prototype (100 to 249 ★)](https://repo.masters-masters.com/en/ranks/rush) |
| View on GitHub | [Enping-Hu/minimind-deep-dive](https://github.com/Enping-Hu/minimind-deep-dive) |

## Repository

从 MiniMind 源码读起,再延伸到现代大模型技术体系的中文学习笔记。主线逐行精读预训练 / SFT / DPO / PPO / GRPO 与训练机制;附录 17 篇进阶卷覆盖量化、投机解码、RLHF 全景、模型代际史等 MiniMind 没涉及、但进阶绕不开的主题。

The tier for repos with 100 to 249 ★. Published odds per card pulled: 70%. Rarity: Common.

## Nearby in Python

- [lee-dro/PatchCore · ★ 217 · Python · Prototype](https://repo.masters-masters.com/en/repos/lee-dro/PatchCore)
- [beckettharriman/openfill · ★ 217 · Python · Prototype](https://repo.masters-masters.com/en/repos/beckettharriman/openfill)
- [christos-antono/NPCTalk · ★ 217 · Python · Prototype](https://repo.masters-masters.com/en/repos/christos-antono/NPCTalk)
- [pkrumins/xgoogle · ★ 217 · Python · Prototype](https://repo.masters-masters.com/en/repos/pkrumins/xgoogle)
- [sublee/hangulize · ★ 217 · Python · Prototype](https://repo.masters-masters.com/en/repos/sublee/hangulize)
- [jpulgarin/django-tokenapi · ★ 217 · Python · Prototype](https://repo.masters-masters.com/en/repos/jpulgarin/django-tokenapi)

---
Data: GitHub · Illustration drawn by RepoMasters from the repo’s ID and its language color.