# MiniMind

> Train a small LLM from scratch in two hours

Code and guides to pretrain and fine-tune a 64M-parameter language model on a single consumer GPU.

- Source: https://github.com/jingyaogong/minimind
- Homepage: https://jingyaogong.github.io/minimind
- License: Apache-2.0
- Language: Python
- Stars: 63142
- Forks: 8192
- Contributors: 20
- Last commit: 2026-09-22
- Latest release: v2 (2025-10-21)
- Purpose: AI & LLM tooling
- Runs on: Library / SDK
- For: Personal

## Worth score: 74/100

How much DigGitHub recommends it, from activity, adoption, docs, license and security signals.

- Popularity: 24/25
- Momentum: 10/20
- Maintenance: 21/25
- Community: 11/15
- Readiness: 8/15

---

Source page: https://diggithub.com/jingyaogong/minimind
Updated: 2026-10-04