🧠「大模型」2小时完全从0训练64M的小参数LLM！Train a 64M-parameter LLM from scratch in just 2h!
Python · active 2025-05-04 → 2026-08-05 (UTC)
Activity over time
Daily event counts in the loaded window
Line chart, 459 days from 2025-05-04 to 2026-08-05. Pushes: 69 total, peak 10 in a day. Pull requests: 25 total, peak 3 in a day. Issues: 220 total, peak 34 in a day. Comments: 282 total, peak 16 in a day. Stars: 10,560 total, peak 441 in a day.
- Pushes
- Pull requests
- Issues
- Comments
- Stars
Stars, PRs, issues and forks are under-captured in the later part of this window. GH Archive progressively stopped capturing non-push events during 2026 — −95% or worse by the end of the window. Every series here except Pushes fades for that reason, so a decline above reflects the archive, not this repository. Pushes stay reliable throughout, so read them, and the contributor counts derived from them, as the real signal. Data health has the measurements.
Top contributors
Pushes, PRs, issues, reviews and comments — stars and forks excluded, so this is contribution rather than popularity
| Contributor | Contributions | Pushes | PRs | Comments |
|---|---|---|---|---|
| jingyaogong | 245 | 69 | 0 | 79 |
| dao86 | 19 | 0 | 0 | 19 |
| tangjinou | 15 | 0 | 0 | 9 |
| ciaoyizhen | 10 | 0 | 0 | 7 |
| dyhuachi | 10 | 0 | 2 | 8 |
| AnnaPatio | 8 | 0 | 0 | 5 |
| yuyu5333 | 8 | 0 | 1 | 6 |
| arthaschan | 7 | 0 | 0 | 2 |
| ds-new | 6 | 0 | 0 | 4 |
| github-luffy | 6 | 0 | 0 | 1 |
| gaochao0609 | 5 | 0 | 0 | 5 |
| zhenyu-02 | 4 | 0 | 0 | 4 |
| lwdnxu | 4 | 0 | 0 | 2 |
| Cui-ruochu | 4 | 0 | 0 | 4 |
| huangchenzhao | 3 | 0 | 0 | 1 |
| GWwangshuo | 3 | 0 | 0 | 2 |
| Copilot | 3 | 0 | 0 | 1 |
| hmyjyghh | 3 | 0 | 0 | 3 |
| hahazei | 3 | 0 | 0 | 2 |
| xuanxh1 | 3 | 0 | 0 | 1 |
Recent activity
Latest issues, pull requests and releases
- Issue comment#786fanshengzhe972026-05-31 11:42[not issue] 训练实验数据分享
- Issue#787fanshengzhe972026-05-31 03:27LoRA实现似乎有些问题
- Issue comment#770yanghw1162026-05-25 09:32适配昇腾NPU
- Pull request#780shiguol2026-05-25 02:17
- Issue#768jingyaogong2026-05-17 05:37关于v2版本的数据
- Issue comment#768jingyaogong2026-05-17 05:37关于v2版本的数据
- Issue#766jingyaogong2026-05-16 10:09建议将requirements 中 modelscope可以更新为1.37.0
- Issue#758TKiteRunner2026-05-06 06:17[Bug] CUDA OOM in train_grpo.py on GPUs with ≤8GB VRAM
- Issue comment#755tangjinou2026-05-05 05:16GRPO训练的效果不收敛，reward和loss等震荡严重
- Issue comment#755tangjinou2026-05-04 22:52GRPO训练的效果不收敛，reward和loss等震荡严重
- Issue comment#755tangjinou2026-05-04 15:48GRPO训练的效果不收敛，reward和loss等震荡严重
- Issue comment#755tangjinou2026-05-04 00:55GRPO训练的效果不收敛，reward和loss等震荡严重
- Issue comment#753jorel6662026-05-03 03:19Main
- Issue comment#751Captainphiora2026-04-29 14:51关于训练阶段是否传入padding_mask
- Issue comment#749shinygem737-art2026-04-28 07:03train_pretrain中args.accumulation_steps梯度累积部分的疑惑
- Issue#751kustcl2026-04-28 00:49关于训练阶段是否传入padding_mask
- Issue#748jingyaogong2026-04-25 06:14看了大佬的新版本中去掉了推理冷启动的部分，背后有什么考量嘛
- Issue#748orzorzor22026-04-25 06:00看了大佬的新版本中去掉了推理冷启动的部分，背后有什么考量嘛
- Issue comment#746JJASMINE222026-04-24 02:19关于SFT-mini数据集的疑问
- Issue#746JJASMINE222026-04-24 00:47关于SFT-mini数据集的疑问
- Issue#743aivolcano2026-04-19 15:02How to execute train_grpo.py via sglang engine?
- Issue#740jingyaogong2026-04-19 06:36Which version of sglang is recommended to install? the requirements.txt does not show this packages.
- Issue comment#740jingyaogong2026-04-19 06:36Which version of sglang is recommended to install? the requirements.txt does not show this packages.
- Issue#740aivolcano2026-04-19 06:17Which version of sglang is recommended to install? the requirements.txt does not show this packages.
- Issue#739Kydon-ai2026-04-18 14:57关于模型参数大小和模型对话表现的疑问
Totals cover only the window loaded into ClickHouse and count events, not GitHub's lifetime totals — 10,560 stars here means stars gained during the window, not the repo's star count.