This is the first paper to explore how to effectively use RL for MLLMs and introduce Vision-R1, a reasoning MLLM that leverages cold-start initialization and RL training to incentivize reasoning capability.
active 2025-03-08 → 2026-07-05 (UTC)
Activity over time
Daily event counts in the loaded window
Line chart, 485 days from 2025-03-08 to 2026-07-05. Pushes: 16 total, peak 4 in a day. Pull requests: 8 total, peak 8 in a day. Issues: 18 total, peak 4 in a day. Comments: 21 total, peak 4 in a day. Stars: 417 total, peak 33 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 |
|---|---|---|---|---|
| Osilly | 33 | 12 | 0 | 12 |
| BhJia | 12 | 4 | 8 | 0 |
| Ustinianer | 3 | 0 | 0 | 1 |
| TUDelftHao | 2 | 0 | 0 | 1 |
| YFCodeDream | 2 | 0 | 0 | 1 |
| zhouyanzhong18 | 2 | 0 | 0 | 1 |
| Naieo | 1 | 0 | 0 | 1 |
| LiuRicky | 1 | 0 | 0 | 0 |
| Forensax | 1 | 0 | 0 | 0 |
| peterant330 | 1 | 0 | 0 | 1 |
| ICE-Bruce | 1 | 0 | 0 | 1 |
| ArmenJeddi | 1 | 0 | 0 | 0 |
| dragen1860 | 1 | 0 | 0 | 0 |
| FoolishMao | 1 | 0 | 0 | 1 |
| NielsRogge | 1 | 0 | 0 | 1 |
Recent activity
Latest issues, pull requests and releases
- Issue#31Osilly2026-01-25 21:32ray爆显存问题
- Issue#32Osilly2026-01-25 21:32RL训练阶段脚本是可以直接使用EasyR1的吗？
- Issue comment#30Osilly2025-09-28 06:40RL阶段使用的数据
- Issue comment#29Osilly2025-09-16 14:48PTST优化问题
- Issue#28LiuRicky2025-08-27 14:25冷启动数据集来源
- Issue#27Osilly2025-08-27 07:17Some inquiries regarding the evaluation
- Issue comment#27YFCodeDream2025-08-24 04:38Some inquiries regarding the evaluation
- Issue#27YFCodeDream2025-08-23 05:50Some inquiries regarding the evaluation
- Issue comment#5NielsRogge2025-08-21 19:44Release artifacts (models, dataset) on Hugging Face
- Issue#5Osilly2025-08-21 10:29Release artifacts (models, dataset) on Hugging Face
- Issue comment#5Osilly2025-08-21 10:21Release artifacts (models, dataset) on Hugging Face
- Issue comment#12peterant3302025-08-21 05:03Request to update experiment scripts, code and data
- Issue#26dragen18602025-08-18 11:01RL verl training scripts?
- Issue comment#16Osilly2025-08-17 08:32RL训练脚本
- Issue comment#25Osilly2025-08-17 08:29冷启动数据集的效果问题？
- Issue#25zhouyanzhong182025-08-17 02:43冷启动数据集的效果问题？
- Issue comment#16zhouyanzhong182025-08-17 02:34RL训练脚本
- Issue#24ArmenJeddi2025-07-31 15:54Trying distillation like R1
- Issue#17Osilly2025-06-29 10:57Vision-R1-cold Prompt
- Issue#9Osilly2025-06-29 10:56数据集开源请求
- Issue#11Osilly2025-06-29 10:56上传一下权重，让家人们尝尝咸淡吧
- Issue#23Osilly2025-06-29 10:56能否发布一下sft冷启之后的模型权重和rl阶段使用的数据集？
- Issue comment#17Osilly2025-06-26 13:13Vision-R1-cold Prompt
- Issue comment#12Osilly2025-06-26 13:11Request to update experiment scripts, code and data
- Issue comment#9Osilly2025-06-26 11:25数据集开源请求
Totals cover only the window loaded into ClickHouse and count events, not GitHub's lifetime totals — 417 stars here means stars gained during the window, not the repo's star count.