verl: Volcano Engine Reinforcement Learning for LLMs——PJ internal usage.
Python · active 2025-04-05 → 2026-01-27 (UTC)
Activity over time
Daily event counts in the loaded window
Line chart, 298 days from 2025-04-05 to 2026-01-27. Pushes: 1,025 total, peak 39 in a day. Pull requests: 1,945 total, peak 23 in a day. Issues: 1,133 total, peak 15 in a day. Comments: 7,257 total, peak 78 in a day. Stars: 6,039 total, peak 83 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 |
|---|---|---|---|---|
| vermouth1992 | 1.7K | 375 | 385 | 381 |
| gemini-code-assist[bot] | 1.7K | 0 | 0 | 1.1K |
| eric-haibin-lin | 1.1K | 199 | 210 | 359 |
| wuxibin89 | 884 | 180 | 67 | 275 |
| zhaochenyang20 | 331 | 43 | 41 | 164 |
| HollowMan6 | 274 | 0 | 44 | 133 |
| tongyx361 | 260 | 108 | 44 | 43 |
| FightingZhen | 225 | 14 | 26 | 124 |
| CLAassistant | 221 | 0 | 0 | 221 |
| ISEEKYAN | 171 | 10 | 15 | 94 |
| chenhaiq | 170 | 8 | 25 | 86 |
| hiyouga | 153 | 15 | 22 | 65 |
| SwordFaith | 130 | 0 | 4 | 76 |
| 0x404 | 98 | 0 | 14 | 69 |
| ccclyu | 97 | 0 | 5 | 54 |
| PeterSH6 | 96 | 15 | 22 | 20 |
| tardis-key | 90 | 0 | 10 | 41 |
| ji-huazhong | 87 | 9 | 23 | 30 |
| yyDing1 | 84 | 6 | 19 | 40 |
| hongpeng-guo | 83 | 0 | 19 | 35 |
Recent activity
Latest issues, pull requests and releases
- Pull request#5055Rhetee2026-01-27 05:03
- Pull request#5053yyyy20002026-01-27 04:12
- Pull request#5054none06632026-01-27 04:07
- Issue comment#4973litianjian2026-01-27 04:00[doc, trainer] fix: shoudn't use rollout routing replay data for R2
- Issue comment#5031wuxibin892026-01-27 03:59[ckpt] feat: add CheckpointEngineManager
- Issue comment#4973litianjian2026-01-27 03:54[doc, trainer] fix: shoudn't use rollout routing replay data for R2
- Issue comment#5007beverly1974-justice02026-01-27 02:05GLM-4.7-flash update transformer==5.0 cannot import name 'AutoModelForVision2Seq' from 'transformers'
- Pull request#5051acisseJZhong2026-01-27 01:47
- Pull request#5049dependabot[bot]2026-01-26 19:21
- Pull request#5049dependabot[bot]2026-01-26 19:21
- Pull request#5048dependabot[bot]2026-01-26 19:21
- Pull request#5048dependabot[bot]2026-01-26 19:21
- Issue comment#5025bithighrr2026-01-26 14:14[rollout, perf, cfg] fix: Add global step info and support more profile control params for rollout profiling (sglang backend)
- Issue comment#4921SchumiDing2026-01-26 08:51(raylet) A worker died or was killed while executing a task by an unexpected system error.
- Issue#5045yangchengping2026-01-26 07:39[Bug] qwen3vl-2B verl megaton save error
- Issue comment#4973HollowMan62026-01-26 07:37[doc, trainer] fix: shoudn't use rollout routing replay data for R2
- Issue comment#5046tardis-key2026-01-26 06:46PTA call acl api failed: Parameter verification failed
- Issue#5045yangchengping2026-01-26 04:13[Bug] qwen3vl-2B verl megaton save error
- Issue comment#5044CLAassistant2026-01-26 03:39[doc]feat: add dapo multi model optimization practice
- Pull request#5043acat-rw2026-01-26 03:26
- Pull request#5041JacobHelwig2026-01-26 01:45
- Issue comment#4897JacobHelwig2026-01-26 01:43[fsdp,trainer,algo] feat: On-Policy Distillation
- Issue comment#4897xvrong2026-01-25 17:35[fsdp,trainer,algo] feat: On-Policy Distillation
- Issue comment#4989liuchaohu2026-01-25 07:09deepeyes的recipe的bug
- Issue comment#5025bithighrr2026-01-25 06:18[rollout, perf, cfg] fix: Add global step info and support more profile control params for rollout profiling (sglang backend)
Totals cover only the window loaded into ClickHouse and count events, not GitHub's lifetime totals — 6,039 stars here means stars gained during the window, not the repo's star count.