DaoyuanLi2816 / mini-verl
verl for a single consumer GPU. PPO, GRPO and on-policy distillation on NVIDIA GPUs.
Why this job: says: distillation; topics: reinforcement-learning, peft
What it does
This Qwen3-0.6B example trains with a simple length reward. Inspect it in rewards.jsonl, or follow the five-minute workflow to use GRPO, resume training and export an adapter.
Get started
No single install command. Read the steps in the README (opens in new tab)
- Maintenance
- Active
- Last commit 56 days ago
- Stars
- 310
- +2 this week
- Licence
- Apache-2.0
- Allows business use
- Downloads for
- None listed
- No release files
- Latest version
- v0.16.0
- Released 59 days ago
- Built with
- Python
- 100% of the code
Is it looked after?
- Contributors
- 1
- Last 90 days
- Issues
- 75% closed
- 1 still open
- Pull requests
- 112 merged
- 0 waiting
- Releases
- 28
- 28 in the last year
Similar repos
| Repo | Stars | Last commit | Licence | Downloads for |
|---|---|---|---|---|
| MakazhanAlpamys / Soup | 10,385 | yesterday | Apache-2.0 | No downloads |
| hiyouga / LlamaFactory | 75,965 | yesterday | Apache-2.0 | No downloads |
| AgentR1 / Agent-R1 | 1,718 | 3 months ago | MIT | No downloads |
More: languages, topics, project record and preview image
Languages
- Python 100%
- Jinja <1%
- PowerShell <1%
Topics
- agentic-rl
- consumer-gpu
- knowledge-distillation
- llm
- llm-agents
- on-policy-distillation
- post-training
- qlora
- qwen
- tool-use
- verl
- single-gpu
Project record
- JobFine-tuning
- StatusListed
- Commits192
- Forks75
- Watching53
- Stars in 28 days+4
- CreatedJul 27, 2026
- First seen by usNov 7, 2026
- CheckedNov 10, 2026