PRIME-RL / TTRL
[NeurIPS 2025] TTRL: Test-Time Reinforcement Learning
Why this job: says: reinforcement learning
What it does
The core challenge of the problem is reward estimation during inference while not having access to ground-truth information. While this setting appears elusive, we find that common practices in Test-Time Scaling (TTS), such as majority voting, yield surprisingly effective rewards suitable for driving RL training.
Get started
No single install command. Read the steps in the README (opens in new tab)
- Maintenance
- Inactive
- Last commit 8 months ago
- Stars
- 1,132
- On GitHub
- Licence
- MIT
- Allows business use
- Downloads for
- None listed
- No release files
- Latest tag
- verl
- Released 16 months ago
- Built with
- Python
- 88% of the code
Is it looked after?
- Contributors
- Not stated
- Last 90 days
- Issues
- 67% closed
- 17 still open
- Pull requests
- 1 merged
- 0 waiting
- Releases
- 2
- 0 in the last year
- No contributing guide
- No security policy
- No code of conduct
Similar repos
| Repo | Stars | Last commit | Licence | Downloads for |
|---|---|---|---|---|
| lasgroup / SDPO | 1,145 | 4 months ago | Apache-2.0 | No downloads |
| hiyouga / EasyR1 | 5,214 | 4 days ago | Apache-2.0 | No downloads |
| inclusionAI / Awex | 179 | 44 days ago | Apache-2.0 | No downloads |
More: languages, topics and project record
Languages
- Python 88%
- Shell 11%
- Jupyter Notebook 1%
- Makefile <1%
Topics
- llm
- reasoning
- rl
Project record
- JobFine-tuning
- StatusListed
- Commits20
- Forks83
- Watching17
- Stars in 28 days+6
- CreatedApr 23, 2025
- First seen by usNov 7, 2026
- CheckedNov 8, 2026