verl
verl/HybridFlow: A Flexible and Efficient RL Post-Training Framework
verl scores 7.6/10 (Strong) in Fine-Tuning.
Language Python License Apache-2.0 Stars 22.3k★ Type heavyweight Self-host Yes Repo verl-project/verl Homepage https://verl.readthedocs.io
Quality
Strong 7.6
Adoption 59
Activity 90
Maturity 88
Community 66
Capability 50
Show the math
Overall 7.6/10
value = 7.6/10
score = value
Adoption 6.3/10
value = 58.7/100
score = 1 + 9 * value / 100
Activity 9.1/10
value = 90/100
score = 1 + 9 * value / 100
Maturity 8.9/10
value = 88/100
score = 1 + 9 * value / 100
Community 6.9/10
value = 65.7/100
score = 1 + 9 * value / 100
Capability 5.5/10
value = 50/100
score = 1 + 9 * value / 100
Key metrics
22.3k★ GitHub stars
404/90d Recent commits
65 devs/90d Recent contributors
1.7y Project age
active Last commit
Gotchas
No gotchas documented yet. Contribute one if you know a constraint we missed.
Labels
- Self-hosted
Self-hosted
You deploy and operate it yourself; there is no hosted option here.
- Heavyweight
Heavyweight
Resource-heavy to run (GPU, large memory, or multi-service stack).
rl grpo agentic-rl ray post-training