NVIDIA-NeMo/RL
Scalable toolkit for efficient model reinforcement
Build with Backblaze B2
SDKs, agent skills, IDE extensions, and reference pipelines from Backblaze Labs. All open source.
Loading star history...
Where it stands
Rank, growth pace and release cadence relative to everything GStars tracks.
Ranks #414 of 2,845 repositories GStars tracks this week (top 15%).
1 release in the last 90 days, the latest 78 days ago, with 174 contributors on GitHub.
Gained 14% fewer stars in the last 90 days than in the 90 days before.
Use Cases & Benefits
- Provides a scalable and efficient post-training reinforcement learning toolkit for large language models from 1 GPU to thousands and up to 100B+ parameters.
- Enables high-performance distributed training with advanced parallelism and seamless Hugging Face integration for flexible, large-scale model fine-tuning and evaluation.
- Use for training and fine-tuning large language models with reinforcement learning algorithms like GRPO, SFT, and DPO on single or multi-node GPU clusters.
- Use for multi-turn reinforcement learning tasks such as tool use or game playing with support for environment and worker isolation to manage complex workflows.
- Use for converting and evaluating trained models in Hugging Face format with optimized inference backends and scalable resource management using Ray.
Should you depend on it?
License terms, project hygiene and risks, from GitHub metadata and GStars' snapshots.
License
Apache 2.0
Permissive: commercial use, modification and redistribution are allowed with attribution; includes an explicit patent grant from contributors.
Community files
- README: present
- Contributing guide: present
Watch-outs
No watch-outs from GStars' checks.
General guidance from repository metadata, not legal advice. How we score
From the README
Nemo RL is a scalable and efficient post-training library designed for models ranging from 1 GPU to thousands, and from tiny to over 100 billion parameters.
✅ Available now | 🔜 Coming in v0.4
Clone NeMo RL.
If you are using the Megatron backend on bare-metal (outside of a container), you may need to install the cudnn headers as well. Here is how you can check as well as install them:
For faster setup and environment isolation, we use uv. Follow these instructions to install uv.
Then, initialize NeMo RL project virtual environment via:
If working outside a container, it can help to build flash-attn and warm the uv cache before your first run.
If sucessful, you should see
✅ flash-attn successfully added to uv cache.Use
uv runto launch all commands. It handles pip installing implicitly and ensures your environment is up to date with our lock file.
Daily snapshots
Stars gained per day, and how forks and open issues moved, from GStars' daily snapshots
Forks
600+121
Open issues
1.3K+522
Compared with peers in Backend
Fastest-growing repositories GStars tracks in this cohort for the week of Oct 11, by 7-day star growth
| Repository | Stars | +7d | Health | Last release |
|---|---|---|---|---|
| NationalSecurityAgency/ghidra | 94.0K | +7.5K | 75 | 2 days ago |
| DuarteSantos8/openGym | 15.5K | +4.4K | 75 | 1 day ago |
| K-Dense-AI/scientific-agent-skills | 54.2K | +3.2K | 75 | 22 days ago |
| 666ghj/MiroFish | 80.7K | +2.4K | 23 | 226 days ago |
| anthropics/claude-plugins-official | 44.4K | +2.3K | 75 | — |
| NVIDIA-NeMo/RL | 2.1K | +0 | 75 | 78 days ago |
| Category median | 54.2K | +3.2K | 75 | — |
Also by NVIDIA-NeMo
Other repositories from this maintainer that GStars tracks
- Python18.6K stars
Speech
A scalable generative AI framework built for researchers and developers working on Large Language Models, Multimodal, and Speech AI (Automatic Speech Recognition and Text-to-Speech)
- Rust3.4K stars
Switchyard
Switchyard lets LLM applications route traffic across models and providers while preserving native OpenAI and Anthropic API compatibility - enabling flexible model selection, benchmarking, and cost/performance optimization.
- Python2.3K stars
DataDesigner
🎨 NeMo Data Designer: Generate high-quality synthetic data from scratch or from seed data.
- Python1.2K stars
Gym
Evaluate and improve models and agents using environments
- Python1.0K stars
Automodel
🚀 Pytorch Distributed native training library for LLMs/VLMs with OOTB Hugging Face support
Discover Repositories
Search across tracked repositories by name or description