TRL
Hugging Face’s Apache-2.0 trainers for SFT, DPO, PPO, and reward modeling.
Fit.
Catalog snapshot
- Platforms
- Linux, Macos, Windows
- Operating systems
- Linux, macOS, Windows
- Ram floor
- 16 GB
- Gpu
- Required · 8 GB VRAM
- Docker
- No
- Setup
- Advanced
- Deploy as
- Local, Self-hosted
- Commercial use
- Allowed
A single 8 GB card handles small-model LoRA; full SFT wants more.
Project.
Fetched 1
About.
TRL is Hugging Face’s trainer library for the full post-training stack: SFT, DPO, PPO, and reward modeling on top of transformers and PEFT. Apache-2.0 and the reference implementation most alignment papers build on.
SFT trainer · DPO and PPO · reward modeling · HF Trainer integration
