New
huggingface/trl v1.16.0 — post-training with SFT·DPO·GRPO
huggingface/trl is a library for post-training models on top of the Transformers ecosystem, providing SFT, DPO, GRPO and KTO trainers plus a CLI, and its latest release is v1.16.0.
huggingface/trl
huggingface/trl is a library for post-training models on top of the Transformers ecosystem, providing SFT, DPO, GRPO and KTO trainers plus a CLI, and its latest release is v1.16.0.
View on GitHub ↗- stars
- ★ 19.6K
- Forks
- 3.1K
- Language
- Python
- License
- Apache-2.0
- Latest release
- v1.16.0
Posts
1No matching stories.