🎖️

TRL

Training & Fine-tuning

Transformer Reinforcement Learning for RLHF and alignment.

10kstars1.3kforksPython

About

TRL by Hugging Face provides tools for RLHF, DPO, PPO, and reward modeling to align language models with human preferences.

Streamlined fine-tuning of LLMs with PEFT and full-param support.

Parameter-efficient fine-tuning methods by Hugging Face.

Deep learning optimization library for large-scale training.