TRL

Other Free

Transformer Reinforcement Learning library for model alignment.

About TRL

TRL supports PPO, DPO and other preference optimization techniques.