TRL - Transformer Reinforcement Learning

TRL: SFT, DPO, GRPO, RLOO reward modeling for LLM RLHF.