Leandro von Werra

TRL library; RLHF tutorials

Machine learning engineer at Hugging Face, where he leads work on post-training and alignment tooling. He created and maintains TRL (Transformer Reinforcement Learning), the widely adopted open-source library for fine-tuning LLMs with PPO, DPO, and related RL methods. He is also known for accessible tutorials demystifying RLHF for the broader ML community.