huggingface/trl

Train transformer language models with reinforcement learning.