GPT-Critic: Offline Reinforcement Learning for End-to-End Task-Oriented Dialogue Systems.

Youngsoo Jang, Jongmin Lee, Kee-Eung Kim

Browse the full ICLR paper archive.