Google, Heidelberg University & NEC Propose Human Feedback for Real-World RL in NLP Systems | Synced

A new study proposes using human feedback and interaction logs to boost offline reinforcement learning (RL) in natural language processing (NLP).