For language models, many methods using teacher forcing (TF) to train.

But with offline RL, which shows a clear performance improvement over teacher forcing while not inducing training instability or sacrificing practical training budgets.

Powerpoint for this talk

Reference Paper

• Decision Transformer: Reinforcement Learning via Sequence Modeling
• Offline Reinforcement Learning with Implicit Q-Learning
• Proximal Policy Optimization Algorithms
• A Learning Algorithm for Continually Running Fully Recurrent Neural Networks