Enhancing Decision-Making of Large Language Models via Actor-Critic

Motivation ▪ LLMs are powerful in NLP but struggle in complex decision-making tasks. ▪ Challenges: ▪ Limited long-term reasoning (auto-regressive bias). ▪ Fragile rollout-based planning. ▪ Goal: Enable robust and scalable decision-making with LLMs using reinforcement learning insights.