Language Assistance in Reinforcement Learning in Dynamic Environments

Language is an intuitive and effective way for humans to communicate. Large Language Models (LLMs) can interpret and respond well to language. However, their use in deep reinforcement learning is limited as they are sample inefficient. State-of-the-art deep reinforcement learning algorithms are more sample efficient but cannot understand...