Adapting LLM Agents with Universal Feedback in Communication

arXiv:2310.01444v3 Announce Type: replace
Abstract: Recent advances in large language models (LLMs) have demonstrated potential for LLM agents. To facilitate the training for these agents with both linguistic feedback and non-linguistic reward signals, we introduce Learning through Communication (LTC). We design a universal buffer to store all the feedback, and an iterative pipeline to enable an LLM agent to explore and update its policy in an given environment. To optimize agent interactions for task-specific learning with our universal buffer and …

abstract advances agents arxiv communication cs.ai cs.cl design feedback iterative language language models large language large language models llm llms store through training type universal