ACC-Collab: An Actor-Critic Approach to Multi-Agent LLM Collaboration
This work addresses the limitation of relying on emergent collaboration in multi-agent LLM systems, offering a learned approach for researchers and practitioners in AI and natural language processing.
The paper tackles the problem of improving multi-agent collaboration in large language models by proposing ACC-Collab, an actor-critic learning framework that trains specialized agents, and it demonstrates superior performance over state-of-the-art methods across various benchmarks.
Large language models (LLMs) have demonstrated a remarkable ability to serve as general-purpose tools for various language-based tasks. Recent works have demonstrated that the efficacy of such models can be improved through iterative dialog between multiple models. While these paradigms show promise in improving model efficacy, most works in this area treat collaboration as an emergent behavior, rather than a learned behavior. In doing so, current multi-agent frameworks rely on collaborative behaviors to have been sufficiently trained into off-the-shelf models. To address this limitation, we propose ACC-Collab, an Actor-Critic based learning framework to produce a two-agent team (an actor-agent and a critic-agent) specialized in collaboration. We demonstrate that ACC-Collab outperforms SotA multi-agent techniques on a wide array of benchmarks.