Existing multi-agent PPO algorithms lack compatibility with different types of parameter sharing when extending the theoretical guarantee of PPO to cooperative multi-agent reinforcement learning (MARL). In this paper, we propose a novel and versatile multi-agent PPO algorithm for cooperative MARL to overcome this limitation. Our approach is achieved upon the proposed full-pipeline paradigm, which establishes multiple parallel optimization pipelines by employing various equivalent decompositions of the advantage function. This procedure successfully formulates the interconnections among agents in a more general manner, i.e., the interconnections among pipelines, making it compatible with diverse types of parameter sharing. We provide a solid theoretical foundation for policy improvement and subsequently develop a practical algorithm called Full-Pipeline PPO (FP3O) by several approximations. Empirical evaluations on Multi-Agent MuJoCo and StarCraftII tasks demonstrate that FP3O outperforms other strong baselines and exhibits remarkable versatility across various parameter-sharing configurations.

Multi-agent reinforcement learning (MARL) has been shown effective for cooperative games in recent years. However, existing state-of-the-art methods face challenges related to sample complexity, training instability, and the risk of converging to a suboptimal Nash Equilibrium. In this paper, we propose a unified framework for learning stochastic policies to resolve these issues. We embed cooperative MARL problems into probabilistic graphical models, from which we derive the maximum entropy (MaxEnt) objective for MARL. Based on the MaxEnt framework, we propose Heterogeneous-Agent Soft Actor-Critic (HASAC) algorithm. Theoretically, we prove the monotonic improvement and convergence to quantal response equilibrium (QRE) properties of HASAC. Furthermore, we generalize a unified template for MaxEnt algorithmic design named Maximum Entropy Heterogeneous-Agent Mirror Learning (MEHAML), which provides any induced method with the same guarantees as HASAC. We evaluate HASAC on six benchmarks: Bi-DexHands, Multi-Agent MuJoCo, StarCraft Multi-Agent Challenge, Google Research Football, Multi-Agent Particle Environment, and Light Aircraft Game. Results show that HASAC consistently outperforms strong baselines, exhibiting better sample efficiency, robustness, and sufficient exploration. See our project page at \url{https://sites.google.com/view/meharl}.

Offline reinforcement learning (RL) leverages previously collected data to extract policies that return satisfying performance in online environments. However, offline RL suffers from the distribution shift between the offline dataset and the online environment. In the multi-agent RL (MARL) setting, this distribution shift may arise from the nonstationary opponents (exogenous agents beyond control) in the online testing who display distinct behaviors from those recorded in the offline dataset. Hence, the key to the broader deployment of offline MARL is the online adaptation to nonstationary opponents. Recent advances in large language models have demonstrated the surprising generalization ability of the transformer architecture in sequence modeling, which prompts one to wonder \textit{whether the offline-trained transformer policy adapts to nonstationary opponents during online testing}. This work proposes the self-confirming loss (SCL) in offline transformer training to address the online nonstationarity, which is motivated by the self-confirming equilibrium (SCE) in game theory. The gist is that the transformer learns to predict the opponents' future moves based on which it acts accordingly. As a weaker variant of Nash equilibrium (NE), SCE (equivalently, SCL) only requires local consistency: the agent's local observations do not deviate from its conjectures, leading to a more adaptable policy than the one dictated by NE focusing on global optimality. We evaluate the online adaptability of the self-confirming transformer (SCT) by playing against nonstationary opponents employing a variety of policies, from the random one to the benchmark MARL policies. Experimental results demonstrate that SCT can adapt to nonstationary opponents online, achieving higher returns than vanilla transformers and offline MARL baselines.

We consider the problem of locating a facility to serve a set of agents located along a line. The Nash welfare objective function, defined as the product of the agents' utilities, is known to provide a compromise between fairness and efficiency in resource allocation problems. We apply this welfare notion to the facility location problem, converting individual costs to utilities and analyzing the facility placement that maximizes the Nash welfare. We give a polynomial-time approximation algorithm to compute this facility location, and prove results suggesting that it achieves a good balance of fairness and efficiency. Finally, we take a mechanism design perspective and propose a strategy-proof mechanism with a bounded approximation ratio for Nash welfare.

In cooperative multi-agent reinforcement learning, a team of agents works together to achieve a common goal. Different environments or tasks may require varying degrees of coordination among agents in order to achieve the goal in an optimal way. The nature of coordination will depend on the properties of the environment -- its spatial layout, distribution of obstacles, dynamics, etc. We term this variation of properties within an environment as heterogeneity. Existing literature has not sufficiently addressed the fact that different environments may have different levels of heterogeneity. We formalize the notions of coordination level and heterogeneity level of an environment and present HECOGrid, a suite of multi-agent RL environments that facilitates empirical evaluation of different MARL approaches across different levels of coordination and environmental heterogeneity by providing a quantitative control over coordination and heterogeneity levels of the environment. Further, we propose a Centralized Training Decentralized Execution learning approach called Stateful Active Facilitator (SAF) that enables agents to work efficiently in high-coordination and high-heterogeneity environments through a differentiable and shared knowledge source used during training and dynamic selection from a shared pool of policies. We evaluate SAF and compare its performance against baselines IPPO and MAPPO on HECOGrid. Our results show that SAF consistently outperforms the baselines across different tasks and different heterogeneity and coordination levels. We release the code for HECOGrid as well as all our experiments.

Large language model (LLM) agents have been shown effective on a wide range of tasks, and by ensembling multiple LLM agents, their performances could be further improved. Existing approaches employ a fixed set of agents to interact with each other in a static architecture, which limits their generalizability to various tasks and requires strong human prior in designing these agents. In this work, we propose to construct a strategic team of agents communicating in a dynamic interaction architecture based on the task query. Specifically, we build a framework named Dynamic LLM-Agent Network ($\textbf{DyLAN}$) for LLM-agent collaboration on complicated tasks like reasoning and code generation. DyLAN enables agents to interact for multiple rounds in a dynamic architecture with inference-time agent selection and an early-stopping mechanism to improve performance and efficiency. We further design an automatic agent team optimization algorithm based on an unsupervised metric termed $\textit{Agent Importance Score}$, enabling the selection of best agents based on the contribution each agent makes. Empirically, we demonstrate that DyLAN performs well in both reasoning and code generation tasks with reasonable computational cost. DyLAN achieves 13.0% and 13.3% improvement on MATH and HumanEval, respectively, compared to a single execution on GPT-35-turbo. On specific subjects of MMLU, agent team optimization in DyLAN increases accuracy by up to 25.0%.

As Natural Language Processing (NLP) systems are increasingly employed in intricate social environments, a pressing query emerges: Can these NLP systems mirror human-esque collaborative intelligence, in a multi-agent society consisting of multiple large language models (LLMs)? This paper probes the collaboration mechanisms among contemporary NLP systems by melding practical experiments with theoretical insights. We fabricate four unique `societies' comprised of LLM agents, where each agent is characterized by a specific `trait' (easy-going or overconfident) and engages in collaboration with a distinct `thinking pattern' (debate or reflection). Evaluating these multi-agent societies on three benchmark datasets, we discern that LLM agents navigate tasks by leveraging diverse social behaviors, from active debates to introspective reflections. Notably, certain collaborative strategies only optimize efficiency (using fewer API tokens), but also outshine previous top-tier approaches. Moreover, our results further illustrate that LLM agents manifest human-like social behaviors, such as conformity or majority rule, mirroring foundational Social Psychology theories. In conclusion, we integrate insights from Social Psychology to contextualize the collaboration of LLM agents, inspiring further investigations into the collaboration mechanism for LLMs. We commit to sharing our code and datasets (already submitted in supplementary materials), hoping to catalyze further research in this promising avenue (All code and data are available at \url{https://github.com/zjunlp/MachineSoM}.).

Highly autonomous generative agents powered by large language models promise to simulate intricate social behaviors in virtual societies. However, achieving real-time interactions with humans at a low computational cost remains challenging. Here, we introduce Lyfe Agents. They combine low-cost with real-time responsiveness, all while remaining intelligent and goal-oriented. Key innovations include: (1) an option-action framework, reducing the cost of high-level decisions; (2) asynchronous self-monitoring for better self-consistency; and (3) a Summarize-and-Forget memory mechanism, prioritizing critical memory items at a low cost. We evaluate Lyfe Agents' self-motivation and sociability across several multi-agent scenarios in our custom LyfeGame 3D virtual environment platform. When equipped with our brain-inspired techniques, Lyfe Agents can exhibit human-like self-motivated social reasoning. For example, the agents can solve a crime (a murder mystery) through autonomous collaboration and information exchange. Meanwhile, our techniques enabled Lyfe Agents to operate at a computational cost 10-100 times lower than existing alternatives. Our findings underscore the transformative potential of autonomous generative agents to enrich human social experiences in virtual worlds.

Recent advancements in robotics have paved the way for robots to replace humans in perilous situations, such as searching for victims in blazing buildings, earthquake-damaged structures, uncharted caves, traversing minefields, or patrolling crime-ridden streets. These challenges can be generalized as problems where agents need to explore unknown mazes. Although various algorithms for single-agent maze exploration exist, extending them to multi-agent systems poses complexities. We propose a solution: a cooperative multi-agent system of automated mobile agents for exploring unknown mazes and locating stationary targets. Our algorithm employs a potential field governing maze exploration, integrating cooperative agent behaviors like collision avoidance, coverage coordination, and path planning. This approach builds upon the Heat Equation Driven Area Coverage (HEDAC) method by Ivi\'c, Crnkovi\'c, and Mezi\'c. Unlike previous continuous domain applications, we adapt HEDAC for discrete domains, specifically mazes divided into nodes. Our algorithm is versatile, easily modified for anti-collision requirements, and adaptable to expanding mazes and numerical meshes over time. Comparative evaluations against alternative maze-solving methods illustrate our algorithm's superiority. The results highlight significant enhancements, showcasing its applicability across diverse mazes. Numerical simulations affirm its robustness, adaptability, scalability, and simplicity, enabling centralized parallel computation in autonomous systems of basic agents/robots.

We propose an efficient framework using the Dehornoy order for homotopy-aware multi-agent path planning in the plane. We developed a method to generate homotopically distinct solutions of multi-agent path planning problem in the plane by combining our framework with revised prioritized planning and proved its completeness under specific assumptions. Experimentally, we demonstrated that the runtime of our method grows approximately quintically with the number of agents. We also confirmed the usefulness of homotopy-awareness by showing experimentally that generation of homotopically distinct solutions by our method contributes to planning low-cost trajectories for a swarm of agents.