ICML 2025 Papers

Leveraging more test-time computation has proven to be an effective way to boost the reasoning capabilities of large language models (LLMs). Among various methods, the verify-and-improve paradigm stands out for enabling dynamic solution exploration and feedback incorporation. However, existing approaches often suffer from restricted feedback spaces and lack of coordinated training of different parties, leading to suboptimal performance. To address this, we model this multi-turn refinement process as a Markov Decision Process and introduce DPSDP (**D**irect **P**olicy **S**earch by **D**ynamic **P**rogramming), a reinforcement learning algorithm that trains an actor-critic LLM system to iteratively refine answers via direct preference learning on self-generated data. Theoretically, DPSDP can match the performance of any policy within the training distribution. Empirically, we instantiate DPSDP with various base models and show improvements on both in- and out-of-distribution benchmarks. For example, on benchmark MATH 500, majority voting over five refinement steps increases first-turn accuracy from 58.2% to 63.2% with Ministral-based models. An ablation study further confirms the benefits of multi-agent collaboration and out-of-distribution generalization.

Recent advances toward foundation models for routing problems have shown great potential of a unified deep model for various VRP variants. However, they overlook the complex real-world customer distributions. In this work, we advance the Multi-Task VRP (MTVRP) setting to the more realistic yet challenging Multi-Task Multi-Distribution VRP (MTMDVRP) setting, and introduce SHIELD, a novel model that leverages both *sparsity* and *hierarchy* principles. Building on a deeper decoder architecture, we first incorporate the Mixture-of-Depths (MoD) technique to enforce sparsity. This improves both efficiency and generalization by allowing the model to dynamically select nodes to use or skip each decoder layer, providing the needed capacity to adaptively allocate computation for learning the task/distribution specific and shared representations. We also develop a context-based clustering layer that exploits the presence of hierarchical structures in the problems to produce better local representations. These two designs inductively bias the network to identify key features that are common across tasks and distributions, leading to significantly improved generalization on unseen ones. Our empirical results demonstrate the superiority of our approach over existing methods on 9 real-world maps with 16 VRP variants each.

Vision Language Model (VLM) Agents are stateful, autonomous entities capable of perceiving and interacting with their environments through vision and language. Multi-agent systems comprise specialized agents who collaborate to solve a (complex) task. A core security property is **robustness**, stating that the system maintains its integrity during adversarial attacks. Multi-agent systems lack robustness, as a successful exploit against one agent can spread and **infect** other agents to undermine the entire system's integrity. We propose a defense Cowpox to provably enhance the robustness of a multi-agent system by a distributed mechanism that improves the **recovery rate** of agents by limiting the expected number of infections to other agents. The core idea is to generate and distribute a special *cure sample* that immunizes an agent against the attack before exposure. We demonstrate the effectiveness of Cowpox empirically and provide theoretical robustness guarantees.

As multimodal large language models (MLLMs) advance, MLLM-based virtual agents have demonstrated remarkable performance. However, existing benchmarks face significant limitations, including uncontrollable task complexity, extensive manual annotation, and a lack of multidimensional evaluation. In response to these challenges, we introduce OmniBench, a self-generating, graph-based benchmark with an automated pipeline for synthesizing tasks of controllable complexity through subtask composition. To evaluate the diverse capabilities of virtual agents on the graph, we further present OmniEval, a multidimensional evaluation framework that includes subtask-level evaluation, graph-based metrics, and comprehensive tests across 10 capabilities. Our synthesized dataset contains 36k graph-structured tasks across 20 scenarios, achieving a 91% human acceptance rate. Training on our graph-structured data shows that it improves generalization across environments. We conduct multidimensional evaluations for virtual agents, revealing their performance across various capabilities and paving the way for future advancements. Our project is available at https://omni-bench.github.io.

We initiate the study of a repeated principal-agent problem over a finite horizon $T$, where a principal sequentially interacts with $K\geq 2$ types of agents arriving in an *adversarial* order. At each round, the principal strategically chooses one of the $N$ arms to incentivize for an arriving agent of *unknown type*. The agent then chooses an arm based on its own utility and the provided incentive, and the principal receives a corresponding reward. The objective is to minimize regret against the best incentive in hindsight. Without prior knowledge of agent behavior, we show that the problem becomes intractable, leading to linear regret. We analyze two key settings where sublinear regret is achievable. In the first setting, the principal knows the arm each agent type would select greedily for any given incentive. Under this setting, we propose an algorithm that achieves a regret bound of $\mathcal{O}(\min(\sqrt{KT\log N},K\sqrt{T}))$ and provide a matching lower bound up to a $\log K$ factor. In the second setting, an agent's response varies smoothly with the incentive and is governed by a Lipschitz constant $L$. Under this setting, we show that there is an algorithm with a regret bound of $\tilde{\mathcal{O}}((LN)^{1/3}T^{2/3})$ and establish a matching lower bound up to logarithmic factors. Finally, we extend our algorithmic results for both settings by allowing the principal to incentivize multiple arms simultaneously in each round.

Practical reinforcement learning (RL) usually requires agents to be optimized for multiple potentially conflicting criteria, e.g. speed vs. safety. Although Multi-Objective RL (MORL) algorithms have been studied in previous works, their trained agents often cover limited Pareto optimal solutions and they lack precise controllability of the delicate trade-off among multiple objectives. Hence, the resulting agent is not versatile in aligning with customized requests from different users. To bridge the gap, we develop the ``Preference controllable (PC) RL'' framework, which trains a preference-conditioned meta-policy that takes user preference as input controlling the generated trajectories within the preference region on the Pareto frontier. The PCRL framework is compatible with advanced Multi-Objective Optimization~(MOO) algorithms that are rarely seen in previous MORL approaches. We also proposed a novel preference-regularized MOO algorithm specifically for PCRL. We provide a comprehensive theoretical analysis to justify its convergence and preference controllability. We evaluate PCRL with different MOO algorithms against state-of-the-art MORL baselines in various challenging environments with up to six objectives. In these experiments, our proposed method exhibits significantly better controllability than existing approaches and can generate Pareto solutions with better diversity and utilities.

Experimentation with interference poses a significant challenge in contemporary online platforms. Prior research on experimentation with interference has concentrated on the final output of a policy. Cumulative performance, while equally important, is less well understood. To address this gap, we introduce the problem of Multi-armed Bandits with Interference (MABI), where the learner assigns an arm to each of $N$ experimental units over $T$ rounds. The reward of each unit in each round depends on the treatments of all units, where the interference between two units decays in their distance. The reward functions are chosen by an adversary and may vary arbitrarily over time and across different units. We first show that the optimal expected regret (against the best fixed-arm policy) is $\tilde O(\sqrt T)$, and can be achieved by a switchback policy. However, the regret (as a random variable) for any switchback policy suffers a high variance, since it does not account for $N$. We propose a policy based on a novel clustered randomization scheme, whose regret (i) is optimal in expectation and (ii) admits a high probability bound that vanishes in $N$.

This work investigates the problem of best arm identification for multi-agent multi-armed bandits. We consider $N$ agents grouped into $M$ clusters, where each cluster solves a stochastic bandit problem. The mapping between agents and bandits is \textit{a priori} unknown. Each bandit is associated with $K$ arms, and the goal is to identify the best arm for each agent under a $\delta$-probably correct ($\delta$-PC) framework, while minimizing sample complexity and communication overhead. We propose two novel algorithms: \emph{Clustering then Best Arm Identification} (\texttt{Cl-BAI}) and \emph{Best Arm Identification then Clustering} (\texttt{BAI-Cl}). \texttt{Cl-BAI} employs a two-phase approach that first clusters agents based on the bandit problems they are learning, followed by identifying the best arm for each cluster. \texttt{BAI-Cl} reverses the sequence by identifying the best arms first and then clustering agents accordingly. Both algorithms exploit the successive elimination framework to ensure computational efficiency and high accuracy. Theoretical analysis establishes $\delta$-PC guarantees for both methods, derives bounds on their sample complexity, and provides a lower bound for the problem class. Moreover, when $M$ is small (a constant), we show that the sample complexity of (a variant of) \texttt{BAI-Cl} is (order-wise) minimax optimal. Experiments on synthetic and real-world (Movie Lens, Yelp) data demonstrates the superior performance of the proposed algorithms in terms of sample and communication efficiency, particularly in settings where $M \ll N$.

Despite the superior capability in complementary information exploration and consistent clustering structure learning, most current weight-based multi-modal clustering methods still contain three limitations: 1) lack of trustworthiness in learned weights; 2) isolated view weight learning; 3) extra weight parameters. Motivated by the peer-review mechanism in the academia, we in this paper give a new peer-review look on the multi-modal clustering problem and propose to iteratively treat one modality as "author" and the remaining modalities as "reviewers" so as to reach a peer-review score for each modality. It essentially explores the underlying relationships among modalities. To improve the trustworthiness, we further design a new trustworthy score with a self-supervision working mechanism. Following that, we propose a novel Peer-review Trustworthy Information Bottleneck (PTIB) method for weighted multi-modal clustering, where both the above scores are simultaneously taken into account for accurate and parameter-free modality weight learning. Extensive experiments on eight multi-modal datasets suggest that PTIB can outperform the state-of-the-art multi-modal clustering methods.

LLM-based multi-agent systems (MAS) have shown significant potential in tackling diverse tasks. However, to design effective MAS, existing approaches heavily rely on manual configurations or multiple calls of advanced LLMs, resulting in inadaptability and high inference costs. In this paper, we simplify the process of building an MAS by reframing it as a generative language task, where the input is a user query and the output is a corresponding MAS. To address this novel task, we unify the representation of MAS as executable code and propose a consistency-oriented data construction pipeline to create a high-quality dataset comprising coherent and consistent query-MAS pairs. Using this dataset, we train MAS-GPT, an open-source medium-sized LLM that is capable of generating query-adaptive MAS within a single LLM inference. The generated MAS can be seamlessly applied to process user queries and deliver high-quality responses. Extensive experiments on 9 benchmarks and 5 LLMs show that the proposed MAS-GPT consistently outperforms 10+ baseline MAS methods on diverse settings, indicating MAS-GPT's high effectiveness, efficiency and strong generalization ability. The codes are released at \url{https://github.com/rui-ye/MAS-GPT}.

Multi-Agent Reinforcement Learning (MARL) struggles with coordination in sparse reward environments. Macro-actions —sequences of actions executed as single decisions— facilitate long-term planning but introduce asynchrony, complicating Centralized Training with Decentralized Execution (CTDE). Existing CTDE methods use padding to handle asynchrony, risking misaligned asynchronous experiences and spurious correlations. We propose the Agent-Centric Actor-Critic (ACAC) algorithm to manage asynchrony without padding. ACAC uses agent-centric encoders for independent trajectory processing, with an attention-based aggregation module integrating these histories into a centralized critic for improved temporal abstractions. The proposed structure is trained via a PPO-based algorithm with a modified Generalized Advantage Estimation for asynchronous environments. Experiments show ACAC accelerates convergence and enhances performance over baselines in complex MARL tasks.

Designing effective reward functions in multi-agent reinforcement learning (MARL) is a significant challenge, often leading to suboptimal or misaligned behaviors in complex, coordinated environments. We introduce Multi-agent Reinforcement Learning from Multi-phase Human Feedback of Mixed Quality ($\text{M}^3\text{HF}$), a novel framework that integrates multi-phase human feedback of mixed quality into the MARL training process. By involving humans with diverse expertise levels to provide iterative guidance, $\text{M}^3\text{HF}$ leverages both expert and non-expert feedback to continuously refine agents' policies. During training, we strategically pause agent learning for human evaluation, parse feedback using large language models to assign it appropriately and update reward functions through predefined templates and adaptive weights by using weight decay and performance-based adjustments. Our approach enables the integration of nuanced human insights across various levels of quality, enhancing the interpretability and robustness of multi-agent cooperation. Empirical results in challenging environments demonstrate that $\text{M}^3\text{HF}$ significantly outperforms state-of-the-art methods, effectively addressing the complexities of reward design in MARL and enabling broader human participation in the training process.

Large Language Models (LLMs) solve complex problems using training-free methods like prompt engineering and in-context learning, yet ensuring reasoning correctness remains challenging. While self-correction methods such as self-consistency and self-refinement aim to improve reliability, they often reinforce biases due to the lack of effective feedback mechanisms. Multi-Agent Debate (MAD) has emerged as an alternative, but we identify two key limitations: bias reinforcement, where debate amplifies model biases instead of correcting them, and lack of perspective diversity, as all agents share the same model and reasoning patterns, limiting true debate effectiveness. To systematically evaluate these issues, we introduce $\textit{MetaNIM Arena}$, a benchmark designed to assess LLMs in adversarial strategic decision-making, where dynamic interactions influence optimal decisions. To overcome MAD’s limitations, we propose $\texttt{\textbf{DReaMAD}}$ ($\textbf{D}$iverse $\textbf{Rea}$soning via $\textbf{M}$ulti-$\textbf{A}$gent $\textbf{D}$ebate with Refined Prompt), a novel framework that (1) refines LLMs’ strategic prior knowledge to improve reasoning quality and (2) promotes diverse viewpoints within a single model by systematically modifying prompts, reducing bias. Empirical results show that $\texttt{\textbf{DReaMAD}}$ significantly improves decision accuracy, reasoning diversity, and bias mitigation across multiple strategic tasks, establishing it as a more effective approach for LLM-based decision-making.

Traditional robust methods in multi-agent reinforcement learning (MARL) often struggle against coordinated adversarial attacks in cooperative scenarios. To address this limitation, we propose the Wolfpack Adversarial Attack framework, inspired by wolf hunting strategies, which targets an initial agent and its assisting agents to disrupt cooperation. Additionally, we introduce the Wolfpack-Adversarial Learning for MARL (WALL) framework, which trains robust MARL policies to defend against the proposed Wolfpack attack by fostering system-wide collaboration. Experimental results underscore the devastating impact of the Wolfpack attack and the significant robustness improvements achieved by WALL. Our code is available at https://github.com/sunwoolee0504/WALL.