NashEval is proposed, a general framework for robust contextual equilibrium learning that frame evaluation as a contextual game between two players, each selecting a distribution over agents as the strategy to receive greater collective preference than the other.
Abstract
Many applications require to evaluate agents under contextual information (e.g., a prompt, task, or user group). We study how to perform such context-dependent agent evaluation from offline feedback. Existing score-based models for this purpose (e.g., Bradley-Terry) impose a transitive preference ordering, which fails to reflect collective preferences when human judgements are heterogeneous. Inspired by social choice theory, we frame evaluation as a contextual game between two players, each selecting a distribution over agents as the strategy to receive greater collective preference than the other. Then, the support of the Nash equilibrium defines a context-specific set of winners. However, learning context-specific equilibria from offline logs is difficult because each context reveals human feedback on only a subset of agents, and, hence, a naive plug-in estimator can therefore be biased. To address these challenges, we propose NashEval, a general framework for robust contextual equilibrium learning. NashEval first constructs debiased estimates of the contextual payoff matrix that characterizes the game. NashEval then learns the context-to-equilibrium mapping with a tailored orthogonal loss, which avoids the need to solve a separate game for each context. We show theoretically that errors in estimating the nuisance functions underlying the payoff matrix affect the risk of the learned equilibrium (i.e., exploitability) only through higher-order terms. Across various experiments, NashEval improves robustness of equilibrium learning and consistently identifies the set of top-performing agents across contexts.
Understanding the objectives that drive agents'behavior is crucial in strategic environments, where multiple agents interact and adapt to one another. This challenge arises in domains such as cybersecurity, autonomous driving, and online markets, where agents'underlying incentives are typically unknown and must be infe...
Antoine Bergerault, Cyrille Kone, G. Ramponi· 0 citations
In many modern multi-agent systems, a mediator seeks to influence strategic decision dynamics using only outcome-level feedback and without knowing agents'utilities. In this work, we study how a mediator can learn an unknown game and steer adaptive agents toward outcomes that maximize her utility using only trajectory...
It is found that correctly inferring user preferences is necessary but not sufficient for task success, as execution failures in downstream web interactions remain a significant bottleneck even when agents align with the target preference.
Dong-Chan Shin, Xing Han Lù, Jiaqi Deng et al.· arXiv.org· 0 citations
When is social learning efficient if agents have heterogeneous preferences over a set of actions? When myopic agents generate information only about the action they choose, efficiency requires balance such that all high-value actions are discovered. We characterize the value of policies that control the flow of past in...
This work introduces Steiner's taxonomy of group tasks as a framework for analyzing multi-agent LLM scaling and shows that task structure is a fundamental determinant of team scaling.