This work proposes RMEval, a benchmark that revisits reward model evaluation through the lens of global ranking consistency, and evaluates existing reward models on RMEval and demonstrates that the metric exhibits a strong correlation with downstream policy performance, with Pearson coefficients exceeding 0.6 across mu...
Wenjie Qiu, Xuqin Zhang, Yihang Zhang et al.· Proceedings of the 32nd ACM...· 0 citations
Reliable policy optimization in real-time bidding (RTB) demands a high-fidelity world model capable of simulating stochastic market dynamics under constraints. However, current generative simulators are limited by their reliance on deterministic point prediction objectives (e.g., mean squared error). These approaches a...
Chenyang Wu, Tianyu Wang, Shengjun Fang et al.· Proceedings of the 32nd ACM...· 0 citations
This work proposes MAFIS, a novel method that addresses limitations for both online and offline MAIL settings by building upon the single-agent IQ-Learn framework and introducing the value decomposition network to factorize the imitation objective at agent level, thus enabling scalable training for multi-agent systems.
Yichen Li, Zhongxiang Ling, Tao Jiang et al.· Neural Information Processin...· 3 citations
Reward Models (RMs) are widely used to guide the alignment of Large Language Models (LLMs) in Reinforcement Learning from Human Feedback (RLHF), and their quality has a substantial impact on the performance of trained policy models. In current practice, reward models are predominantly evaluated using local metrics such...
Wenjie Qiu, Xuqin Zhang, Yihang Zhang et al.· Proceedings of the 32nd ACM...· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
