8 papers
Rethinking Groups in Critic-Free RLVR
Yihong Wu, Liheng Ma, Lingfeng Xiao +4
Reinforcement learning (RL) has become a central paradigm for post-training large language models. Existing critic-free RL methods typically generate a group of rollouts for the sa…
TrasMuon: Trust-Region Adaptive Scaling for Orthogonalized Momentum Optimizers
Peng Cheng, Jiucheng Zang, Qingnan Li +6
Muon-style optimizers leverage Newton-Schulz (NS) iterations to orthogonalize updates, yielding update geometries that often outperform Adam-series methods. However, this orthogona…
The Value of Variance: Mitigating Debate Collapse in Multi-Agent Systems via Uncertainty-Driven Policy Optimization
Luoxi Tang, Yuqiao Meng, Joseph Costa +3
Multi-agent debate (MAD) systems improve LLM reasoning through iterative deliberation, but remain vulnerable to debate collapse, a failure type where final agent decisions are comp…
Mobility-Aware Cache Framework for Scalable LLM-Based Human Mobility Simulation
Hua Yan, Heng Tan, Yingxue Zhang +1
Simulating large-scale human mobility is fundamental to understanding population movement patterns and supporting real-world geospatial applications such as urban planning, epidemi…
It Takes Two: Your GRPO Is Secretly DPO
Yihong Wu, Liheng Ma, Lei Ding +9
GRPO has emerged as a prominent reinforcement learning algorithm for post-training LLMs. Unlike critic-based methods, GRPO computes advantages by estimating the \emph{value baselin…
From Evidence to Trajectory: Abductive Reasoning Path Synthesis for Retrieval-Augmented Generation Agents Development
Muzhi Li, Jinhu Qi, Yihong Wu +9
Retrieval-augmented generation (RAG) agent development is hindered by the lack of executable ground-truth agent-environment interaction trajectories. Existing datasets provide ques…
An Entity Linking Agent for Question Answering
Yajie Luo, Yihong Wu, Muzhi Li +5
Some Question Answering (QA) systems rely on knowledge bases (KBs) to provide accurate answers. Entity Linking (EL) plays a critical role in linking natural language mentions to KB…
Advancing Multi-Agent RAG Systems with Minimalist Reinforcement Learning
Yihong Wu, Liheng Ma, Muzhi Li +7
Large Language Models (LLMs) equipped with modern Retrieval-Augmented Generation (RAG) systems often employ multi-turn interaction pipelines to interface with search engines for co…