25 papers
Judged Useless, Queried Anyway: Tool-Using Agents Rarely Turn Their Own Evidence Judgments into Stopping Decisions
Chubin Zhang, Zhenglin Wan, Xingrui Yu +4
An agent whose tool keeps returning nothing useful should stop relying on it. In a retrieval environment with controlled source failures, we separate how agents judge results from…
CUA-Sandbox: Efficient Environments for Computer-Use Agent Reinforcement Learning
Xin Yan, Zhengbo Jiao, Jiaqi Liu +11
Reinforcement learning enables computer-use agents to improve through interaction with real software environments, including websites and desktop applications. However, conventiona…
SpikeOPD: Stable On-Policy Distillation for Autoregressive Spiking Language Models
Enqiao Lu, Xingrui Yu, Yiwei Fu +7
Spiking neural networks (SNNs) offer a path to energy-efficient language modeling through sparse encoding and event-driven computation, but training capable spiking language models…
SkillDAG: Self-Evolving Typed Skill Graphs for LLM Skill Selection at Scale
Tong Bai, Zhenglin Wan, Pengfei Zhou +3
Large skill libraries are now standard equipment for LLM agents, and which subset reaches the prompt decides what the agent can do. Enumeration and top- embedding retrieval both…
Don't Blindly Trust It: How Unreliable Feedback Breaks Tool-Using LLM Agents
Chubin Zhang, Zhenglin Wan, Xingrui Yu +5
Tool-augmented agents are typically evaluated by their gains under reliable external feedback. Yet these gains leave open a key counterfactual: when feedback is unreliable, would t…
Calibration Is Not Control: Intervention Value for LLM-Agent Oversight
Chubin Zhang, Zhenglin Wan, Xingrui Yu +5
Runtime oversight often intervenes when an LLM agent's calibrated failure score crosses a threshold. Yet states with the same failure risk can differ in whether intervention helps.…
Dirichlet-Guided Group Forecasting for Alleviating Over-smoothing in Time Series Forecasting
Xingyu Zhang, Jingyao Wang, Xin Yu +4
Time series forecasting often suffers from over-smoothing, especially when future dynamics are multi-modal. Forecasts may follow the coarse trend of the observed future, but fail t…
STaR-Quant: State-Time Consistent Post-Training Quantization for Diffusion Large Language Models
Xin Yan, Aqiang Wang, Zhenglin Wan +2
Diffusion large language models (DLLMs) have recently emerged as a promising alternative to autoregressive LLMs by generating text through iterative masked denoising with bidirecti…
Vision-OPD: Learning to See Fine Details for Multimodal LLMs via On-Policy Self-Distillation
Qianhao Yuan, Jie Lou, Xing Yu +4
Multimodal Large Language Models (MLLMs) still struggle with fine-grained visual understanding, where answers often depend on small but decisive evidence in the full image. We obse…
Adversarial Dual On-Policy Distillation from Expressive Teacher
Zhenglin Wan, Jingxuan Wu, Xingrui Yu +5
Learning from demonstrations in embodied control is often cast as behavioral cloning, and recent diffusion or flow-matching policies improve this paradigm by modeling multi-modal e…
Your Teacher Can't Help You Here: Combating Supervision Fidelity Decay in On-Policy Distillation
Yanjiang Liu, Jie Lou, Xinyan Guan +7
On-policy distillation transfers reasoning capabilities by training a student model on its own generated trajectories using token-level feedback from a teacher. However, we identif…
Learning from Failures: Correction-Oriented Policy Optimization with Verifiable Rewards
Mengjie Ren, Jie Lou, Boxi Cao +6
Reinforcement Learning with Verifiable Rewards (RLVR) has emerged as an effective paradigm for improving the reasoning capabilities of large language models. However, RLVR training…
From Generic Correlation to Input-Specific Credit in On-Policy Self Distillation
Guobin Shen, Lei Huang, Xiang Cheng +4
On-policy self-distillation has emerged as a promising paradigm for post-training language models, in which the model conditions on environment feedback to serve as its own teacher…
Anti-Self-Distillation for Reasoning RL via Pointwise Mutual Information
Guobin Shen, Xiang Cheng, Chenxiao Zhao +4
On-policy self-distillation, where a student is pulled toward a copy of itself conditioned on privileged context (e.g., a verified solution or feedback), offers a promising directi…
Towards Backdoor-Based Ownership Verification for Vision-Language-Action Models
Ming Sun, Rui Wang, Xingrui Yu +5
Vision-Language-Action models (VLAs) support generalist robotic control by enabling end-to-end decision policies directly from multi-modal inputs. As trained VLAs are increasingly…
Lifting Traces to Logic: Programmatic Skill Induction with Neuro-Symbolic Learning for Long-Horizon Agentic Tasks
Jie-Jing Shao, Haiyan Yin, Yueming Lyu +5
Foundation model-driven agents often struggle with long-horizon planning due to the transient nature of purely prompting-based reasoning. While existing skill induction methods mit…
HBVLA: Pushing 1-Bit Post-Training Quantization for Vision-Language-Action Models
Xin Yan, Zhenglin Wan, Feiyang Ye +4
Vision-Language-Action (VLA) models enable instruction-following embodied control, but their large compute and memory footprints hinder deployment on resource-constrained robots an…
Advancing Analytic Class-Incremental Learning through Vision-Language Calibration
Binyu Zhao, Wei Zhang, Xingrui Yu +2
Class-incremental learning (CIL) with pre-trained models (PTMs) faces a critical trade-off between efficient adaptation and long-term stability. While analytic learning enables rap…
SA-VLA: Spatially-Aware Flow-Matching for Vision-Language-Action Reinforcement Learning
Xu Pan, Zhenglin Wan, Xingrui Yu +6
Vision-Language-Action (VLA) models exhibit strong generalization in robotic manipulation, yet reinforcement learning (RL) fine-tuning often degrades robustness under spatial distr…
Denoising Time Matters:Diverse Generation in Diffusion Language Models
Jingxuan Wu, Zhenglin Wan, Xingrui Yu +5
Diffusion language models (Diffusion-LMs) generate text through iterative denoising, exposing a temporal structure that is largely absent from autoregressive decoding. In this pape…
Reflection-Driven Control for Trustworthy Code Agents
Bin Wang, Jiazheng Quan, Xingrui Yu +3
Contemporary large language model (LLM) agents are remarkably capable, but they still lack reliable safety controls and can produce unconstrained, unpredictable, and even actively…
Training Diffusion Policies via Prior-Mapping Co-Evolution
Chubin Zhang, Zhenglin Wan, Feng Chen +7
Reinforcement learning (RL) faces a persistent tension: policies that are stable to optimize (e.g., Gaussians) are often too simple to represent the multimodal action distributions…
FM-IRL: Flow-Matching for Reward Modeling and Policy Regularization in Reinforcement Learning
Zhenglin Wan, Jingxuan Wu, Xingrui Yu +4
Flow Matching (FM) has shown remarkable ability in modeling complex distributions and achieves strong performance in offline imitation learning for cloning expert behaviors. Howeve…
Letting Trajectories Spread: Quality-Preserving Control for Diverse Flow Matching
Jingxuan Wu, Zhenglin Wan, Xingrui Yu +4
Flow-based text-to-image models follow deterministic trajectories, making it costly to explore diverse modes under limited sampling budgets. Existing approaches to improving divers…