Author

Xinya Du

h-index 8 View on Semantic Scholar

We have 2 of 19 papers

We haven’t gathered this author’s papers yet. Follow them and we’ll fetch their work.

Not the right person? Other researchers publish under this name.

Jul 2026

Be Consistent! Enhancing Robust Visual Reasoning in LVLMs with Consistency Constraints

ConVLM is presented, which improves LVLM reasoning through Group Relative Policy Optimization-based reinforcement learning with a novel consistency reward, and ConVBench, a complex vision-centric reasoning benchmark in which each image is paired with two logically equivalent questions across six categories.

Li-Qiang Jing, Xiong Zhou, Siddharth Varia et al. · arXiv.org · 1 citation · ⚡1

Conference Open access 2026

Your Reasoning Model is Secretly a Reward Model - Optimization-Free Verification from Experience

This paper introduces C LUE (Clustering and Experience-based Verification) , a training-free, non-parametric verifier that improves selection and reranking in Large Language Model outputs and finds that correct and incorrect solutions exhibit measurable geometric differences in their hidden-state trajectories.

Zhenwen Liang, Ruosen Li, Yujun Zhou et al. · Annual Meeting of the Associ... · 0 citations