Weiqing Luo

I am a PhD student at Arizona State University. I received my M.Sc. and B.Sc. in Computer Science from Nanjing University.

My research focuses on large language models and multimodal AI, with particular interests in test-time learning, adaptive computation and model interpretability. I have interned at Tencent (WeChat), SenseTime and ByteDance (TikTok).

Seeking research internships for 2027

Portrait of Weiqing Luo

Research

LLMs and multimodal AI

Multimodal adaptation

I adapt multimodal models to the task at hand: selecting visual evidence by its estimated downstream utility beyond semantic relevance, using a training-free, lightweight surrogate, and fine-tuning pretrained vision-language models to the image resolution each task needs. (ACL 2026, EMNLP 2025)

Test-time and
post-training learning

Models can keep improving after training by learning at test time from self-generated supervision, but that supervision can reinforce its own bias. My structured exploration framework combines multi-view reasoning with instance-adaptive sampling to mitigate this bias, improving reasoning performance while reducing sampling cost by over 30%. (under review)

Model interpretability

I also study how evidence and context shape a model's answer: tracing evidence across semantic input regions in vision-language models, and analyzing bounded responses to repeated assertions and the sparse pathways that mediate contextual override. (ICML 2026, EMNLP 2026, under review)

Earlier work brought external knowledge into recommendation and dataset retrieval, including TRAWL, deployed in WeChat Subscriptions (CIKM 2025, Findings of EMNLP 2023).

Publications

Selected papers; filled labels mark first-author work. The full list is on Google Scholar.

  1. Under review

    Beyond Self-Consensus: Mitigating Self-Reinforcing Exploration Bias in Test-Time Reinforcement Learning

    Weiqing Luo, Zongye Hu, Haofeng Zhang, Ziyi Huang

  2. Under review

    Semantic Route Attribution: Tracing Value Flow across Input Regions in Vision-Language Models

    Weiqing Luo, Zongye Hu, Haofeng Zhang, Ziyi Huang

  3. ACL 2026

    Utility-Oriented Visual Evidence Selection for Multimodal Retrieval-Augmented Generation

    Weiqing Luo, Zongye Hu, Xiao Wang, Zhiyuan Yu, Haofeng Zhang, Ziyi Huang

    paper code

  4. EMNLP 2026

    Tracing Contextual Override: Mediated Causal Attribution in Language Models

    Zongye Hu, Weiqing Luo, Haofeng Zhang, Ziyi Huang

    paper

  5. ICML 2026

    Not All Answers Are Contextually Persuadable: Inference Dynamics in Large Language Models under Contextual Influence

    Zongye Hu, Weiqing Luo, Yanjie Fu, Yu Gan, Haofeng Zhang, Ziyi Huang

    paper openreview

  6. EMNLP 2025

    Task-Aware Resolution Optimization for Visual Large Language Models

    Weiqing Luo, Zhen Tan, Yifan Li, Xinyu Zhao, Kwonjoon Lee, Behzad Dariush, Tianlong Chen

    paper

  7. CIKM 2025

    TRAWL: External Knowledge-Enhanced Recommendation with LLM Assistance

    Weiqing Luo, Chonggang Song, Lingling Yi, Gong Cheng

    Deployed in WeChat Subscriptions paper

  8. SIGIR 2025

    Benchmarking Recommendation, Classification, and Tracing Based on Hugging Face Knowledge Graph

    Qiaosheng Chen, Kaijia Huang, Xiao Zhou, Weiqing Luo, Yuanning Cui, Gong Cheng

    paper code

  9. SIGIR 2024

    ACORDAR 2.0: A Test Collection for Ad Hoc Dataset Retrieval with Densely Pooled Datasets and Question-Style Queries

    Qiaosheng Chen, Weiqing Luo, Zixian Huang, Tengteng Lin, Xiaxia Wang, Ahmet Soylu, Basil Ell, Baifan Zhou, Evgeny Kharlamov, Gong Cheng

    paper code

  10. Findings of EMNLP 2023

    An Empirical Investigation of Implicit and Explicit Knowledge-Enhanced Methods for Ad Hoc Dataset Retrieval

    Weiqing Luo, Qiaosheng Chen, Zhiyang Zhang, Zixian Huang, Gong Cheng

    paper code

  11. ISWC 2023

    Dense Re-Ranking with Weak Supervision for RDF Dataset Search

    Qiaosheng Chen, Zixian Huang, Zhiyang Zhang, Weiqing Luo, Tengteng Lin, Qing Shi, Gong Cheng

    Best Research Paper Nominee paper code

Earlier publications (2)
  1. Data Intelligence 2022

    CKGSE: A Prototype Search Engine for Chinese Knowledge Graphs

    Xiaxia Wang, Tengteng Lin, Weiqing Luo, Gong Cheng, Yuzhong Qu

    paper

  2. CCKS 2021

    Content-Based Open Knowledge Graph Search: A Preliminary Study with OpenKG.CN

    Xiaxia Wang, Tengteng Lin, Weiqing Luo, Gong Cheng, Yuzhong Qu

    paper

Education and experience

  1. Aug 2025 – Present

    Ph.D., Arizona State University

    Tempe, AZ

    • Graduate Research Assistant, working on multimodal adaptation, test-time and post-training learning, and model interpretability. See Research.
  2. Sep 2022 – Jun 2025

    M.Sc. in Computer Science, Nanjing University

    Nanjing, China · GPA 90.83/100

    • Graduate Researcher, Websoft Lab. Evaluated dense retrieval and knowledge-based methods for ad hoc dataset search, and contributed to work on structured data retrieval and evaluation benchmarks (Findings of EMNLP 2023, first author; SIGIR 2024; ISWC 2023).
    1. Jun 2024 – Oct 2024

      University of North Carolina at Chapel Hill

      Research assistant

      UNITES Lab · Remote

      • Task-aware visual resolution. Investigated resolution preferences across eight vision-language tasks, developed a selection criterion based on image complexity and model uncertainty variance, and built a parameter-efficient fine-tuning approach for adapting pretrained models to higher resolutions without exhaustive retraining (EMNLP 2025, first author).
    2. Jul 2023 – May 2024

      Tencent

      Research intern

      WeChat Machine Learning Application Group · Shenzhen, China

      • LLM-enhanced recommendation. Developed TRAWL, which uses LLMs to extract recommendation-relevant knowledge from external text and contrastive learning to align semantic and behavioral representations (CIKM 2025, first author).
      • Deployed TRAWL in WeChat Subscriptions; online A/B tests showed 0.31% more clicks and 0.36% more read sessions over the baseline.
  3. Sep 2018 – Jun 2022

    B.Sc. in Computer Science, Nanjing University

    Elite Program · Nanjing, China · GPA 4.45/5.0

    1. Jan 2022 – Jul 2022

      SenseTime

      Research intern

      Decision Intelligence Group · Shenzhen, China

      • Contributed to DI-engine, an open-source reinforcement learning framework, and to algorithm research on hybrid action spaces.
    2. Jul 2021 – Sep 2021

      ByteDance

      Engineering intern

      TikTok Recommendation Group · Shanghai, China

      • Built and maintained an internal debugging dashboard and supported backend engineering for TikTok's live-streaming services.

Honors

  • Scholarly Advancement Award, Arizona State University2026
  • Outstanding Graduate Student, Nanjing University2023–2024
  • First Prize, Graduate Talent Scholarship, Nanjing University2024
  • Outstanding Undergraduate Thesis, Nanjing University2022
  • Elite Program Scholarship, Nanjing University2019–2021

Service

  • AAAI 2027Program Committee
  • ICLR 2027Reviewer
  • NeurIPS 2026, Evaluations and Datasets TrackReviewer
  • ACL Rolling Review, 2025–2026Reviewer
  • DASFAA 2024Reviewer