ZLZheyuan Liu

Zheyuan Liu

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Zheyuan Liu has a total of 97 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Top Cited Papers
    Open Access

    Towards Trustworthy Retrieval Augmented Generation for Large Language Models: A Survey

    by Bo Ni, Zheyuan Liu, Yongjia Lei, Leyao Wang, Yuying Zhao, Xueqi Cheng, Qingkai Zeng, Xin Dong, Yinglong Xia, Krishnaram Kenthapadi, et al.

    Retrieval-Augmented Generation (RAG) enhances AI-generated content by integrating external knowledge, improving relevance, and reducing hallucinations. However, RAG also introduces risks related to reliability, safety, privacy, fairness, explainability, and accountability, which impact trustworthiness. While various methods aim to address these concerns, a unified framework is lacking. This survey bridges that gap by presenting a comprehensive roadmap for trustworthy RAG systems. We provide a structured analysis of key challenges, existing solutions, and future directions across these aspects. We further organize the field around how the retrieval and generation stages and the six trustworthiness dimensions interact, demonstrating their synergies and trade-offs through a controlled cross-dimensional analysis. Additionally, we highlight downstream applications where trustworthy RAG can make a significant impact, encouraging further research and adoption in real-world AI systems.


  • Open Access

    Inference-Time Control for Trustworthy Large Language Models

    by Yuyang Bai, Zheyuan Liu, Han Yan, Zhangchen Xu, Yixin Wan, Canyu Chen, Zehong Wang, Xiangchi Yuan, Yue Huang, Guangyao Dou, et al.

    Once a large language model is released, training-time alignment is hard to revise; yet deployment introduces context-specific risks that the original training cannot anticipate: evolving safety policies, jurisdictional constraints, retrieval contamination, and adaptive adversarial prompting. In this paper, we unify inference-time techniques for trustworthy generation across safety, privacy, fairness, and factuality under a single framework: the inference-time control plane, with three tiers of intervention -- External Controls (Context Engineering, Guardrails, Decoding Strategies), which act around the model; Internal Manipulations (Representation Engineering, Unlearning, Pruning), which act inside it; and System-Level Orchestration (Multi-Agent Systems), which coordinate several models. We also introduce a meta-axis evaluation framework that crosses the four trustworthiness dimensions with five evaluation axes (effectiveness, locality, generality, interpretability, efficiency), and describe representative metrics at each intersection. We identify four cross-cutting open problems: brittleness under adaptive adversaries, the control-utility tradeoff, verification of removal, and the composition of layered interventions. A curated paper list is available at https://github.com/leopoldwhite/Awesome-Inference-Time-Trustworthiness.

  • Open Access

    OpenDecoder: Open Large Language Model Decoding to Incorporate Document Quality in RAG

    by Fengran Mo, Zhan Su, Yuchen Hui, Jinghan Zhang, Jia Ao Sun, Zheyuan Liu, Chao Zhang, Tetsuya Sakai, Jian-Yun Nie

    The development of large language models (LLMs) has achieved superior performance in a range of downstream tasks, including LLM-based retrieval-augmented generation (RAG). The quality of generated content heavily relies on the usefulness of the retrieved information and the capacity of LLMs' internal information processing mechanism to incorporate it in answer generation. It is generally assumed that the retrieved information is relevant to the question. However, the retrieved information may have a variable degree of relevance and usefulness, depending on the question and the document collection. It is important to take into account the relevance of the retrieved information in answer generation. In this paper, we propose OpenDecoder, a new approach that leverages explicit evaluation of the retrieved information as quality indicator features for generation. We aim to build a RAG model that is more robust to varying levels of noisy context. Three types of explicit evaluation information are considered: relevance score, ranking score, and QPP (query performance prediction) score. The experimental results on five benchmark datasets demonstrate the effectiveness and better robustness of OpenDecoder by outperforming various baseline methods. Importantly, this paradigm is flexible to be integrated with the post-training of LLMs for any purposes and incorporated with any type of external indicators.


  • Open Access

    MTMCS-Bench: Evaluating Contextual Safety of Multimodal Large Language Models in Multi-Turn Dialogues

    by Zheyuan Liu, Dongwhi Kim, Yixin Wan, Xiangchi Yuan, Zhaoxuan Tan, Fengran Mo, Meng Jiang

    Multimodal large language models (MLLMs) are increasingly deployed as assistants that interact through text and images, making it crucial to evaluate contextual safety when risk depends on both the visual scene and the evolving dialogue. Existing contextual safety benchmarks are mostly single-turn and often miss how malicious intent can emerge gradually or how the same scene can support both benign and exploitative goals. We introduce the Multi-Turn Multimodal Contextual Safety Benchmark (MTMCS-Bench), a benchmark of realistic images and multi-turn conversations that evaluates contextual safety in MLLMs under two complementary settings, escalation-based risk and context-switch risk. MTMCS-Bench offers paired safe and unsafe dialogues with structured evaluation. It contains over 30 thousand multimodal (image+text) and unimodal (text-only) samples, with metrics that separately measure contextual intent recognition, safety-awareness on unsafe cases, and helpfulness on benign ones. Across eight open-source and seven proprietary MLLMs, we observe persistent trade-offs between contextual safety and utility, with models tending to either miss gradual risks or over-refuse benign dialogues. Finally, we evaluate five current guardrails and find that they mitigate some failures but do not fully resolve multi-turn contextual risks.

  • Open Access

    Steering Multimodal Large Language Models Decoding for Context-Aware Safety

    by Zheyuan Liu, Zhangchen Xu, Guangyao Dou, Xiangchi Yuan, Zhaoxuan Tan, Radha Poovendran, Meng Jiang

    Multimodal Large Language Models (MLLMs) are increasingly deployed in real-world applications, yet their ability to make context-aware safety decisions remains limited. Existing methods often fail to balance oversensitivity (unjustified refusals of benign queries) and undersensitivity (missed detection of visually grounded risks), leaving a persistent gap in safety alignment. To address this issue, we introduce Safety-aware Contrastive Decoding (SafeCoDe), a lightweight and model-agnostic decoding framework that dynamically adjusts token generation based on multimodal context. SafeCoDe operates in two stages: (1) a contrastive decoding mechanism that highlights tokens sensitive to visual context by contrasting real and Gaussian-noised images, and (2) a global-aware token modulation strategy that integrates scene-level reasoning with token-level adjustment to adapt refusals according to the predicted safety verdict. Extensive experiments across diverse MLLM architectures and safety benchmarks, covering undersensitivity, oversensitivity, and general safety evaluations, show that SafeCoDe consistently improves context-sensitive refusal behaviors while preserving model helpfulness.

  • Incorporating Rather Than Eliminating: Achieving Fairness for Skin Disease Diagnosis Through Group-Specific Experts

    by Gelei Xu, Yuying Duan, Zheyuan Liu, Xueyang Li, Meng Jiang, Michael Lemmon, Wei Jin, Yiyu Shi


  • Open Access

    Knowing More, Acting Better: Hierarchical Representation for Embodied Decision-Making

    by Chunhui Zhang, Zhongyu Ouyang, Xingjian Diao, Zheyuan Liu, Soroush Vosoughi

    Chunhui Zhang, Zhongyu Ouyang, Xingjian Diao, Zheyuan Liu, Soroush Vosoughi. Findings of the Association for Computational Linguistics: EMNLP 2025. 2025.

  • Open Access

    Superficial Self-Improved Reasoners Benefit from Model Merging

    by Xiangchi Yuan, Chunhui Zhang, Zheyuan Liu, Dachuan Shi, Leyan Pan, Soroush Vosoughi, Wenke Lee

    Xiangchi Yuan, Chunhui Zhang, Zheyuan Liu, Dachuan Shi, Leyan Pan, Soroush Vosoughi, Wenke Lee. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing. 2025.


  • Top Cited Papers
    Open Access

    Modality-Aware Neuron Pruning for Unlearning in Multimodal Large Language Models

    by Zheyuan Liu, Guangyao Dou, Xiangchi Yuan, Chunhui Zhang, Zhaoxuan Tan, Meng Jiang

    Zheyuan Liu, Guangyao Dou, Xiangchi Yuan, Chunhui Zhang, Zhaoxuan Tan, Meng Jiang. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2025.


  • Open Access

    Protecting Privacy in Multimodal Large Language Models with MLLMU-Bench

    by Zheyuan Liu, Guangyao Dou, Mengzhao Jia, Zhaoxuan Tan, Qingkai Zeng, Yongle Yuan, Meng Jiang

    Zheyuan Liu, Guangyao Dou, Mengzhao Jia, Zhaoxuan Tan, Qingkai Zeng, Yongle Yuan, Meng Jiang. Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 2025.