Papers by Wenxuan Huang

ParroT: Translating during Chat using Large Language Models tuned with Human Translation and Feedback (2023.findings-emnlp)

Challenge:	Existing benchmarks conflate factual correctness and normative fairness . a model may generate responses that are factually accurate but socially unfair .
Approach:	They propose a benchmark to examine the boundary between fact and fair . they draw on representativeness bias, attribution bias and ingroup–outgroup bias to explain why models often misalign fact and faireness.
Outcome:	The proposed model is based on ten frontier models and is available on github . it is compared with a standard model that generates people of color in Nazi-era uniforms .

Copied to clipboard

Wenxiang Jiao, Jen-tse Huang, Wenxuan Wang, Zhiwei He, Tian Liang, Xing Wang, Shuming Shi, Zhaopeng Tu

Refuse Whenever You Feel Unsafe: Improving Safety in LLMs via Decoupled Refusal Training (2025.acl-long)

Challenge:	Large language models (LLMs) like ChatGPT are only accessible through restricted APIs, which creates barriers to new research and advancements in the field.
Approach:	They propose a framework to enhance and regulate the translation abilities during chat . they reformulate translation data into the instruction-following style and introduce a "Hint" field .
Outcome:	The proposed framework enhances and regulates the translation abilities during chat . it reformulates translation data into the instruction-following style and introduces a "Hint" field .

Copied to clipboard

Youliang Yuan, Wenxiang Jiao, Wenxuan Wang, Jen-tse Huang, Jiahao Xu, Tian Liang, Pinjia He, Zhaopeng Tu

Chain-of-Jailbreak Attack for Image Generation Models via Step by Step Editing (2025.findings-acl)

Challenge:	Large Language Models exhibit a level of intelligence that is both impressive and everevolving, but their ability to refuse generating unsafe content is a double-edged sword.
Approach:	They propose a method to tackle a refusal position bias within safety tuning data that compromises the models’ ability to appropriately refuse generating unsafe content.
Outcome:	The proposed method significantly improves model safety without compromising performance and surpasses baseline methods in defending against attacks.

Copied to clipboard

Wenxuan Wang, Kuiyi Gao, Youliang Yuan, Jen-tse Huang, Qiuzhi Liu, Shuai Wang, Wenxiang Jiao, Zhaopeng Tu

Can’t See the Forest for the Trees: Benchmarking Multimodal Safety Awareness for Multimodal LLMs (2025.acl-long)

Challenge:	Text-based image generation models, such as Stable Diffusion and DALL-E 3, hold significant potential in content creation and publishing workflows . however, considerable efforts are being made to prevent the generation of harmful content, such abusive, violent, or pornographic material.
Approach:	They propose a chain-of-jailbreak method which decomposes malicious queries into multiple sub-queries and iteratively edits images based on these sub-questions.
Outcome:	The proposed method can bypass safeguards of image generation models for over 60% cases, significantly outperforms other jailbreaking methods (14%)

Copied to clipboard

Wenxuan Wang, Xiaoyuan Liu, Kuiyi Gao, Jen-tse Huang, Youliang Yuan, Pinjia He, Shuai Wang, Zhaopeng Tu

On the Reliability of Psychological Scales on Large Language Models (2024.emnlp-main)

Challenge:	Multimodal Large Language Models (MLLMs) have expanded the capabilities of traditional language models by enabling interaction through both text and images.
Approach:	They propose a multimodal safety awareness benchmark to evaluate MLLMs across 29 safety scenarios with 1,500 carefully curated image-prompt pairs.
Outcome:	The proposed model is able to identify unsafe content and avoid over-sensitivity that can hinder helpfulness.

Copied to clipboard

Jen-tse Huang, Wenxiang Jiao, Man Ho Lam, Eric John Li, Wenxuan Wang, Michael Lyu

MT3: A Synergistic Multi-Task RL Framework for Specializing MLLMs in Text Image Machine Translation (2026.acl-long)

Challenge:	Recent research has focused on examining Large Language Models’ characteristics from a psychological standpoint, acknowledging the necessity of understanding their behavioral characteristics.
Approach:	They propose to examine the reliability of personality tests to LLMs by using psychological scales.
Outcome:	The proposed model can represent diverse personalities with specific prompt instructions.

Copied to clipboard

Zhaopeng Feng, Yupu Liang, Shaosheng Cao, Jiayuan Su, Jiahan Ren, Zhijie Zhou, Wenxuan Huang, Jian Wu, Zuozhu Liu

Multilinguality

Challenge:	Text Image Machine Translation (TIMT) is a critical subfield of machine translation . it requires accurate optical character recognition, robust visual-text reasoning, and high-quality translation a challenge .
Approach:	They propose a multi-task optimization framework to specialize MLLMs into expert TIMT models.
Outcome:	The proposed model outperforms baselines on the latest in-domain MIT-10M benchmark.

DimA: A Parameter-efficient Fine-tuning Method with Knowledge Transfer Based on Transformer (2024.lrec-main)

Copied to clipboard

Wenxuan Zhang, Min Huang, Zhuoyang Song, Qinghai Miao

Semantic Text Processing

Challenge:	Pre-trained language models (PLMs) have demonstrated impressive performance across various downstream tasks, but fine-tuning is computationally expensive and storage-intensive.
Approach:	They propose a parameter-efficient method called DimA which enhances the transformer architecture by increasing the dimensionality.
Outcome:	The proposed method achieves state-of-the-art results in GLUE and XSUM tasks while utilizing less than 1% of the original model’s parameters.

Reframe Your Life Story: Interactive Narrative Therapist and Innovative Moment Assessment with Large Language Models (2025.emnlp-main)

Copied to clipboard

Yi Feng, Jiaqi Wang, Wenxuan Zhang, Zhuang Chen, Shen Yutong, Xiyao Xiao, Minlie Huang, Liping Jing, Jian Yu

Probing Multimodal Large Language Models on Cognitive Biases in Chinese Short-Video Misinformation (2026.findings-acl)

Challenge:	Existing approaches to mental health support lack realism and capture therapeutic progression over time.
Approach:	They propose a framework that simulates expert narrative therapists by planning therapeutic stages, guiding reflection levels, and generating contextually appropriate responses through retrieval-augmentation.
Outcome:	The proposed framework outperforms standard methods in quality and depth on 260 simulated clients and 230 human participants.

Copied to clipboard

Jen-tse Huang, Chang Chen, Shiyang Lai, Wenxuan Wang, Michelle R Kaufman, Mark Dredze

MetaScale: Test-Time Scaling with Evolving Meta-Thoughts (2026.findings-acl)

Challenge:	Short-video platforms have become major channels for misinformation, but their robustness against misinformation entangled with cognitive biases remains under-explored.
Approach:	They propose a framework for evaluation of short-video platforms that use visual cues and social cue.
Outcome:	The proposed framework evaluates MLLMs across five modality settings.

Copied to clipboard

Qin Liu, Wenxuan Zhou, Nan Xu, James Y. Huang, Fei Wang, Sheng Zhang, Hoifung Poon, Muhao Chen

LogicAsker: Evaluating and Improving the Logical Reasoning Ability of Large Language Models (2024.emnlp-main)

Challenge:	Existing approaches impose fixed cognitive structures that enhance performance in specific tasks but lack adaptability across diverse scenarios.
Approach:	They propose a test-time scaling framework based on meta-thoughts to improve performance . meta-thinkts are adaptive thinking strategies tailored to a given task .
Outcome:	Experimental results show that MetaScale outperforms standard inference approaches . it can scale more effectively with increasing sampling budgets and produces more structured responses .

Copied to clipboard

Yuxuan Wan, Wenxuan Wang, Yiliu Yang, Youliang Yuan, Jen-tse Huang, Pinjia He, Wenxiang Jiao, Michael Lyu

Pardon? Evaluating Conversational Repair in Large Audio-Language Models (2026.findings-acl)

Challenge:	LogicAsker examines and improves the reasoning abilities of large language models such as ChatGPT and GPT-4.
Approach:	They propose a set of atomic reasoning skills grounded in propositional and predicate logic to examine and improve the reasoning abilities of large language models such as ChatGPT and GPT-4.
Outcome:	The proposed approach improves reasoning abilities in large language models such as ChatGPT and GPT-4 by up to 5%.

Copied to clipboard

Shuanghong Huang, Jinlei Xu, Youchao Zhou, Yanghao Zhou, Xuan Zhao, Chong Feng, Wenxuan Zhang

Natural Language Interfaces

Challenge:	Existing evaluations of large audio-language models focus on answer accuracy and robustness to acoustic perturbations, but they assume that inputs remain semantically answerable.
Approach:	They propose a repair-aware evaluation setting that explicitly distinguishes between answerable and unanswerable audio inputs.
Outcome:	The proposed evaluation setting distinguishes between answerable and unanswerable audio inputs.

JARVIS or Ultron? A Survey on the Safety and Security Threats of Computer-Using Agents (2026.acl-long)

Copied to clipboard

Ada Chen, Yongjiang Wu, Junyuan Zhang, Jingyu Xiao, Shu Yang, Jen-tse Huang, Kun Wang, Wenxuan Wang, Shuai Wang

Challenge:	Recent advances in computer-using agents have created new safety and security risks . despite the impressive capabilities of CUAs, there are still significant security risks.
Approach:	They propose a systematization of knowledge on the safety and security threats of Computer-Using Agents.
Outcome:	The proposed framework provides a framework for assessing the safety and security risks of computer-using agents.

Social Welfare Function Leaderboard: On the Emergence of LLM Agents as the Welfare Dictator (2026.findings-acl)

Copied to clipboard

Zhengliang Shi, Ruotian Ma, Jen-tse Huang, Xinbei Ma, Xingyu Chen, Mengru Wang, Qu Yang, Yue Wang, Fanghua Ye, Ziyang Chen, Shanyi Wang, Cixing LI, Wenxuan Wang, Zhaopeng Tu, Xiaolong Li, Zhaochun Ren, Liefeng Bo

Identifying the Achilles’ Heel: An Iterative Method for Uncovering Factual Errors in Large Language Models (2026.findings-acl)

Challenge:	Large language models (LLMs) are increasingly entrusted with high-stakes decisions that affect human welfare.
Approach:	They evaluate 20 state-of-the-art Large language models (LLMs) and 20 LLM dictators to create a social welfare function benchmark.
Outcome:	The proposed model creates dilemma between maximizing collective efficiency and ensuring distributive fairness.

Copied to clipboard

Wenxuan Wang, Yuk-Kit Chan, Zixuan Ling, Shi Juluan, Youliang Yuan, Jen-tse Huang, Yifei Zhang, Wenxiang Jiao, Zhaopeng Tu, Michael R. Lyu

LongMP-Bench: A Benchmark for Multimodal Persona Understanding in Long-Term Dialogues (2026.findings-acl)

Challenge:	Current methods for evaluating LLMs’ veracity are limited by the need for extensive human labor, test data contamination, or limited scope, hindering efficient and effective exposure of errors.
Approach:	They propose a framework that extracts fact triplets to generate diverse question types using rule-based natural language processing techniques.
Outcome:	The proposed framework can trigger factual errors in up to 55% of questions in large LLMs while maintaining coverage of questions.

Copied to clipboard

Zhuoqun Li, Zhaopei Huang, Wenxuan Wang, Qin Jin

Learning to Ask: When LLM Agents Meet Unclear Instruction (2025.emnlp-main)

Challenge:	Existing datasets suffer from limited persona diversity and static, overly simplified settings, making them insufficient for capturing the complexity of real-world interactions.
Approach:	They propose a benchmark to evaluate models' ability to understand evolving user personas within long-term multimodal dialogues by using a dataset that contains long conversations from 150 users.
Outcome:	The proposed benchmark aims to assess models' ability to track persona evolution, integrate visual and textual inputs, and apply persona understanding in realistic dialogue scenarios.

Copied to clipboard

Wenxuan Wang, Shi Juluan, Zixuan Ling, Yuk-Kit Chan, Chaozheng Wang, Cheryl Lee, Youliang Yuan, Jen-tse Huang, Wenxiang Jiao, Michael R. Lyu

Challenge:	Despite their impressive capabilities, LLMs struggle with complex computations and delivering accurate, timely information.
Approach:	They propose a framework that prompts LLM agents to ask questions when they encounter obstacles due to unclear instructions and an automated evaluation tool called ToolEvaluator.
Outcome:	The proposed framework outperforms existing frameworks for tool learning in the Noisy ToolBench.

mDPO: Conditional Preference Optimization for Multimodal Large Language Models (2024.emnlp-main)

Copied to clipboard

Fei Wang, Wenxuan Zhou, James Y. Huang, Nan Xu, Sheng Zhang, Hoifung Poon, Muhao Chen

AI Sees Your Location—But With A Bias Toward The Wealthy World (2025.emnlp-main)

Challenge:	Recent studies have attempted to apply DPO to multimodal scenarios but have found it challenging to achieve consistent improvement.
Approach:	They propose a multimodal DPO objective that prevents the over-prioritization of language-only preferences by also optimizing image preference.
Outcome:	The proposed method significantly improves performance on two multimodal LLMs of different sizes and three widely used benchmarks.

Copied to clipboard

Jingyuan Huang, Jen-tse Huang, Ziyi Liu, Xiaoyuan Liu, Wenxuan Wang, Jieyu Zhao

Curing Miracle Steps in LLM Mathematical Reasoning with Rubric Rewards (2026.acl-long)

Challenge:	Visual Language Models (VLMs) have shown remarkable performance across various tasks, particularly in recognizing geographic information from images.
Approach:	They propose to use 1,200 images paired with detailed geographic metadata to evaluate VLMs' performance.
Outcome:	The models achieve 53.8% accuracy in city prediction, but exhibit significant biases in regional tasks.

Copied to clipboard

Youliang Yuan, Qiuyang Mang, Jingbang Chen, Hong Wan, Xiaoyuan Liu, Junjielong Xu, Jen-tse Huang, Wenxuan Wang, Wenxiang Jiao, Pinjia He

Reasoning

Challenge:	Existing models are susceptible to reward hacking, leading to a substantial overestimation of a model's reasoning ability.
Approach:	They propose a Rubric Reward Model that rewards the entire reasoning trajectory against problem-specific rubrics.
Outcome:	The proposed model outperforms outcome-only supervision on four math benchmarks and boosts Verified Pass@1024 from 26.7% to 62.6% and reduces the incidence of Miracle Steps by 71%.

Not All Countries Celebrate Thanksgiving: On the Cultural Dominance in Large Language Models (2024.acl-long)

Copied to clipboard

Wenxuan Wang, Wenxiang Jiao, Jingyuan Huang, Ruyi Dai, Jen-tse Huang, Zhaopeng Tu, Michael Lyu

MedEinst: Benchmarking the Einstellung Effect in Medical LLMs through Counterfactual Differential Diagnosis (2026.acl-long)

Challenge:	e.g., ChatGPT often provides inappropriate English-culture-related answers when users ask in non-English languages.
Approach:	They build a benchmark of concrete and abstract cultural objects to evaluate the cultural dominance issue in large language models.
Outcome:	The proposed model can significantly mitigate cultural dominance issue in large language models . the model can provide accurate answers in English, while the model is ethically sound .

Copied to clipboard

Wenting Chen, Guolin Huang, Wenxuan Wang, Zhongrui Zhu

Structured Data in NLP

Challenge:	Existing medical benchmarks fail to detect the Einstellung Effect in clinical diagnosis . Existing models exhibit the Einstellung effect, relying on statistical shortcuts rather than logical reasoning.
Approach:	They propose a counterfactual benchmark that uses statistical shortcuts to diagnose patients . they propose CGME-based system that iteratively refines reasoning paths .
Outcome:	The proposed model achieves high baseline accuracy but severe bias trap rates . iteratively refines reasoning paths in an exemplar base and consolidates disease-specific knowledge into illness graphs.

Contrastive Instruction Tuning (2024.findings-acl)

Copied to clipboard

Tianyi Yan, Fei Wang, James Y. Huang, Wenxuan Zhou, Fan Yin, Aram Galstyan, Wenpeng Yin, Muhao Chen

UniCorn: Towards Self-Improving Unified Multimodal Models through Self-Generated Supervision (2026.acl-long)

Challenge:	Current LLMs exhibit limited robustness to unseen instructions, generating inconsistent outputs when the same instruction is phrased with slightly varied forms or language styles.
Approach:	They propose a method which maximizes the similarity between the hidden representations of semantically equivalent instruction-instance pairs while minimizing the similarities between semantically different ones.
Outcome:	Experiments on the PromptBench benchmark show that Contrastive Instruction Tuning improves LLMs’ robustness to unseen instructions with variations across character, word, sentence, and semantic levels by +2.5% in accuracy.

Copied to clipboard

Zhen Fang, Ruiyan Han, XinYu Sun, Yuchen Ma, Ziheng Wang, Yu Zeng, Zehui Chen, Lin Chen, Wenxuan Huang, Wei-Jie Xu, Yi Cao, Feng Zhao

Parameter-Efficient Tuning with Special Token Adaptation (2023.eacl-main)

Challenge:	Unified Multimodal Models have achieved remarkable success in cross-modal comprehension, but a gap persists in their ability to translate internal knowledge into faithful and controllable synthesis.
Approach:	They propose a self-improvement framework that partitions a single UMM into three collaborative roles: Proposer, Solver, and Judge.
Outcome:	The proposed framework improves on TIIF, DPG, CompBench and UniCycle benchmarks.

Copied to clipboard

Xiaocong Yang, James Y. Huang, Wenxuan Zhou, Muhao Chen

Towards Generalizable and Robust Text-to-SQL Parsing (2022.findings-emnlp)

Challenge:	a recent study shows that parameter-efficient tuning is a challenge for multitask deployments.
Approach:	They propose a parameter-efficient tuning technique that only updates a small subset of parameters when adapting a pretrained model to downstream tasks.
Outcome:	The proposed method achieves comparable performance to fine-tuning in natural language understanding tasks including text classification and NER with only 0.029% of parameters trained.

Copied to clipboard

Chang Gao, Bowen Li, Wenxuan Zhang, Wai Lam, Binhua Li, Fei Huang, Luo Si, Yongbin Li

Programming Languages in NLP

Challenge:	Text-to-SQL parsers must be generalizable and robust against input perturbations.
Approach:	They propose a novel framework to learn text-to-SQL parsing in stages to improve parser's ability to acquire general SQL knowledge instead of capturing spurious patterns.
Outcome:	The proposed framework achieves state-of-the-art performance on the Spider, SParC, and CoSQL datasets.

Robust Natural Language Understanding with Residual Attention Debiasing (2023.findings-acl)

Copied to clipboard

Fei Wang, James Y. Huang, Tianyi Yan, Wenxuan Zhou, Muhao Chen

Robustness in NLP

Challenge:	Existing ensemble-based debiasing methods do not address unintended dataset biases . attention plays a crucial role in providing robust prediction in NLU models .
Approach:	They propose an end-to-end debiasing method that mitigates unintended biases from attention.
Outcome:	The proposed method improves the OOD performance of BERT-based models on three benchmarks.

VisBias: Measuring Explicit and Implicit Social Biases in Vision Language Models (2025.emnlp-main)

Copied to clipboard

Jen-tse Huang, Jiantong Qin, Jianping Zhang, Youliang Yuan, Wenxuan Wang, Jieyu Zhao

Visual Data in NLP

Challenge:	Identifying and addressing potential social biases is essential to prevent harm to users.
Approach:	They examine explicit and implicit biases exhibited by Vision-Language Models . they pose questions related to gender and racial differences to test their models .
Outcome:	The proposed models are used in image description tasks, form completion tasks and medical applications.

Hiring Now: A Skill-Aware Multi-Attention Model for Job Posting Generation (2020.acl-main)

Copied to clipboard

Liting Liu, Jie Liu, Wenzheng Zhang, Ziming Chi, Wenxuan Shi, Yalou Huang

Challenge:	Creating job requirements is a crucial step in the recruiting process, but it is difficult to specify the level of education, experience, relevant skills per the job description.
Approach:	They propose a conditional text generation task to generate job requirements based on job descriptions . they use a hierarchical decoder to label the job description with multiple skills . a skill knowledge graph is constructed to capture the global prior knowledge about skills based upon the model .
Outcome:	The proposed method is evaluated on real-world job posting data.

Asclepius: A Spectrum Evaluation Benchmark for Medical Multi-Modal Large Language Models (2025.acl-long)

Copied to clipboard

Jie Liu, Wenxuan Wang, Su Yihang, Jingyuan Huang, Yudi Zhang, Cheng-Yi Li, Wenting Chen, Xiaohan Xing, Kao-Jung Chang, Linlin Shen, Michael R. Lyu

Semantic Text Processing

Challenge:	Medical Multi-Modal Large Language Models (Med-MLLMs) are a promising new form of artificial general intelligence due to their ability to tackle complex tasks.
Approach:	They propose a new benchmark that comprehensively assesses medical multi-modal large language models in terms of distinct medical specialties and different diagnostic capacities.
Outcome:	The proposed model covers 15 medical specialties and different diagnostic capacities, and excludes overlap with existing VQA dataset.

All Languages Matter: On the Multilingual Safety of LLMs (2024.findings-acl)

Copied to clipboard

Wenxuan Wang, Zhaopeng Tu, Chang Chen, Youliang Yuan, Jen-tse Huang, Wenxiang Jiao, Michael Lyu