Papers by Shihan Dou
Muse: Towards Reproducible Long-Form Song Generation with Fine-Grained Style Control (2026.findings-acl)
Copied to clipboard
Changhao Jiang, Jiahao Chen, Zhenghao Xiang, Zhixiong Yang, Hanchen Wang, Jiabao Zhuang, Xinmeng Che, Jiajun Sun, Hui Li, Yifei Cao, Shihan Dou, Ming Zhang, Junjie Ye, Tao Ji, Tao Gui, Qi Zhang, Xuanjing Huang
| Challenge: | Recent commercial systems such as Suno demonstrate strong capabilities in long-form song generation, but academic research remains non-reproducible due to the lack of publicly available training data. |
| Approach: | They propose a system for long-form song generation with fine-grained style conditioning that includes a licensed synthetic dataset and a song generation model, Muse. |
| Outcome: | The proposed system achieves competitive performance on phoneme error rate, text–music style similarity, and audio aesthetic quality while enabling controllable segment-level generation across different musical structures. |
LoRAMoE: Alleviating World Knowledge Forgetting in Large Language Models via MoE-Style Plugin (2024.acl-long)
Copied to clipboard
Shihan Dou, Enyu Zhou, Yan Liu, Songyang Gao, Wei Shen, Limao Xiong, Yuhao Zhou, Xiao Wang, Zhiheng Xi, Xiaoran Fan, Shiliang Pu, Jiang Zhu, Rui Zheng, Tao Gui, Qi Zhang, Xuanjing Huang
| Challenge: | Experimental results show that, as the instruction data increases, LoRAMoE can significantly improve the ability to process downstream tasks, while maintaining the world knowledge stored in the LLM. |
| Approach: | They propose a framework that introduces several low-rank adapters and integrates them by using a router network to freeze the backbone model and force a portion of LoRAs to focus on leveraging world knowledge to solve downstream tasks. |
| Outcome: | The proposed framework freezes the backbone model and forces a portion of LoRAs to focus on leveraging world knowledge to solve downstream tasks, to alleviate world knowledge forgetting. |
Decorrelate Irrelevant, Purify Relevant: Overcome Textual Spurious Correlations from a Feature Perspective (2022.coling-1)
Copied to clipboard
| Challenge: | Existing methods to debiase samples with biased features obstructs the model in learning from non-biased parts of the samples. |
| Approach: | They propose to eliminate spurious correlations in a fine-grained manner from a feature space perspective by using Random Fourier Features and weighted re-sampling to decorrelate dependencies between features. |
| Outcome: | The proposed method eliminates spurious correlations in a fine-grained manner from a feature space perspective. |
MINER: Improving Out-of-Vocabulary Named Entity Recognition from an Information Theoretic Perspective (2022.acl-long)
Copied to clipboard
Xiao Wang, Shihan Dou, Limao Xiong, Yicheng Zou, Qi Zhang, Tao Gui, Liang Qiao, Zhanzhan Cheng, Xuanjing Huang
| Challenge: | Named Entity Recognition models are feature-engineering and machine learning based. |
| Approach: | They propose a new NER learning framework that uses entity mentions to improve model performance. |
| Outcome: | The proposed model achieves better performance on OOV entities on various settings and datasets. |
PRISM: Probabilistic Reward Model with Inherent Structural Modeling (2026.acl-long)
Copied to clipboard
| Challenge: | Existing evaluators compress diverse human judgments into a single scalar, leading to brittle alignment and reward hacking. |
| Approach: | They propose a Gaussian-based reinterpretation of reward evaluation as a conditional distribution and a mixture of Gaussians to capture conflicting preference dimensions. |
| Outcome: | The proposed model outperforms scalar baselines in accuracy and generalization. |
Governance in Motion: Co-evolution of Constitutions and AI models for Scalable Safety (2025.emnlp-main)
Copied to clipboard
Chenhao Huang, Ziyu Shen, Yicong Ren, Huiyuan Zheng, Jiazheng Zhang, Mingxu Chai, Ming Zhang, Shihan Dou, Fan Mo, Jie Shi, Tao Gui, Qi Zhang, Xuanjing Huang
| Challenge: | Existing approaches to align large language models with human preferences lack flexibility . static alignment preferences lack the ability to correct misaligned behaviors as they emerge . |
| Approach: | They propose a framework that enables dynamic and continuous alignment of large language models with human preferences. |
| Outcome: | The proposed framework improves safety and accuracy of a 7B model with human annotations. |
DARM: Distribution-Aware Reward Modeling by Alleviating Biases from Low Preference-Context Dependency Data (2026.acl-long)
Copied to clipboard
Shaofan Liu, Guoqiang Zhang, Shihan Dou, Huiyuan Zheng, Yiming Zhou, Junjie Ye, Shaowen Wang, Shichun Liu, Jiazheng Zhang, Tao Gui, Qi Zhang, Xuanjing Huang
| Challenge: | Existing methods for training reward models are vulnerable to context neglect and degraded accuracy. |
| Approach: | They propose distribution-aware reward modeling that augments the RM objective with a conditional mutual information regularizer that maximizes context and the predicted reward conditioned on the response. |
| Outcome: | The proposed model improves performance in RLHF and improves accuracy in other settings. |
TransferTOD: A Generalizable Chinese Multi-Domain Task-Oriented Dialogue System with Transfer Capabilities (2024.emnlp-main)
Copied to clipboard
Ming Zhang, Caishuang Huang, Yilong Wu, Shichun Liu, Huiyuan Zheng, Yurui Dong, Yujiong Shen, Shihan Dou, Jun Zhao, Junjie Ye, Qi Zhang, Tao Gui, Xuanjing Huang
| Challenge: | Current datasets cater to user-led systems and are limited to predefined specific scenarios and slots. |
| Approach: | They propose to use a Chinese dialogue dataset to train a model that authentically simulates human-computer dialogues in 30 popular life service scenarios. |
| Outcome: | The proposed model achieves a joint accuracy of 75.09% in out-of-domain evaluations . it also achieves notable abilities in slot filling and questioning . |
Kernel-Whitening: Overcome Dataset Bias with Isotropic Sentence Embedding (2022.emnlp-main)
Copied to clipboard
| Challenge: | Existing approaches to reduce dataset bias rely on spurious correlations and obstruct valid feature information while mitigating bias. |
| Approach: | They propose a representation normalization method which disentangles correlations between features of encoded sentences and a kernel approximation method which provides isotropic data distribution. |
| Outcome: | The proposed method eliminates the bias problem by providing isotropic data distribution while maintaining in-distribution accuracy. |
PFDial: A Structured Dialogue Instruction Fine-tuning Method Based on UML Flowcharts (2025.findings-acl)
Copied to clipboard
Ming Zhang, Yuhui Wang, Yujiong Shen, Tingyi Yang, Changhao Jiang, Yilong Wu, Shihan Dou, Qinhao Chen, Zhiheng Xi, Zhihao Zhang, Yi Dong, Zhen Wang, Zhihui Fei, Mingyang Wan, Tao Liang, Guojun Ma, Qi Zhang, Tao Gui, Xuanjing Huang
| Challenge: | Large Language Models (LLMs) have shown remarkable progress in dialogue and reasoning, but they struggle to solve strictly constrained dialogue tasks. |
| Approach: | They construct a dataset that contains 12,705 high-quality Chinese dialogue instructions from 440 flowcharts containing 5,055 process nodes. |
| Outcome: | The proposed model outperforms GPT-4o models on backward transitions and outperformed GPT-42 models on the same dataset. |
OctoBench: Benchmarking Scaffold-Aware Instruction Following in Repository-Grounded Agentic Coding (2026.acl-long)
Copied to clipboard
Deming Ding, Shichun Liu, Enhui Yang, Jiahang Lin, Ziying Chen, Shihan Dou, Honglin Guo, Weiyu Cheng, Pengyu Zhao, Chengjun Xiao, Qunhong Zeng, Qi Zhang, Xuanjing Huang, Qidi Xu, Tao Gui
| Challenge: | coding scaffolds that follow heterogeneous instructions remain under-examined in software engineering . coding models are capable software agents, but their ability to follow constraints remains under-explored . |
| Approach: | They introduce OctoBench, which benchmarks scaffold-aware instruction following in agentic coding. |
| Outcome: | The proposed benchmark aims to accelerate the development of more scaffold-aware agents. |
Multi-Programming Language Sandbox for LLMs (2025.acl-demo)
Copied to clipboard
Shihan Dou, Jiazheng Zhang, Jianxiang Zang, Yunbo Tao, Weikang Zhou, Haoxiang Jia, Shichun Liu, Yuming Yang, Shenxi Wu, Zhiheng Xi, Muling Wu, Rui Zheng, Changze Lv, Limao Xiong, Shaoqing Zhang, Lin Zhang, Wenyu Zhan, Rongxiang Weng, Jingang Wang, Xunliang Cai, Yueming Wu, Ming Wen, Yixin Cao, Tao Gui, Xipeng Qiu, Qi Zhang, Xuanjing Huang
| Challenge: | MPLSandbox is an out-of-the-box multi-programming language sandbox designed to provide unified and comprehensive feedback from compiler and analysis tools for Large Language Models (LLMs). |
| Approach: | They propose a multi-programming language sandbox that provides unified feedback from compilers and analysis tools for Large Language Models. |
| Outcome: | The proposed multi-language sandbox can provide comprehensive feedback from compilers and analysis tools for large language models (LLMs). |
LLMEval-Fair: A Large-Scale Longitudinal Study on Robust and Fair Evaluation of Large Language Models (2026.acl-long)
Copied to clipboard
Ming Zhang, Yujiong Shen, Jingyi Deng, Yuhui Wang, Huayu Sha, Kexin Tan, Qiyuan Peng, Yue Zhang, Junzhe Wang, Shichun Liu, Yueyuan Huang, Jingqi Tong, Changhao Jiang, Yilong Wu, Zhihao Zhang, Mingqi Wu, Mingxu Chai, Zhiheng Xi, Shihan Dou, Tao Gui, Qi Zhang, Xuanjing Huang
| Challenge: | Existing evaluation of Large Language Models on static benchmarks is vulnerable to data contamination and leaderboard overfitting. |
| Approach: | LLMEval-Fair framework provides a framework for dynamic evaluation of Large Language Models . evaluators use a proprietary bank of 220k graduate-level questions to analyze model data . |
| Outcome: | LLMEval-Fair provides robust and credible evaluation framework for Large Language Models . it provides a strong empirical validation for the dynamic evaluation paradigm . |
Measuring Data Diversity for Instruction Tuning: A Systematic Analysis and A Reliable Metric (2025.acl-long)
Copied to clipboard
Yuming Yang, Yang Nan, Junjie Ye, Shihan Dou, Xiao Wang, Shuo Li, Huijie Lv, Tao Gui, Qi Zhang, Xuanjing Huang
| Challenge: | Existing studies have explored various diversity-aware data selection methods to construct high-quality datasets and enhance model performance. |
| Approach: | They propose to use data diversity to measure instruction tuning of large language models. |
| Outcome: | The proposed diversity metric outperforms existing methods on simulated and real-world data and shows that it captures diversity variations and achieves a 0.97 correlation with instruction tuning. |
Detecting Adversarial Samples through Sharpness of Loss Landscape (2023.findings-acl)
Copied to clipboard
Rui Zheng, Shihan Dou, Yuhao Zhou, Qin Liu, Tao Gui, Qi Zhang, Zhongyu Wei, Xuanjing Huang, Menghan Zhang
| Challenge: | Existing studies have shown that adversarial samples are more vulnerable than normal ones to textual adversarials. |
| Approach: | They propose a simple and effective sharpness-based detector that can distinguish adversarial samples by maximizing the loss increment within the region where the inference sample is located. |
| Outcome: | The proposed method outperforms previous detection methods by large margins on three text classification tasks. |
JanusMM: A Benchmark for Self-Deprecation Understanding in Real-World Multimodal Conversations (2026.acl-long)
Copied to clipboard
Xinyi Xu, Bingguang Hao, Yongyi Xiong, Zimo Chen, Xinchen Liu, Hongxin Guo, Xuelong Wang, Silin Zhou, Shihan Dou
| Challenge: | Multimodal large language models (MLLMs) are a common communicative strategy in human society, often using image-text interplay to express emotions and intentions. |
| Approach: | They propose to evaluate multimodal large language models (MLLMs)' understanding of self-deprecation in real-world conversations using 2,016 bilingual memes. |
| Outcome: | The proposed framework evaluates MLLMs' understanding of self-deprecation in real-world conversations. |
Enhancing LLM-based Search Agents via Contribution Weighted Group Relative Policy Optimization (2026.acl-long)
Copied to clipboard
| Challenge: | Existing approaches to training large language models suffer from unstable value estimation, whereas outcome supervision struggles with credit assignment due to sparse, trajectory-level rewards. |
| Approach: | They propose a framework that integrates process supervision into group relative policy optimization. |
| Outcome: | The proposed framework outperforms standard GRPO on knowledge-intensive benchmarks by 5.0% and 6.3% on Qwen3-1.7B. |
MM-Doc-R1: Training Agents for Long Document Visual Question Answering through Multi-turn Reinforcement Learning (2026.findings-acl)
Copied to clipboard
Jiahang Lin, Kai Hu, Binghai Wang, Yuhao Zhou, Zhiheng Xi, Honglin Guo, Shichun Liu, Junzhe Wang, Shihan Dou, Enyu Zhou, Hang Yan, Zhenhua Han, Tao Gui, Qi Zhang, Xuanjing Huang
| Challenge: | Existing work on long document visual question answering is based on Retrieval-Augmented Generation (RAG) where textual or visual content is encoded into embeddings and relevance is determined by similarity scores with respect to the original query. |
| Approach: | They propose a framework that employs an agentic, vision-aware workflow to address long document visual question answering through iterative information discovery and synthesis. |
| Outcome: | The proposed framework outperforms existing RL systems by 10.4% on the MMLongbench-Doc benchmark and demonstrates superior training performance over GRPO. |
LLMEval-Med: A Real-world Clinical Benchmark for Medical LLMs with Physician Validation (2025.findings-emnlp)
Copied to clipboard
Ming Zhang, Yujiong Shen, Zelin Li, Huayu Sha, Binze Hu, Yuhui Wang, Chenhao Huang, Shichun Liu, Jingqi Tong, Changhao Jiang, Mingxu Chai, Zhiheng Xi, Shihan Dou, Tao Gui, Qi Zhang, Xuanjing Huang
| Challenge: | Current medical benchmarks have limitations in question design, data sources and evaluation methods. |
| Approach: | They propose a new benchmark covering five core medical areas . it includes 2,996 questions created from real-world electronic health records . |
| Outcome: | The proposed model covers five core medical areas and includes 2,996 questions created from real-world electronic health records and expert-designed clinical scenarios. |
UPLex: Fine-Grained Personality Control in Large Language Models via Unsupervised Lexical Modulation (2025.findings-emnlp)
Copied to clipboard
Tianlong Li, Wenhao Liu, Muling Wu, Shihan Dou, Zhenghua Wang, Changze Lv, Xiaohua Wang, Xiaoqing Zheng, Xuanjing Huang
| Challenge: | Personality is a crucial factor that shapes human communication patterns, thereby regulating the personalities of large language models (LLMs). |
| Approach: | They propose a method that uses an Unsupervisedly-Built Personalized Lexicon (UPL) during the decoding phase to manipulate LLM’s personality traits. |
| Outcome: | The proposed method can modulate the personality expression of large language models by dynamically altering their predicted probability of upcoming words in a pluggable fashion. |
ToolEyes: Fine-Grained Evaluation for Tool Learning Capabilities of Large Language Models in Real-world Scenarios (2025.coling-main)
Copied to clipboard
Junjie Ye, Guanyu Li, SongYang Gao, Caishuang Huang, Yilong Wu, Sixian Li, Xiaoran Fan, Shihan Dou, Tao Ji, Qi Zhang, Tao Gui, Xuanjing Huang
| Challenge: | Existing evaluations of tool learning focus on validation of tools for large language models with expected outcomes, but this focus ignores the complex capabilities required for LLMs to effectively use tools. |
| Approach: | They propose a fine-grained system for evaluation of large language models’ tool learning capabilities in authentic scenarios. |
| Outcome: | The proposed system examines seven real-world scenarios, analyzing five dimensions crucial to LLMs in tool learning: format alignment, intent comprehension, behavior planning, tool selection, and answer organization. |
Lost in the Context: Insufficient and Distracted Attention to Contexts in Preference Modeling (2025.acl-long)
Copied to clipboard
Shihan Dou, Jiayi Chen, Chenhao Huang, Feng Chen, Wei Chengzhi, Huiyuan Zheng, Shichun Liu, Yan Liu, Chenxiao Liu, Chao Xin, Lin Yan, Zongzhang Zhang, Tao Gui, Qi Zhang, Xuanjing Huang
| Challenge: | Existing reward models concatenate contexts and responses, but they often ignore crucial segments of the context that are important for evaluating the response quality. |
| Approach: | They propose a reward model that evaluates the response quality based on a given context and assigns a rewards reward. |
| Outcome: | The proposed framework significantly improves preference modeling by increasing attention to relevant information within the context and achieves better generalizability. |
On the Universal Adversarial Perturbations for Efficient Data-free Adversarial Detection (2023.findings-acl)
Copied to clipboard
| Challenge: | Existing adversarial detection methods require access to training data, which brings noteworthy concerns regarding privacy leakage and generalizability. |
| Approach: | They propose a data-agnostic adversarial detection framework which induces different responses between normal and adversarials to UAPs. |
| Outcome: | The proposed framework achieves competitive detection performance on various text classification tasks, and maintains equivalent time consumption to normal inference. |
StepCoder: Improving Code Generation with Reinforcement Learning from Compiler Feedback (2024.acl-long)
Copied to clipboard
Shihan Dou, Yan Liu, Haoxiang Jia, Enyu Zhou, Limao Xiong, Junjie Shan, Caishuang Huang, Xiao Wang, Xiaoran Fan, Zhiheng Xi, Yuhao Zhou, Tao Ji, Rui Zheng, Qi Zhang, Tao Gui, Xuanjing Huang
| Challenge: | Existing work integrates reinforcement learning with compiler feedback to enhance code generation quality but the long code generated by LLMs makes RL exploration ineffective. |
| Approach: | They propose a framework that integrates reinforcement learning and compiler feedback to enhance code generation quality. |
| Outcome: | The proposed framework outperforms state-of-the-art approaches in corresponding benchmarks and integrates reinforcement learning with compiler feedback to improve code generation quality. |
Revisiting Jailbreaking for Large Language Models: A Representation Engineering Perspective (2025.coling-main)
Copied to clipboard
Tianlong Li, Zhenghua Wang, Wenhao Liu, Muling Wu, Shihan Dou, Changze Lv, Xiaohua Wang, Xiaoqing Zheng, Xuanjing Huang
| Challenge: | Recent surge in jailbreaking attacks has revealed significant vulnerabilities in Large Language Models (LLMs) however, limited research into the underlying mechanisms that make LLMs vulnerable to such attacks has been conducted. |
| Approach: | They propose that LLMs' self-safeguarding capability is linked to specific activity patterns within their representation space. |
| Outcome: | The proposed models can be detected with a few pairs of contrastive queries, and the robustness can be manipulated by weakening or strengthening these patterns. |
VRPO: Rethinking Value Modeling for Robust RL under Noisy Supervision in LLM Post-Training (2026.acl-long)
Copied to clipboard
Dingwei Zhu, Shihan Dou, Zhiheng Xi, Senjie Jin, Guoqiang Zhang, Jiazheng Zhang, Junjie Ye, Mingxu Chai, Enyu Zhou, Ming Zhang, Yuhui Wang, Caishuang Huang, Chenhao Huang, Yunke Zhang, Yuran Wang, Tao Gui, Qi Zhang, Xipeng Qiu, Xuanjing Huang
| Challenge: | Reinforcement Learning (RL) in real-world environments often suffers from ambiguous or incomplete supervision. |
| Approach: | They propose a framework that enhances value modeling for robust RL in LLM post-training by integrating auxiliary losses guided by entropy and perplexity from a frozen language model and variational information bottleneck. |
| Outcome: | The proposed framework outperforms baselines on multi-turn dialogue, math reasoning, and science QA with rule-based and model-based rewards. |
Loose lips sink ships: Mitigating Length Bias in Reinforcement Learning from Human Feedback (2023.findings-emnlp)
Copied to clipboard
| Challenge: | Experimental results prove that language models can learn from human feedback better, irrespective of sequence length . emergence of length bias often induces the model to favor longer outputs . |
| Approach: | They propose to separate reward modeling from the influence of sequence length by using the Product-of-Experts technique. |
| Outcome: | The proposed approach shows that language models perform better regardless of sequence length . the main expert is focused on understanding human intents, while the biased expert targets the identification and capture of length bias. |
From Scores to Preferences: Redefining Evaluation Paradigm for Speech Quality Reward Modeling (2026.findings-acl)
Copied to clipboard
Yifei Cao, Changhao Jiang, Jiabao Zhuang, Jiajun Sun, Ming Zhang, Zhiheng Xi, Hui Li, Shihan Dou, Yuran Wang, Yunke Zhang, Tao Ji, Tao Gui, Qi Zhang, Xuanjing Huang
| Challenge: | Experimental results show that the MOS-aware GRM significantly improves fine-grained speech quality discrimination. |
| Approach: | They propose a MOS-aware reward model that incorporates MOS gap into reward function during reinforcement learning. |
| Outcome: | The proposed model significantly improves fine-grained speech quality discrimination. |
DocFusion: A Unified Framework for Document Parsing Tasks (2025.findings-acl)
Copied to clipboard
Mingxu Chai, Ziyu Shen, Chong Zhang, Yue Zhang, Xiao Wang, Shihan Dou, Jihua Kang, Jiazheng Zhang, Qi Zhang
| Challenge: | Existing methods for document parsing often employ multiple models, limiting performance . Existing models often employ discrete tokens, whereas recognition relies on continuous coordinates . |
| Approach: | They propose a Gaussian-Kernel Cross-Entropy Loss (GK-CEL) that unifies detection and recognition by enabling generative frameworks to handle both tasks simultaneously. |
| Outcome: | The proposed model performs competitively across four core document parsing tasks. |
DSRM: Boost Textual Adversarial Training with Distribution Shift Risk Minimization (2023.acl-long)
Copied to clipboard
| Challenge: | Existing adversarial training methods require multi-step gradient ascents or word substitutions to obtain adversarials, which impairs the effectiveness of adversariarial training. |
| Approach: | They propose a procedure for instead adversarial training with only clean data that estimates the adversarials loss by perturbing the input data’s probability distribution rather than their embeddings. |
| Outcome: | The proposed procedure reduces time consumption by up to 70% compared to current best-performing adversarial training methods. |
Beyond Scaling: Measuring and Predicting the Upper Bound of Knowledge Retention in Language Model Pre-Training (2026.acl-long)
Copied to clipboard
Changhao Jiang, Ming Zhang, Yifei Cao, Junjie Ye, Xiaoran Fan, Shihan Dou, Zhiheng Xi, Jiajun Sun, Yi Dong, Yujiong Shen, Jingqi Tong, Baoyu Fan, Tao Gui, Qi Zhang, Xuanjing Huang
| Challenge: | Existing methods to predict performance of large language models are lacking . authors propose a size-dependent mutual information predictor for closed-book question answering accuracy . |
| Approach: | They propose a size-dependent mutual information predictor that integrates knowledge frequency, knowledge specificity, and model size to forecast closed-book question answering accuracy. |
| Outcome: | The proposed method outperforms baseline models and achieves R2 > 0.7 in predicting QA accuracy without additional training. |