Papers by Xuanjing Huang

Math Word Problem Solving with Explicit Numerical Values (2021.acl-long)

Challenge:	Currently, pre-trained language model (PLM) based metrics are widely adopted in text generation tasks.
Approach:	They propose to use PLMs to encode stereotypical societal biases in PLM-based metrics . they show that popular metrics exhibit higher social bias than traditional metrics based on 6 attributes .
Outcome:	The proposed method shows that PLM-based metrics exhibit higher social bias than traditional metrics on 6 attributes.

Copied to clipboard

Qinzhuo Wu, Qi Zhang, Zhongyu Wei, Xuanjing Huang

Numerical Reasoning

Challenge:	Existing methods for solving math word problems ignore numerical values in solving problems.
Approach:	They propose a numerically-based approach that explicitly incorporates numerical values into a sequence-to-tree network and uses a mathematical properties prediction mechanism to capture category and comparison information of numerals.
Outcome:	The proposed model outperforms existing state-of-the-art models on the Math23K and APE datasets.

HAF-RM: A Hybrid Alignment Framework for Reward Model Training (2025.acl-long)

Copied to clipboard

Shujun Liu, Xiaoyu Shen, Yuhang Lai, Siyuan Wang, Shengbin Yue, Zengfeng Huang, Xuanjing Huang, Zhongyu Wei

Convolutional Interaction Network for Natural Language Inference (D18-1)

Challenge:	Recent studies have focused on enhancing reward models through data improvements, following the conventional training framework for reward models that directly optimizes the predicted rewards.
Approach:	They propose a hybrid alignment framework HAF-RM that incorporates additional constraint on token-level policy probabilities in addition to the reward score.
Outcome:	The proposed framework can supervise the internal preference model at the token level and optimize the mapping layer of the reward model at sequence level.

Copied to clipboard

Jingjing Gong, Xipeng Qiu, Xinchi Chen, Dong Liang, Xuanjing Huang

Better Process Supervision with Bi-directional Rewarding Signals (2025.findings-acl)

Challenge:	Attention-based neural models have achieved great success in natural language inference (NLI).
Approach:	They propose a general model to capture the interaction between two sentences, which can be an alternative to the attention mechanism for NLI.
Outcome:	The proposed model can capture complex interactions on three large datasets.

Copied to clipboard

Wenxiang Chen, Wei He, Zhiheng Xi, Honglin Guo, Boyang Hong, Jiazheng Zhang, Nijun Li, Tao Gui, Yun Li, Qi Zhang, Xuanjing Huang

FinToolSyn: A forward synthesis Framework for Financial Tool-Use Dialogue Data with Dynamic Tool Retrieval (2026.findings-acl)

Challenge:	Existing processes that reward for each step are one-directional and lack a mechanism to model the distance to the final target.
Approach:	They propose a process supervision model that evaluates the correctness of previous steps and the probability of future success.
Outcome:	The proposed model outperforms existing supervision models like ORM and PRM on reasoning tasks and improves solution re-design.

Copied to clipboard

Caishuang Huang, Yang Qiao, Rongyu Zhang, Junjie Ye, Pu Lu, null Wuwenxi, Meng Zhou, Xiku Du, Qi Zhang, Tao Gui, Xuanjing Huang

A Graph-based Model for Joint Chinese Word Segmentation and Dependency Parsing (2020.tacl-1)

Challenge:	Existing data synthesis methods rely on static tools to generate queries . this approach fails to capture the implicit, event-driven nature of real-world needs .
Approach:	They propose a forward synthesis framework to generate high-quality financial dialogues . they construct a repository of 43,066 tools and synthesize over 148k dialogue instances .
Outcome:	Experiments show that models trained on FinToolSyn achieve a 21.06% improvement . the framework is designed to generate high-quality financial dialogues .

Copied to clipboard

Hang Yan, Xipeng Qiu, Xuanjing Huang

TripTailor: A Real-World Benchmark for Personalized Travel Planning (2025.findings-acl)

Challenge:	Chinese word segmentation and dependency parsing suffer from error propagation . a graph-based model can integrate both tasks, but it suffers from performance limitations .
Approach:	They propose a graph-based model to integrate Chinese word segmentation and dependency parsing . their model achieves better performance than previous joint models .
Outcome:	The proposed model achieves better performance than previous joint models and state-of-the-art results in both Chinese word segmentation and dependency parsing.

Copied to clipboard

Kaimin Wang, Yuanzhe Shen, Changze Lv, Xiaoqing Zheng, Xuanjing Huang

Challenge:	Existing evaluation metrics for travel planning rely on unrealistic simulated data . fewer than 10% of the itineraries generated by the latest state-of-the-art LLMs achieve human-level performance.
Approach:	They propose a benchmark for personalized travel planning in real-world scenarios . they identify several critical challenges in travel planning including feasibility and rationality .
Outcome:	The proposed benchmarks show that fewer than 10% of the itineraries generated by the latest state-of-the-art LLMs achieve human-level performance.

AgentGym2: Benchmarking Large Language Model Agents in De-Idealized Real-World Environments (2026.acl-long)

Copied to clipboard

Zhiheng Xi, Dingwen Yang, Jiaqi Liu, Jixuan Huang, Honglin Guo, Baodai Huang, Tinggang Chen, Qi Zhang, Zhonghang Lu, Chenyu Liu, Jiajun Sun, Jiazheng Zhang, Dingwei Zhu, Xin Guo, Junzhe Wang, Zhihao Zhang, Yuming Yang, Junjie Ye, Minghe Gao, Dongrui Liu, Jiaming Ji, Guohao Li, Tao Gui, Qi Zhang, Xuanjing Huang

LoRAMoE: Alleviating World Knowledge Forgetting in Large Language Models via MoE-Style Plugin (2024.acl-long)

Challenge:	Existing benchmarks evaluate agents in simplified, idealized settings, relying on pre-packaged tool interfaces, overlooking critical steps, and assume inputs are clean and fully specified.
Approach:	They propose a framework that evaluates language agents in simplified, idealized settings . they show that even SOTA systems like Gemini and GPT-5 struggle on AgentGym2 .
Outcome:	Experiments on 15 proprietary and open-source models show that even SOTA systems like Gemini and GPT-5 struggle on AgentGym2 .

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

Argue with Me Tersely: Towards Sentence-Level Counter-Argument Generation (2023.emnlp-main)

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.

Copied to clipboard

Jiayu Lin, Rong Ye, Meng Han, Qi Zhang, Ruofei Lai, Xinyu Zhang, Zhao Cao, Xuanjing Huang, Zhongyu Wei

Decorrelate Irrelevant, Purify Relevant: Overcome Textual Spurious Correlations from a Feature Perspective (2022.coling-1)

Challenge:	Existing work describes paragraph-level counter-argument generation task as paragraph-based . however, sentence-level generation can be quite different due to its unique constraints and brevity-focused challenges.
Approach:	They propose a benchmark framework for sentence-level counter-argument generation . they use an annotated debate forum dataset to generate high-quality counter-argments .
Outcome:	The proposed framework and evaluator are competitive in counter-argument generation tasks.

Copied to clipboard

Shihan Dou, Rui Zheng, Ting Wu, SongYang Gao, Junjie Shan, Qi Zhang, Yueming Wu, Xuanjing Huang

Enhancing Model Privacy in Federated Learning with Random Masking and Quantization (2025.findings-emnlp)

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.

Copied to clipboard

Zhibo Xu, Zhu JianHao, Jingwen Xu, Changze Lv, Zhenghua Wang, Zisu Huang, Xiaohua Wang, Muling Wu, Qi Qian, Xiaoqing Zheng, Xuanjing Huang

Causal Intervention Improves Implicit Sentiment Analysis (2022.coling-1)

Challenge:	federated learning approaches are limited by the complexity of large language models and the need for specialized expertise to protect intellectual property.
Approach:	They propose a federated learning approach that leverages random masking to obscure a subnetwork of model parameters and applies quantization to the remaining parameters.
Outcome:	The proposed approach maintains strong model performance in federated learning settings and achieves enhanced protection of model parameters compared to baseline methods.

Copied to clipboard

Siyin Wang, Jie Zhou, Changzhi Sun, Junjie Ye, Tao Gui, Qi Zhang, Xuanjing Huang

Sentiment Analysis

Challenge:	Existing neural models struggle with implicit sentiment analysis because they latch onto spurious correlations, resulting in poor generalization and robustness.
Approach:	They propose a CausaL intervention model for implicit sEntiment ANalysis using instrumental variable to eliminate confounding causal effects and extract the pure causal effect between sentence and sentiment.
Outcome:	The proposed model extracts the pure causal effect between sentence and sentiment using instrumental variable.

TextMixer: Mixing Multiple Inputs for Privacy-Preserving Inference (2023.findings-emnlp)

Copied to clipboard

Xin Zhou, Yi Lu, Ruotian Ma, Tao Gui, Qi Zhang, Xuanjing Huang

Heterogeneous Graph Neural Networks for Extractive Document Summarization (2020.acl-main)

Challenge:	Pre-trained language models (PLMs) are often deployed as cloud services, enabling users to upload textual data and perform inference remotely.
Approach:	They propose a privacy-preserving inference framework called MixPi which aims to obfuscate a user's private input by mixing it with multiple other inputs.
Outcome:	The proposed framework surpasses existing privacy-preserving methods on token and sentence classification tasks.

Copied to clipboard

Danqing Wang, Pengfei Liu, Yining Zheng, Xipeng Qiu, Xuanjing Huang

MINER: Improving Out-of-Vocabulary Named Entity Recognition from an Information Theoretic Perspective (2022.acl-long)

Challenge:	Existing models capture cross-sentence relations with recurrent neural networks, but they are hard to capture sentence-level long-distance dependency.
Approach:	They propose a graph-based neural network for extractive summarization which contains semantic nodes apart from sentences.
Outcome:	The proposed graph-based neural network is the first to incorporate different types of nodes into it and perform a qualitative analysis.

Copied to clipboard

Xiao Wang, Shihan Dou, Limao Xiong, Yicheng Zou, Qi Zhang, Tao Gui, Liang Qiao, Zhanzhan Cheng, Xuanjing Huang

Length Generalization of Causal Transformers without Position Encoding (2024.findings-acl)

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.

Copied to clipboard

Jie Wang, Tao Ji, Yuanbin Wu, Hang Yan, Tao Gui, Qi Zhang, Xuanjing Huang, Xiaoling Wang

ALaRM: Align Language Models via Hierarchical Rewards Modeling (2024.findings-acl)

Challenge:	Besides Transformers without position encodings, the success of NoPE provides a new way to overcome the challenge of generalizing to longer sentences.
Approach:	They propose a parameter-efficient tuning for searching attention heads’ best temperature hyper-parameters, which substantially expands NoPE’s context size.
Outcome:	The proposed tuning significantly expands NoPE's context size, allowing it to generalize to longer sentences with state-of-the-art generalization algorithms.

Copied to clipboard

Yuhang Lai, Siyuan Wang, Shujun Liu, Xuanjing Huang, Zhongyu Wei

A Lexicon-Based Graph Neural Network for Chinese NER (D19-1)

Challenge:	Current alignment approaches struggle with inconsistency and sparsity of human supervision signals.
Approach:	They propose a framework modeling hierarchical rewards in reinforcement learning from human feedback (RLHF) it integrates holistic rewards with aspect-specific rewards to enhance alignment of large language models with human preferences.
Outcome:	The proposed framework improves the alignment of large language models with human preferences by integrating holistic rewards with aspect-specific rewards.

Copied to clipboard

Tao Gui, Yicheng Zou, Qi Zhang, Minlong Peng, Jinlan Fu, Zhongyu Wei, Xuanjing Huang

Unleashing the Power of Language Models in Text-Attributed Graph (2023.findings-emnlp)

Challenge:	Chinese named entity recognition models are vulnerable to word ambiguities due to the lack of global semantics and chain structure.
Approach:	They propose a lexicon-based graph neural network with global semantics to solve word ambiguities in Chinese named entity recognition (NER) Lexicons are used to construct the graph and provide word-level features.
Outcome:	The proposed model improves on four NER datasets on Chinese characters, potential words, and the whole-sentence semantics.

Copied to clipboard

Haoyu Kuang, Jiarong Xu, Haozhe Zhang, Zuyu Zhao, Qi Zhang, Xuanjing Huang, Zhongyu Wei

Unveiling the Truth and Facilitating Change: Towards Agent-based Large-scale Social Movement Simulation (2024.findings-acl)

Challenge:	Existing studies on graph learning on text-attributed graphs have been limited by memory cost and underutilization of relationships between nodes and words.
Approach:	They propose a Node Representation Update Pre-training Architecture based on Co-modeling text and graph to learn representations of papers and words simultaneously.
Outcome:	The proposed model outperforms baselines on the ogbn-arxiv benchmark dataset.

Copied to clipboard

Xinyi Mou, Zhongyu Wei, Xuanjing Huang

Challenge:	Existing methods for simulating social movements encounter challenges in capturing behavior of participants.
Approach:	They propose a hybrid framework for social media user simulation wherein users are categorized into two types: core and ordinary users.
Outcome:	The proposed framework is able to simulate the behavior of social media users across real-world datasets and demonstrate its effectiveness and flexibility.

Exploring the Compositional Deficiency of Large Language Models in Mathematical Reasoning Through Trap Problems (2024.emnlp-main)

Copied to clipboard

Jun Zhao, Jingqi Tong, Yurong Mou, Ming Zhang, Qi Zhang, Xuanjing Huang

Improving Discriminative Capability of Reward Models in RLHF Using Contrastive Learning (2024.emnlp-main)

Challenge:	Current LLMs lack systematic compositionality, and therefore cannot serve as reliable cognitive models.
Approach:	They propose to introduce logical traps into the original problems of MATH and GSM8K to investigate the compositionality of large language models in mathematical reasoning.
Outcome:	The proposed model can generate infinite combinations from finite learned components.

Copied to clipboard

Lu Chen, Rui Zheng, Binghai Wang, Senjie Jin, Caishuang Huang, Junjie Ye, Zhihao Zhang, Yuhao Zhou, Zhiheng Xi, Tao Gui, Qi Zhang, Xuanjing Huang

Error Classification of Large Language Models on Math Word Problems: A Dynamically Adaptive Framework (2025.findings-emnlp)

Challenge:	Current methods rely on ranking losses to teach reward model to assess preferences, but they are susceptible to noise and ambiguous data, often failing to deeply understand human intentions.
Approach:	They propose a method that incorporates contrastive learning into the reward modeling process to enhance generalization and stabilize the reinforcement learning training process.
Outcome:	The proposed method enhances generalization of the reward model, stabilizes the reinforcement learning training process, and improves the final alignment with human preferences.

Copied to clipboard

Zhangyue Yin, YuHong Sun, Xuanjing Huang, Xipeng Qiu, Hui Zhao

Locate, Steer, and Improve: A Practical Survey of Actionable Mechanistic Interpretability in Large Language Models (2026.findings-acl)

Challenge:	Current error classification methods rely on static and predefined categories to capture error patterns.
Approach:	They propose a framework for automated dynamic error classification in mathematical reasoning that incorporates common error patterns as explicit guidance.
Outcome:	The proposed framework reduces human bias and fine-grained analysis of error patterns.

Copied to clipboard

Hengyuan Zhang, Zhihao Zhang, Ercong Nie, Mingyang Wang, Zunhai Su, Yiwei Wang, Qianli Wang, Shuzhou Yuan, Xufeng Duan, Qibo Xue, Zeping Yu, Chenming Shang, Xiao Liang, Jing Xiong, Hui Shen, Chaofan Tao, Zhengwu Liu, Senjie Jin, Zhiheng Xi, Dongdong Zhang, Sophia Ananiadou, Tao Gui, Ruobing Xie, Hayden Kwok-Hay So, Hinrich Schuetze, Xuanjing Huang, Qi Zhang, Ngai Wong

Explainability & Interpretability in NLP

Challenge:	Existing literature on mechanistic interpretation (MI) treats it as an observational science, leaving practical applications underexplored.
Approach:	They propose a survey structured around the pipeline to identify and improve MI models.
Outcome:	The proposed framework enables tangible improvements in Alignment, Capability, and Efficiency.

Towards Efficient NLP: A Standard Evaluation and A Strong Baseline (2022.naacl-main)

Copied to clipboard

Xiangyang Liu, Tianxiang Sun, Junliang He, Jiawen Wu, Lingling Wu, Xinyu Zhang, Hao Jiang, Zhao Cao, Xuanjing Huang, Xipeng Qiu

Governance in Motion: Co-evolution of Constitutions and AI models for Scalable Safety (2025.emnlp-main)

Challenge:	Rather than pursuing the reachless SOTA accuracy, researchers are focusing on model efficiency and usability.
Approach:	They propose an evaluation and a public leaderboard for efficient NLP models that depicts the Pareto Frontier for various language understanding tasks.
Outcome:	The proposed model outperforms or performs on par with SOTA compressed and early exiting models.

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

ProofInfer: Generating Proof via Iterative Hierarchical Inference (2022.emnlp-main)

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.

Copied to clipboard

Zichu Fei, Qi Zhang, Xin Zhou, Tao Gui, Xuanjing Huang

Explicit Memory Learning with Expectation Maximization (2024.emnlp-main)

Challenge:	Existing proof generation models focus on generating several proof paths instead of a whole tree.
Approach:	They propose a method that generates the proof tree via iterative hierarchical inference . they propose coding the proof as plain text without losing structure information .
Outcome:	The proposed proof generation model significantly improves performance on widely-used datasets.

Copied to clipboard

Zhangyue Yin, Qiushi Sun, Qipeng Guo, Zhiyuan Zeng, Qinyuan Cheng, Xipeng Qiu, Xuanjing Huang

Challenge:	Large Language Models lack reliable learning mechanisms for updating information across interactions.
Approach:	They propose a framework that enhances explicit memory updates via the Expectation-Maximization algorithm.
Outcome:	The proposed framework outperforms existing methods without memory or with static external memory on streaming inference tasks.

Inductive Relation Inference of Knowledge Graph Enhanced by Ontology Information (2023.findings-emnlp)

Copied to clipboard

Wentao Zhou, Jun Zhao, Tao Gui, Qi Zhang, Xuanjing Huang

Knowledge Representation

Challenge:	Existing methods to inference knowledge graphs lack ontology information, which is often too sparse.
Approach:	They propose a knowledge graph inductive inference method that fuses ontology information to learn the semantic information of entities.
Outcome:	The proposed method outperforms large language models like ChatGPT on two benchmark datasets and improves the MRR metrics by 15.4% and 44.1%, respectively.

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

Aligning Large Language Models with Human Preferences through Representation Engineering (2024.acl-long)

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 .

Copied to clipboard

Wenhao Liu, Xiaohua Wang, Muling Wu, Tianlong Li, Changze Lv, Zixuan Ling, Zhu JianHao, Cenyuan Zhang, Xiaoqing Zheng, Xuanjing Huang

Negative Sample is Negative in Its Own Way: Tailoring Negative Sentences for Image-Text Retrieval (2022.findings-naacl)

Challenge:	Existing methods for achieving this alignment involve employing reinforcement learning from human feedback (RLHF) Existing approaches involve using RLHF to fine-tune LLMs based on human labels . however, RLRF is susceptible to instability during fine- tuning and presents challenges in implementation.
Approach:	They propose to use reinforcement learning from human feedback to fine-tune large language models with human preferences to achieve precise control of model behavior.
Outcome:	Experiments show that RAHF can be used to capture and manipulate representations to align with a broad spectrum of human preferences or values rather than being confined to a single concept or function.

Copied to clipboard

Zhihao Fan, Zhongyu Wei, Zejun Li, Siyuan Wang, Xuanjing Huang, Jianqing Fan

A Confidence-based Partial Label Learning Model for Crowd-Annotated Named Entity Recognition (2023.findings-acl)

Challenge:	Existing approaches to retrieve hard negative sentences are limited in the scale of the dataset thus fail to identify negative samples of high difficulty for every image.
Approach:	They propose to use a model to generate synthetic negative sentences with higher difficulty by masking and refilling the images and performing word discrimination and word correction tasks to improve retrieval and generation.
Outcome:	The proposed model generates synthetic negative sentences with higher difficulty on MS-COCO and Flickr30K and is robust and faithful to state-of-the-art training.

Copied to clipboard

Limao Xiong, Jie Zhou, Qunxi Zhu, Xiao Wang, Yuanbin Wu, Qi Zhang, Tao Gui, Xuanjing Huang, Jin Ma, Ying Shan

A Multi-Format Transfer Learning Model for Event Argument Extraction via Variational Information Bottleneck (2022.coling-1)

Challenge:	Existing models for named entity recognition (NER) are based on large-scale labeled datasets, which always obtain using crowdsourcing.
Approach:	They propose a CONfidence-based partial Label Learning method to integrate prior and posterior confidences for crowd-annotated named entity recognition models.
Outcome:	The proposed model improves on real-world and synthetic datasets compared with baselines.

Copied to clipboard

Jie Zhou, Qi Zhang, Qin Chen, Qi Zhang, Liang He, Xuanjing Huang

Kernel-Whitening: Overcome Dataset Bias with Isotropic Sentence Embedding (2022.emnlp-main)

Challenge:	Event argument extraction (EAE) aims to extract arguments with given roles from texts.
Approach:	They propose a multi-format transfer learning model with variational information bottleneck to learn from existing datasets.
Outcome:	The proposed model improves on three benchmark datasets and obtains state-of-the-art performance on EAE.

Copied to clipboard

SongYang Gao, Shihan Dou, Qi Zhang, Xuanjing Huang

AgentV-RL: Scaling Reward Modeling with Agentic Verifier (2026.findings-acl)

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.

Copied to clipboard

Jiazheng Zhang, Ziche Fu, Zhiheng Xi, Wenqing Jing, Mingxu Chai, Wei He, Guoqiang Zhang, Chenghao Fan, Chenxin An, Wenxiang Chen, Zhicheng Liu, Haojie Pan, Dingwei Zhu, Tao Gui, Qi Zhang, Xuanjing Huang

LFKQG: A Controlled Generation Framework with Local Fine-tuning for Question Generation over Knowledge Bases (2022.coling-1)

Challenge:	Existing approaches to improve LLM reasoning are limited in complex domains and lack external grounding makes verifiers unreliable on computation-intensive tasks.
Approach:	They propose a framework that transforms reward modeling into a multi-turn, tool-augmented deliberative process.
Outcome:	The proposed framework surpasses state-of-the-art ORMs by 25.2% under parallel and sequential TTS.

Copied to clipboard

Zichu Fei, Xin Zhou, Tao Gui, Qi Zhang, Xuanjing Huang

Unveiling and Consulting Core Experts in Retrieval-Augmented MoE-based LLMs (2024.emnlp-main)

Challenge:	Existing KBQG models focus on the most relevant part of the answer entity, while neglecting the rest of the subgraph.
Approach:	They propose a controlled generation framework for Question Generation over Knowledge Bases that generates questions with out-of-vocabulary (OOV) predicates.
Outcome:	The proposed framework outperforms existing methods significantly on three widely-used benchmark datasets SimpleQuestion, PathQuestions, and WebQuestIONS.

Copied to clipboard

Xin Zhou, Ping Nie, Yiwen Guo, Haojie Wei, Zhanqiu Zhang, Pasquale Minervini, Ruotian Ma, Tao Gui, Qi Zhang, Xuanjing Huang

Orthogonal Subspace Learning for Language Model Continual Learning (2023.findings-emnlp)

Challenge:	Existing research seeks to enhance RAG performance by retrieving higher-quality documents or designing RAG-specific LLMs, but internal mechanisms that contribute to RAG’s effectiveness remain underexplored.
Approach:	They propose to examine the internal mechanisms within the popular Mixture-of-Expert (MoE)-based LLMs and examine their ability to improve RAG by examining expert activations.
Outcome:	The proposed method significantly improved the ability of Large Language Models (LLMs) to solve knowledge-intensive tasks.

Copied to clipboard

Xiao Wang, Tianze Chen, Qiming Ge, Han Xia, Rong Bao, Rui Zheng, Qi Zhang, Tao Gui, Xuanjing Huang

PFDial: A Structured Dialogue Instruction Fine-tuning Method Based on UML Flowcharts (2025.findings-acl)

Challenge:	Existing methods for continual learning in language models suffer catastrophic forgetting when learning sequential tasks.
Approach:	They propose an orthogonal low-rank adaptation approach for continual learning in language models that uses orthogons to learn sequentially.
Outcome:	The proposed approach outperforms state-of-the-art methods on continual learning benchmarks and preserves generalization ability of LLMs on unseen tasks.

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

Flooding-X: Improving BERT’s Resistance to Adversarial Attacks via Loss-Restricted Fine-Tuning (2022.acl-long)

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.

Copied to clipboard

Qin Liu, Rui Zheng, Bao Rong, Jingyi Liu, ZhiHua Liu, Zhanzhan Cheng, Liang Qiao, Tao Gui, Qi Zhang, Xuanjing Huang

Domain Generalization via Causal Adjustment for Cross-Domain Sentiment Analysis (2024.lrec-main)

Challenge:	Existing approaches to generating adversarial perturbations scale up the cost of training computational complexity by the number of gradient steps it takes to obtain the adversarials.
Approach:	They propose a flood method which aims at better generalization and a criterion to bring hyper-parameter-dependent flooding into effect with a narrowed-down search space by measuring how the gradient steps taken within one epoch affect the loss of each batch.
Outcome:	The proposed method improves BERT’s resistance to textual adversarial attacks by a large margin and achieves state-of-the-art robust accuracy on various text classification and GLUE tasks.

Copied to clipboard

Siyin Wang, Jie Zhou, Qin Chen, Qi Zhang, Tao Gui, Xuanjing Huang

Sentiment Analysis

Challenge:	Existing approaches to domain adaptation fail to generalize well on unknown test data.
Approach:	They propose a backdoor adjustment-based causal model to disentangle domain-specific and domain-invariant representations that play essential roles in tackling domain shift.
Outcome:	The proposed model disentangles domain-specific and domain-invariant representations that play essential roles in tackling domain shift.

VoCoT: Unleashing Visually Grounded Multi-Step Reasoning in Large Multi-Modal Models (2025.naacl-long)

Copied to clipboard

Zejun Li, Ruipu Luo, Jiwen Zhang, Minghui Qiu, Xuanjing Huang, Zhongyu Wei

Parameter Efficient Multi-task Fine-tuning by Learning to Transfer Token-wise Prompts (2023.findings-emnlp)

Challenge:	Despite the impressive capabilities of large multi-modal models, their effectiveness in handling complex tasks has been limited by the prevailing singlestep reasoning paradigm.
Approach:	They propose a visuallygrounded object-centric Chain-of-Thought reasoning framework for LMMs that is based on a multi-modal interleaved and aligned representation of object concepts.
Outcome:	The proposed model outperforms SOTA models in CLEVR and EmbSpatial benchmarks.

Copied to clipboard

Muling Wu, Wenhao Liu, Jianhan Xu, Changze Lv, Zixuan Ling, Tianlong Li, Longtao Huang, Xiaoqing Zheng, Xuanjing Huang

Counteracting the Matthew Effect in Self-Improvement of LVLMs through Head-Tail Re-balancing (2026.acl-long)

Challenge:	Prompt tuning has been proven to be successful on various tasks by incorporating a small number of trainable parameters while freezing large pre-trained language models.
Approach:	They propose a token-wise prompt tuning method that uses a bank of finer-grained soft prompt tokens to generate an instance-dependent prompt.
Outcome:	The proposed method performs far better than full parameter fine-tuned models and achieves state-of-the-art by tuning only 0.035% parameters on 14 datasets.

Copied to clipboard

Xin Guo, Zhiheng Xi, Yiwen Ding, Yitao Zhai, Xiaowei Shi, Xunliang Cai, Tao Gui, Qi Zhang, Xuanjing Huang

Query Structure Modeling for Inductive Logical Reasoning Over Knowledge Graphs (2023.acl-long)

Challenge:	Large vision language models have impressive reasoning capabilities across complex multimodal tasks.
Approach:	They propose to use distribution-reshaping and trajectory-rebalancing to improve visual reasoning capabilities.
Outcome:	Experiments on Qwen2-VL-7B-Instruct and InternVL2.5-4B models show that their methods outperform baselines by 3.86 points.

Copied to clipboard

Siyuan Wang, Zhongyu Wei, Meng Han, Zhihao Fan, Haijun Shan, Qi Zhang, Xuanjing Huang

Mitigating Tail Narrowing in LLM Self-Improvement via Socratic-Guided Sampling (2025.naacl-long)

Challenge:	Existing methods for inductive reasoning over knowledge graphs lack the ability to model the logical structures of complex queries.
Approach:	They propose a structure-modeled textual encoding framework for inductive logical reasoning over KGs that encodes linearized query structures and entities using pre-trained language models to find answers.
Outcome:	The proposed framework encodes query structures and entities using pre-trained language models to find answers.

Copied to clipboard

Yiwen Ding, Zhiheng Xi, Wei He, Lizhuoyuan Lizhuoyuan, Yitao Zhai, Shi Xiaowei, Xunliang Cai, Tao Gui, Qi Zhang, Xuanjing Huang

OctoBench: Benchmarking Scaffold-Aware Instruction Following in Repository-Grounded Agentic Coding (2026.acl-long)

Challenge:	Large language models (LLMs) generate solutions themselves and iteratively train on filtered, high-quality rationales, but performance reaches a ceiling after a few iterations.
Approach:	They propose a strategy to improve the efficiency of sampling heavy-tailed data by using Socratic-style guidance signals to help LLMs reasoning with complex queries.
Outcome:	The proposed approach is effective on difficult queries and on held-out tasks, while requiring human supervision.

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

Programming Languages in NLP

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.

AgentSense: Benchmarking Social Intelligence of Language Agents through Interactive Scenarios (2025.naacl-long)

Copied to clipboard

Xinyi Mou, Jingcong Liang, Jiayu Lin, Xinnong Zhang, Xiawei Liu, Shiyue Yang, Rong Ye, Lei Chen, Haoyu Kuang, Xuanjing Huang, Zhongyu Wei

Linguistics & Cognitive NLP

Challenge:	Large language models are increasingly employed to empower autonomous agents to simulate human behavior.
Approach:	They propose to evaluate LLM-driven agents through multi-turn interactions using a bottom-up approach to create diverse social scenarios constructed from extensive scripts.
Outcome:	The proposed model evaluates LLM-driven agents through multi-turn interactions emphasizing goal completion and implicit reasoning.

Self-Polish: Enhance Reasoning in Large Language Models via Problem Refinement (2023.findings-emnlp)

Copied to clipboard

Zhiheng Xi, Senjie Jin, Yuhao Zhou, Rui Zheng, Songyang Gao, Jia Liu, Tao Gui, Qi Zhang, Xuanjing Huang

AgentGym: Evaluating and Training Large Language Model-based Agents across Diverse Environments (2025.acl-long)

Challenge:	Existing prompting methods have been used to enhance multistep reasoning capabilities of large language models, but they have overlooked the potential of formulating higher-quality problems.
Approach:	They propose a method that starts from the problem side and refines problems to be more comprehensible and solvable for models.
Outcome:	The proposed method achieves notable and consistent effectiveness on five reasoning benchmarks across different models.

Copied to clipboard

Zhiheng Xi, Yiwen Ding, Wenxiang Chen, Boyang Hong, Honglin Guo, Junzhe Wang, Xin Guo, Dingwen Yang, Chenyang Liao, Wei He, Songyang Gao, Lu Chen, Rui Zheng, Yicheng Zou, Tao Gui, Qi Zhang, Xipeng Qiu, Xuanjing Huang, Zuxuan Wu, Yu-Gang Jiang

Do Large Language Models Know What They Don’t Know? (2023.findings-acl)

Challenge:	Large language models (LLMs) are promising foundations to build generally-capable agents . however, the community lacks a unified interactive framework that covers diverse environments for comprehensive evaluation of agents.
Approach:	They propose a framework that features 7 real-world scenarios, 14 environments, and 89 tasks for unified, real-time, and concurrent agent interaction.
Outcome:	The proposed framework features 7 real-world scenarios, 14 environments, and 89 tasks for unified, real-time, and concurrent agent interaction.

Copied to clipboard

Zhangyue Yin, Qiushi Sun, Qipeng Guo, Jiawen Wu, Xipeng Qiu, Xuanjing Huang

Multi-Programming Language Sandbox for LLMs (2025.acl-demo)

Challenge:	Large language models (LLMs) have vast knowledge that allows them to excel in various NLP tasks.
Approach:	They propose an automated method to detect uncertainty in the responses of large language models and a dataset to measure their self-knowledge.
Outcome:	The proposed method detects uncertainty in the responses of large language models and provides a novel measure of their self-knowledge.

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

Modeling Evolution of Message Interaction for Rumor Resolution (2020.coling-main)

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).

Copied to clipboard

Lei Chen, Zhongyu Wei, Jing Li, Baohua Zhou, Qi Zhang, Xuanjing Huang

Challenge:	Existing methods for rumor resolution ignore local interactions during the message diffusion which is important for the identification of rumors.
Approach:	They propose to model confrontation and reciprocity between message pairs via discrete variational autoencoders which effectively reflects the diversified opinion interactivity.
Outcome:	Experiments on a PHEME dataset show that the proposed model achieves higher accuracy than existing methods.

Align Voting Behavior with Public Statements for Legislator Representation Learning (2021.acl-long)

Copied to clipboard

Xinyi Mou, Zhongyu Wei, Lei Chen, Shangyi Ning, Yancheng He, Changjian Jiang, Xuanjing Huang

Asynchronous Deep Interaction Network for Natural Language Inference (D19-1)

Challenge:	Existing studies rely on roll call data to estimate political preference of legislators.
Approach:	They propose to integrate voting behavior and public statements on Twitter to jointly model legislators.
Outcome:	The proposed model improves on the task of roll call vote prediction . it also shows that the model captures nuances in statements .

Copied to clipboard

Di Liang, Fubao Zhang, Qi Zhang, Xuanjing Huang

PathQG: Neural Question Generation from Facts (2020.emnlp-main)

Challenge:	Existing methods have framed the reasoning problem as a semantic matching task.
Approach:	They propose an asynchronous deep interaction network (ADIN) to deconstruct the reasoning process and implement asynchron and multi-step reasoning.
Outcome:	The proposed model outperforms strong baselines on three popular benchmarks: SNLI, MultiNLI, and SciTail.

Copied to clipboard

Siyuan Wang, Zhongyu Wei, Zhihao Fan, Zengfeng Huang, Weijian Sun, Qi Zhang, Xuanjing Huang

Tell Me What You Don’t Know: Enhancing Refusal Capabilities of Role-Playing Agents via Representation Space Analysis and Editing (2025.findings-acl)

Challenge:	Existing research for question generation encodes text as a sequence of tokens without explicitly modeling fact information.
Approach:	They propose to incorporate facts in the input text for question generation in a comprehensive way.
Outcome:	The proposed model outperforms state-of-the-art models and human evaluation shows it generates relevant and informative questions.

Copied to clipboard

Wenhao Liu, Siyu An, Junru Lu, Muling Wu, Tianlong Li, Xiaohua Wang, Changze Lv, Xiaoqing Zheng, Di Yin, Xing Sun, Xuanjing Huang

Enhancing Contrastive Learning with Noise-Guided Attack: Towards Continual Relation Extraction in the Wild (2024.acl-long)

Challenge:	Role-playing Agents (RPAs) struggle to recognize and respond to hard queries that conflict with their role-play knowledge.
Approach:	They propose a lightweight representation editing approach that conveniently shifts conflicting requests to the rejection region, thereby enhancing the model’s refusal accuracy.
Outcome:	The proposed model improves RPAs’ refusal ability of conflicting requests while maintaining their general role-playing capabilities.

Copied to clipboard

Ting Wu, Jingyi Liu, Rui Zheng, Tao Gui, Qi Zhang, Xuanjing Huang

Searching for Best Practices in Retrieval-Augmented Generation (2024.emnlp-main)

Challenge:	Existing methods for continual relation extraction (CRE) excel in preserving old knowledge but falter when confronted with contaminated data streams.
Approach:	They propose a noise-resistant contrastive framework for continual relation extraction (CRE) that preserves old knowledge while learning incremental corrupted relations.
Outcome:	The proposed framework outperforms state-of-the-art methods on various benchmarks with increasing noise rates.

Copied to clipboard

Xiaohua Wang, Zhenghua Wang, Xuan Gao, Feiran Zhang, Yixin Wu, Zhibo Xu, Tianyuan Shi, Zhengyuan Wang, Shizheng Li, Qi Qian, Ruicheng Yin, Changze Lv, Xiaoqing Zheng, Xuanjing Huang

Multi-Objective Forward Reasoning and Multi-Reward Backward Refinement for Product Review Summarization (2024.lrec-main)

Challenge:	Retrieval-augmented generation (RAG) techniques have proven to be effective in integrating up-to-date information, mitigating hallucinations, and enhancing response quality, especially in specialized domains.
Approach:	They propose several strategies for deploying RAG that balance performance and efficiency.
Outcome:	The proposed approaches can significantly enhance question-answering capabilities and accelerate the generation of multimodal content using a “retrieval as generation” strategy.

Copied to clipboard

Libo Sun, Siyuan Wang, Meng Han, Ruofei Lai, Xinyu Zhang, Xuanjing Huang, Zhongyu Wei

Evaluating and Enhancing the Robustness of Neural Network-based Dependency Parsing Models with Adversarial Examples (2020.acl-main)

Challenge:	Product review summarization aims to generate a concise summary based on product reviews . factual accuracy, aspect comprehensiveness, and content relevance are challenges .
Approach:	They propose an FB-Thinker framework to improve product review summarization ability . they propose two Chinese product review summary datasets for instruction-tuning and evaluation .
Outcome:	The proposed framework improves product review summarization with forward reasoning and backward refinement.

Copied to clipboard

Xiaoqing Zheng, Jiehang Zeng, Yi Zhou, Cho-Jui Hsieh, Minhao Cheng, Xuanjing Huang

Outcome Accuracy is Not Enough: Aligning the Reasoning Process of Reward Models (2026.acl-long)

Challenge:	Previously studies focused on semantic tasks such as sentiment analysis, question answering and reading comprehension.
Approach:	They propose two approaches to study where and how adversarial examples exist in dependency parsing . they use a state-of-the-art parser to find adversarials in existing texts .
Outcome:	The proposed approaches show that adversarial examples exist in dependency parsing . they show that up to 77% of input examples admit adversarials .

Copied to clipboard

Binghai Wang, Yantao Liu, Yuxuan Liu, Tianyi Tang, Shenzhi Wang, Chang Gao, Chujie Zheng, Yichang Zhang, Le Yu, Shixuan Liu, Tao Gui, Qi Zhang, Xuanjing Huang, Bowen Yu, Fei Huang, Junyang Lin

SFMSS: Service Flow aware Medical Scenario Simulation for Conversational Data Generation (2025.findings-naacl)

Challenge:	Recent studies observe a phenomenon where reward models achieve high accuracy on static datasets but fail to generalize effectively during RLHF.
Approach:	They propose a method that combines rationale consistency with outcome accuracy to improve performance on RM-Bench and JudgeBench.
Outcome:	The proposed method surpasses baselines on RM-Bench and JudgeBench by an average of 5% and improves creative writing tasks by 7%.

Copied to clipboard

Zhijie Bao, Qingyun Liu, Xuanjing Huang, Zhongyu Wei

Natural Language Interfaces

Challenge:	Medical-specific Large Language Models (LLMs) have demonstrated impressive performance on medical-related exams and tasks.
Approach:	They propose a framework for medical conversational data generation that uses Authentic Seed Data to ensure quality of the data.
Outcome:	The proposed model outperforms all baselines and human evaluations, and aligns with human preferences and clinical demands.

A Concise Model for Multi-Criteria Chinese Word Segmentation with Transformer Encoder (2020.findings-emnlp)

Copied to clipboard

Xipeng Qiu, Hengzhi Pei, Hang Yan, Xuanjing Huang

Benchmark Self-Evolving: A Multi-Agent Framework for Dynamic LLM Evaluation (2025.coling-main)

Challenge:	Existing work on multicriteria Chinese word segmentation focuses on combining multiple heterogeneous segmentation criteria into a single task.
Approach:	They propose a unified model for multi-criteria Chinese word segmentation which is fully-shared for all criteria.
Outcome:	The proposed model outperforms existing models on eight datasets with different criteria.

Copied to clipboard

Siyuan Wang, Zhuohan Long, Zhihao Fan, Xuanjing Huang, Zhongyu Wei

Inference-Time Decontamination: Reusing Leaked Benchmarks for Large Language Model Evaluation (2024.findings-emnlp)

Challenge:	Recent advances in Large Language Models have demonstrated remarkable performance across tasks.
Approach:	They propose a benchmark self-evolving framework to dynamically evaluate rapidly advancing Large Language Models.
Outcome:	The proposed framework extends existing benchmarks to extend models across tasks and tasks.

Copied to clipboard

Qin Zhu, Qinyuan Cheng, Runyu Peng, Xiaonan Li, Ru Peng, Tengxiao Liu, Xipeng Qiu, Xuanjing Huang

BBTv2: Towards a Gradient-Free Future with Large Language Models (2022.emnlp-main)

Challenge:	Current LLMs are achieving better performance on various benchmarks, but their performance in practical applications does not always match their benchmark results.
Approach:	They propose to detect and rewrite leaked benchmarks without altering their difficulties by using Inference-Time Decontamination (ITD) to mitigate performance inflation caused by memorizing leaked samples.
Outcome:	The proposed method reduces inflated accuracy by 22.9% on GSM8K and 19.0% on MMLU.

Copied to clipboard

Tianxiang Sun, Zhengfu He, Hong Qian, Yunhua Zhou, Xuanjing Huang, Xipeng Qiu

MulDimIF: A Multi-Dimensional Constraint Framework for Evaluating and Improving Instruction Following in Large Language Models (2026.findings-acl)

Challenge:	Recent work on parameter-efficient tuning (PET) only tunes a small portion of parameters while keeping most of the parameters of the LLM unchanged.
Approach:	They propose an improved version of Black-Box Tuning to tune PTMs through gradient descent . they prepend continuous prompts to every layer of the PTM and propose a divide-and-conquer gradient-free algorithm to optimize the prompts alternately.
Outcome:	The proposed method achieves comparable performance to full model tuning and state-of-the-art parameter-efficient methods under few-shot settings while maintaining much fewer tunable parameters.

Copied to clipboard

Junjie Ye, Caishuang Huang, Zhuohan Chen, Wenjie Fu, Chenyuan Yang, Leyi Yang, Yilong Wu, Peng Wang, Meng Zhou, Xiaolong Yang, Tao Gui, Qi Zhang, Zhongchao Shi, Jianping Fan, Xuanjing Huang

Cross-Lingual Dependency Parsing by POS-Guided Word Reordering (2020.findings-emnlp)

Challenge:	Existing research has focused on constraint categories, offering little guidance for improving instruction following abilities.
Approach:	They propose a multi-dimensional constraint framework that allows for instruction following . they construct 9,106 code-verifiable samples and evaluate 18 LLMs .
Outcome:	The proposed framework improves instruction following performance without compromising general performance.

Copied to clipboard

Lu Liu, Yi Zhou, Jianhan Xu, Xiaoqing Zheng, Kai-Wei Chang, Xuanjing Huang

LongHeads: Multi-Head Attention is Secretly a Long Context Processor (2024.findings-emnlp)

Challenge:	Existing approaches to cross-lingual dependency parsing rely on large corpus size and cost.
Approach:	They propose a cross-lingual dependency parsing approach based on word reordering . they propose to train a model that transfers knowledge learned in one or multiple languages to target languages .
Outcome:	The proposed approach outperforms the baseline approach in Hindi and Latin by 15.3% and 6.7%.

Copied to clipboard

Yi Lu, Xin Zhou, Wei He, Jun Zhao, Tao Ji, Tao Gui, Qi Zhang, Xuanjing Huang

Green & Sustainable NLP

Challenge:	Large language models struggle to process lengthy inputs due to limited length generalization and attention’s quadratic computational demands.
Approach:	They propose a training-free framework that allows each head to attend to important context chunks instead of allowing each head a full sentence .
Outcome:	The proposed framework unlocks multi-head attention's untapped potential by allowing each head to attend to important context chunks instead of the full sentence.

Making Parameter-efficient Tuning More Efficient: A Unified Framework for Classification Tasks (2022.coling-1)

Copied to clipboard

Xin Zhou, Ruotian Ma, Yicheng Zou, Xuanting Chen, Tao Gui, Qi Zhang, Xuanjing Huang, Rui Xie, Wei Wu

LONGAGENT: Achieving Question Answering for 128k-Token-Long Documents through Multi-Agent Collaboration (2024.emnlp-main)

Challenge:	Large pre-trained language models (PLMs) have demonstrated superior performance in industrial applications.
Approach:	They propose a framework that re-uses existing parameter-efficient methods with a unified classifier.
Outcome:	The proposed framework improves the efficiency of existing parameter-efficient methods with a unified classifier.

Copied to clipboard

Jun Zhao, Can Zu, Xu Hao, Yi Lu, Wei He, Yiwen Ding, Tao Gui, Qi Zhang, Xuanjing Huang

Question Answering

Challenge:	Large language models (LLMs) have been successful in understanding language and processing text, but their cost prohibits their practical applications.
Approach:	They propose a multi-agent collaboration method that breaks down lengthy documents into smaller, more manageable chunks and organizes the member agents to read their assigned chunks.
Outcome:	The proposed method achieves 16.42% and 1.63% accuracy gains over existing models on single-hop and multi-hop QA settings.

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

Beyond Boundaries: Learning a Universal Entity Taxonomy across Datasets and Languages for Open Named Entity Recognition (2025.coling-main)

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 .

Copied to clipboard

Yuming Yang, Wantong Zhao, Caishuang Huang, Junjie Ye, Xiao Wang, Huiyuan Zheng, Yang Nan, Yuran Wang, Xueying Xu, Kaixin Huang, Yunke Zhang, Tao Gui, Qi Zhang, Xuanjing Huang

Multi-Agent Simulator Drives Language Models for Legal Intensive Interaction (2025.findings-naacl)

Challenge:	Current Large Language Models struggle with complex entity taxonomies in open domains and lack NER capabilities.
Approach:	They propose a dataset to guide LLMs' generalization in Open NER under a universal entity taxonomy.
Outcome:	The proposed model outperforms GPT-4 in 3 out-of-domain benchmarks across 15 datasets and 6 languages.

Copied to clipboard

ShengbinYue ShengbinYue, Ting Huang, Zheng Jia, Siyuan Wang, Shujun Liu, Yun Song, Xuanjing Huang, Zhongyu Wei

Measuring Data Diversity for Instruction Tuning: A Systematic Analysis and A Reliable Metric (2025.acl-long)

Challenge:	Large Language Models (LLMs) have advanced legal intelligence, but the scarcity of scenario data impedes the progress toward interactive legal scenarios.
Approach:	They propose a Multi-agent Legal Simulation Driver to generate synthetic data by simulating interactive legal scenarios.
Outcome:	The proposed framework ensures consistency of legal attributes between participants and introduces a supervisory mechanism to align participants’ characters and behaviors as well as addressing distractions.

Copied to clipboard

Yuming Yang, Yang Nan, Junjie Ye, Shihan Dou, Xiao Wang, Shuo Li, Huijie Lv, Tao Gui, Qi Zhang, Xuanjing Huang

CDEvalSumm: An Empirical Study of Cross-Dataset Evaluation for Neural Summarization Systems (2020.findings-emnlp)

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.

Copied to clipboard

Yiran Chen, Pengfei Liu, Ming Zhong, Zi-Yi Dou, Danqing Wang, Xipeng Qiu, Xuanjing Huang

VLSBench: Unveiling Visual Leakage in Multimodal Safety (2025.acl-long)

Challenge:	Existing evaluation methods for text summarization systems are limited to in-domain setting, where supervised pre-trained models are evaluated on the same dataset.
Approach:	They propose to use a cross-dataset evaluation approach to evaluate different summarization systems in a multi-domain setting.
Outcome:	The proposed model can be used to evaluate text summarization systems on different datasets.

Copied to clipboard

Xuhao Hu, Dongrui Liu, Hao Li, Xuanjing Huang, Jing Shao

CQG: A Simple and Effective Controlled Generation Framework for Multi-hop Question Generation (2022.acl-long)

Challenge:	Existing studies show that textual unlearning does not achieve comparable safety performance with image-text alignment.
Approach:	They propose to use textual unlearning to align MLLMs with image-text pairs to explain this problem . they construct a visual leakless safety bench with 2.2k image- text pairs to test this problem.
Outcome:	The proposed model can refuse image-text pairs according to textual queries, leading to unreliable safety evaluations.

Copied to clipboard

Zichu Fei, Qi Zhang, Tao Gui, Di Liang, Sirui Wang, Wei Wu, Xuanjing Huang

Mask Attention Networks: Rethinking and Strengthen Transformer (2021.naacl-main)

Challenge:	Current models can not ensure the complexity of generated questions, so they may generate shallow questions that can be answered without multi-hop reasoning.
Approach:	They propose a controlled framework to generate multi-hop questions that contain key entities in multi- hop reasoning chains and a novel Transformer-based decoder to guarantee that key entities appear in the questions.
Outcome:	The proposed model outperforms the state-of-the-art model 25% on HotpotQA.

Copied to clipboard

Zhihao Fan, Yeyun Gong, Dayiheng Liu, Zhongyu Wei, Siyuan Wang, Jian Jiao, Nan Duan, Ruofei Zhang, Xuanjing Huang

PDF-to-Tree: Parsing PDF Text Blocks into a Tree (2024.findings-emnlp)

Challenge:	Existing research explores to enhance the two sublayers separately to improve the capability of Transformer for text representation.
Approach:	They propose to combine SAN and Feed-Forward Networks to create a dynamic mask attention network with a learnable mask matrix which can model localness adaptively.
Outcome:	The proposed model outperforms the original Transformer on translation and text summarization tasks.

Copied to clipboard

Yue Zhang, Zhihao Zhang, Wenbin Lai, Chong Zhang, Tao Gui, Qi Zhang, Xuanjing Huang

Challenge:	Existing studies try to extract one universal reading order for PDF files, however, some applications, like Retrieval Augmented Generation, require breaking long articles into sections and subsections for better indexing.
Approach:	They propose a new task and dataset, PDF-to-Tree, which organizes the text blocks of a PDF into a tree structure.
Outcome:	The proposed parser achieves 93.93% accuracy, surpassing baseline methods by 6.72%.

EcoLANG: Efficient and Effective Agent Communication Language Induction for Social Simulation (2025.findings-emnlp)

Copied to clipboard

Xinyi Mou, Chen Qian, Wei Liu, Ling Yan, Yao Hu, Xuanjing Huang, Zhongyu Wei

Green & Sustainable NLP

Challenge:	Large language models (LLMs) have demonstrated impressive ability to role-play humans and replicate complex social dynamics.
Approach:	They propose an efficient agent communication language induction for social simulations that reduces token consumption by over 20%.
Outcome:	The proposed model reduces token consumption by over 20% while preserving human language.

Reasoning in Flux: Enhancing Large Language Models Reasoning through Uncertainty-aware Adaptive Guidance (2024.acl-long)

Copied to clipboard

Zhangyue Yin, Qiushi Sun, Qipeng Guo, Zhiyuan Zeng, Xiaonan Li, Junqi Dai, Qinyuan Cheng, Xuanjing Huang, Xipeng Qiu

Transferring from Formal Newswire Domain with Hypernet for Twitter POS Tagging (D18-1)

Challenge:	Extensive experiments across various reasoning tasks demonstrate that UAG not only enhances the reasoning abilities of LLMs but consistently outperforms several strong baselines with minimal computational overhead.
Approach:	They propose an approach to guide LLMs onto an accurate and reliable trajectory by identifying and adjusting uncertainty signals within each step of the reasoning chain.
Outcome:	The proposed approach outperforms strong baselines and outperformed strong models with minimal computational overhead.

Copied to clipboard

Tao Gui, Qi Zhang, Jingjing Gong, Minlong Peng, Di Liang, Keyu Ding, Xuanjing Huang

Safe Inputs but Unsafe Output: Benchmarking Cross-modality Safety Alignment of Large Vision-Language Models (2025.findings-naacl)

Challenge:	Existing POS tagging methods for Twitter use labeled newswire text . however, Twitter users tend to mimic formal media expressions and develop linguistically informal styles.
Approach:	They propose to use newswire text to learn POS tagging for Twitter while twitter users are developing linguistically informal styles.
Outcome:	The proposed method achieves better performance than state-of-the-art methods on three different datasets.

Copied to clipboard

Siyin Wang, Xingsong Ye, Qinyuan Cheng, Junwen Duan, Shimin Li, Jinlan Fu, Xipeng Qiu, Xuanjing Huang

Exploration and Exploitation: Two Ways to Improve Chinese Spelling Correction Models (2021.acl-short)

Challenge:	Recent studies focus on single-modality threats, but this approach fails to address cross-modal safety alignment.
Approach:	They propose a safety alignment challenge to evaluate cross-modality safety alignment . they propose 'Safe Inputs but Unsafe Output' to consider safety of single modalities .
Outcome:	The proposed safety alignment challenge examines cases where modalities are safe independently but could lead to unsafe outputs when combined.

Copied to clipboard

Chong Li, Cenyuan Zhang, Xiaoqing Zheng, Xuanjing Huang

Modeling the Q-Diversity in a Min-max Play Game for Robust Optimization (2023.findings-acl)

Challenge:	Experimental results show that a sequence-to-sequence learning framework with neural networks can be effective for Chinese Spelling Correction (CSC)
Approach:	They propose a sequence-to-sequence learning framework with neural networks that generates more valuable training instances and adds task-specific examples to enhance the model.
Outcome:	The proposed method improves generalization and robustness of multiple CSC models across three datasets.

Copied to clipboard

Ting Wu, Rui Zheng, Tao Gui, Qi Zhang, Xuanjing Huang

Unveiling Linguistic Regions in Large Language Models (2024.acl-long)

Challenge:	Existing methods for minimizing the worst-case loss of annotated groups are lacking in practice due to expensive annotations and privacy issues.
Approach:	They propose a distributionally robust optimization framework that relaxes group identification into direct parameterization by using an interactive training mode.
Outcome:	The proposed method outperforms state-of-the-art methods on synthetic and real-world text classification tasks.

Copied to clipboard

Zhihao Zhang, Jun Zhao, Qi Zhang, Tao Gui, Xuanjing Huang

AI-Press: A Multi-Agent News Generating and Feedback Simulation System Powered by Large Language Models (2025.coling-demos)

Challenge:	Existing studies on how LLMs achieve cross-lingual alignment and generalization have not explored the intrinsic mechanisms of how they achieve crosslingual alignment.
Approach:	They propose to remove a core region that corresponds to linguistic competence and set parameters to zero to reduce performance across 30 different languages.
Outcome:	The proposed model can be used to perform tasks requiring abstract knowledge and reasoning in complex languages.

Copied to clipboard

Xiawei Liu, Shiyue Yang, Xinnong Zhang, Haoyu Kuang, Libo Sun, Yihang Yang, Siming Chen, Xuanjing Huang, Zhongyu Wei

A Structure-Aware Argument Encoder for Literature Discourse Analysis (2022.coling-1)

Challenge:	We introduce AI-Press, an automated news drafting and polishing system based on multi-agent collaboration and Retrieval-Augmented Generation.
Approach:	They introduce AI-Press, an automated news drafting and polishing system based on multi-agent collaboration and Retrieval-Augmented Generation.
Outcome:	The proposed system generates public responses considering demographic distributions.

Copied to clipboard

Yinzi Li, Wei Chen, Zhongyu Wei, Yujun Huang, Chujun Wang, Siyuan Wang, Qi Zhang, Xuanjing Huang, Libo Wu

LLMs-as-Instructors: Learning from Errors Toward Automating Model Improvement (2024.findings-emnlp)

Challenge:	Existing research for argument representation learning treats tokens in sentences equally and ignores the implied structure information of argumentative context.
Approach:	They propose to separate tokens into two groups to capture structural information of arguments and to incorporate paragraph-level position information into the model.
Outcome:	The proposed model captures structural information of arguments and is able to identify arguments automatically.

Copied to clipboard

Jiahao Ying, Mingbao Lin, Yixin Cao, Wei Tang, Bo Wang, Qianru Sun, Xuanjing Huang, Shuicheng Yan

Challenge:	Using advanced Large Language Models, instructors can improve training of smaller models by analyzing their own model's errors.
Approach:	They propose a framework that leverages advanced Large Language Models to enhance training of smaller target models.
Outcome:	The proposed framework outperforms ChatGPT on multiple benchmarks and shows that it improves on both in-domain and out-of-domain benchmarks.

Exchange-of-Thought: Enhancing Large Language Model Capabilities through Cross-Model Communication (2023.emnlp-main)

Copied to clipboard

Zhangyue Yin, Qiushi Sun, Cheng Chang, Qipeng Guo, Junqi Dai, Xuanjing Huang, Xipeng Qiu

RethinkingTMSC: An Empirical Study for Target-Oriented Multimodal Sentiment Classification (2023.findings-emnlp)

Challenge:	Large Language Models (LLMs) have made significant strides in complex reasoning tasks, but their reasoning is often constrained by their intrinsic understanding, lacking external insights.
Approach:	They propose a framework that enables cross-model communication during problem-solving.
Outcome:	The proposed framework surpasses established baselines in complex reasoning tasks and is cost-effective.

Copied to clipboard

Junjie Ye, Jie Zhou, Junfeng Tian, Rui Wang, Qi Zhang, Tao Gui, Xuanjing Huang

PlugAT: A Plug and Play Module to Defend against Textual Adversarial Attack (2022.coling-1)

Challenge:	Recent studies have shown that current TMSC systems rely on textual information, and the progress in tackling this task has slowed down.
Approach:	They propose to integrate both visual and textual information to improve the performance of TMSC by considering multimodal information.
Outcome:	The proposed model integrates both visual and textual information to improve performance.

Copied to clipboard

Rui Zheng, Rong Bao, Qin Liu, Tao Gui, Qi Zhang, Xuanjing Huang, Rui Xie, Wei Wu

Subspace Defense: Discarding Adversarial Perturbations by Learning a Subspace for Clean Signals (2024.lrec-main)

Challenge:	Recent studies have shown that adversarial examples can be easily fooled by adversarially perturbed examples.
Approach:	They propose a pluggable defense module PlugAT to provide robust predictions by adding a few trainable parameters to the model inputs while keeping the original model frozen.
Outcome:	The proposed model improves robustness over several strong baselines whilst training only 9.1% parameters.

Copied to clipboard

Rui Zheng, Yuhao Zhou, Zhiheng Xi, Tao Gui, Qi Zhang, Xuanjing Huang

Detecting Adversarial Samples through Sharpness of Loss Landscape (2023.findings-acl)

Challenge:	Existing models that extract discrete inputs into fixed-length representations are vulnerable to adversarial attacks that place perturbations on clean inputs to fool DNNs.
Approach:	They propose to inspect the subspaces of sample features through spectral analysis to better understand adversarial attacks.
Outcome:	The proposed strategy enables the model to inherently suppress adversaries, which boosts model robustness and motivates new directions of effective adversarial defense.

Copied to clipboard

Rui Zheng, Shihan Dou, Yuhao Zhou, Qin Liu, Tao Gui, Qi Zhang, Zhongyu Wei, Xuanjing Huang, Menghan Zhang

Toward Recognizing More Entity Types in NER: An Efficient Implementation using Only Entity Lexicons (2020.findings-emnlp)

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.

Copied to clipboard

Minlong Peng, Ruotian Ma, Qi Zhang, Lujun Zhao, Mengxi Wei, Changlong Sun, Xuanjing Huang

Dynamic and Generalizable Process Reward Modeling (2025.acl-long)

Challenge:	Existing named entity recognition systems require large scale labeled data to perform, while annotation of NER data is laborious and time-consuming.
Approach:	They propose to adjust an existing named entity recognition system to recognize entity types not defined in the system.
Outcome:	The proposed method can be quickly adjusted to a named entity recognition system.

Copied to clipboard

Zhangyue Yin, Qiushi Sun, Zhiyuan Zeng, Qinyuan Cheng, Xipeng Qiu, Xuanjing Huang

Challenge:	Existing Process Reward Models lack cross-domain generalization and focus on feedback results.
Approach:	They propose a process reward model that uses a reward tree to capture and store fine-grained, multi-dimensional reward criteria.
Outcome:	The proposed model performs on prevailing benchmarks and out-of-distribution scenarios.

A Simple Hash-Based Early Exiting Approach For Language Understanding and Generation (2022.findings-acl)

Copied to clipboard

Tianxiang Sun, Xiangyang Liu, Wei Zhu, Zhichao Geng, Lingling Wu, Yilong He, Yuan Ni, Guotong Xie, Xuanjing Huang, Xipeng Qiu

Challenge:	Existing methods to measure instance difficulty use generalization and threshold-tuning . a new approach to learn to exit is based on hash functions to assign tokens to a fixed exiting layer.
Approach:	They propose a Hash-based Early Exiting approach that replaces learn-to-exit modules with hash functions to assign each token to a fixed exiting layer.
Outcome:	The proposed approach improves on learning to exit and predicting instance difficulty.

Hi-ArG: Exploring the Integration of Hierarchical Argumentation Graphs in Language Pretraining (2023.emnlp-main)

Copied to clipboard

Jingcong Liang, Rong Ye, Meng Han, Qi Zhang, Ruofei Lai, Xinyu Zhang, Zhao Cao, Xuanjing Huang, Zhongyu Wei

Task-oriented Dialogue System for Automatic Diagnosis (P18-2)

Challenge:	Recent studies have discussed its capability to assist language models for various applications.
Approach:	They propose a structure to organize arguments using the Hierarchical Argumentation Graph (Hi-ArG) and propose two approaches to exploit Hi-AarG, including a text-graph multi-modal model GreaseArR and a framework augmented with graph information.
Outcome:	The proposed structure supersedes existing language models on two argumentation tasks while incorporating graph information during further training improves vanilla language models.

Copied to clipboard

Zhongyu Wei, Qianlong Liu, Baolin Peng, Huaixiao Tou, Ting Chen, Xuanjing Huang, Kam-fai Wong, Xiangying Dai

Natural Language Interfaces

Challenge:	Existing methods to identify phenotypes using electronic health records (EHRs) are expensive and difficult to transfer models from one disease to another.
Approach:	They propose a task-oriented dialogue system framework to make diagnosis for patients automatically, which can converse with patients to collect additional symptoms beyond their self-reports.
Outcome:	The proposed system can collect additional symptoms from conversation and improve disease identification accuracy.

A Reinforcement Learning Framework for Natural Question Generation using Bi-discriminators (C18-1)

Copied to clipboard

Zhihao Fan, Zhongyu Wei, Siyuan Wang, Yang Liu, Xuanjing Huang

Weight Perturbation as Defense against Adversarial Word Substitutions (2022.findings-emnlp)

Challenge:	Existing research on visual question generation is focused on training models to fit the annotated data set that makes them indifferent from other language generation tasks.
Approach:	They propose to use two discriminators to enhance the training of a visual question generator to ask natural questions about an image.
Outcome:	The proposed model outperforms state-of-the-art models in terms of automatic and human evaluation metrics.

Copied to clipboard

Jianhan Xu, Linyang Li, Jiping Zhang, Xiaoqing Zheng, Kai-Wei Chang, Cho-Jui Hsieh, Xuanjing Huang

Connectivity Patterns are Task Embeddings (2023.findings-acl)

Challenge:	Existence and pervasiveness of textual adversarial examples have raised serious concerns to security-critical applications.
Approach:	They propose to perform weight perturbations in the parameter space rather than the input feature space to improve adversarial robustness of NLP models.
Outcome:	The proposed method improves adversarial robustness of models by performing weight perturbations in the parameter space rather than the input feature space.

Copied to clipboard

Zhiheng Xi, Rui Zheng, Yuansen Zhang, Xuanjing Huang, Zhongyu Wei, Minlong Peng, Mingming Sun, Qi Zhang, Tao Gui

Reward Modeling Requires Automatic Adjustment Based on Data Quality (2024.findings-emnlp)

Challenge:	Existing methods for predicting inter-task transferability are sparse and task-specific.
Approach:	They propose a method that uses connectivity patterns of neurons as a unique identifier associated with a task.
Outcome:	The proposed method outperforms baselines in predicting inter-task transferability across data regimes and transfer settings while keeping high efficiency in computation and storage.

Copied to clipboard

Binghai Wang, Rui Zheng, Lu Chen, Zhiheng Xi, Wei Shen, Yuhao Zhou, Dong Yan, Tao Gui, Qi Zhang, Xuanjing Huang

Challenge:	Reinforcement Learning from Human Feedback (RLHF) is a method for aligning language models with human values.
Approach:	They propose a method that automatically adjusts reward modeling based on data quality . they use preference data to train a reward model that is more aligned with human values .
Outcome:	The proposed method stabilizes reward model training and significantly improves alignment performance on human preference datasets.

One-Model-Connects-All: A Unified Graph Pre-Training Model for Online Community Modeling (2023.findings-emnlp)

Copied to clipboard

Ruoxue Ma, Jiarong Xu, Xinnong Zhang, Haozhe Zhang, Zuyu Zhao, Qi Zhang, Xuanjing Huang, Zhongyu Wei

Characterizing the Impacts of Instances on Robustness (2023.findings-acl)

Challenge:	Existing studies on integrating online community to solve social problems have not fully utilized these three components and the relationship among them.
Approach:	They propose a framework that simultaneously considers communities, users, and texts and can easily connect with a variety of downstream tasks related to social media.
Outcome:	The proposed model can be used to perform violation detection, sentiment analysis, and community recommendation across multiple tasks.

Copied to clipboard

Rui Zheng, Zhiheng Xi, Qin Liu, Wenbin Lai, Tao Gui, Qi Zhang, Xuanjing Huang, Jin Ma, Ying Shan, Weifeng Ge

Open Set Relation Extraction via Unknown-Aware Training (2023.acl-long)

Challenge:	Existing defense approaches focus on developing new model structures or training algorithms, but they do little to tap the potential of training instances.
Approach:	They propose a method that can distinguish between robust and non-robust instances according to the model’s sensitivity to perturbations on individual instances during training.
Outcome:	The proposed method can distinguish between robust and non-robust instances according to the model’s sensitivity to perturbations on individual instances during training.

Copied to clipboard

Jun Zhao, Xin Zhao, WenYu Zhan, Qi Zhang, Tao Gui, Zhongyu Wei, Yun Wen Chen, Xiang Gao, Xuanjing Huang

Discrete Argument Representation Learning for Interactive Argument Pair Identification (2021.naacl-main)

Challenge:	Existing supervised relation extraction methods can still misclassify unknown relations into known relations due to the lack of supervision signals.
Approach:	They propose a method that regularizes the model by dynamically synthesizing negative instances that can provide the missing supervision signals.
Outcome:	The proposed method achieves SOTA unknown relation detection without compromising the classification of known relations.

Copied to clipboard

Lu Ji, Zhongyu Wei, Jing Li, Qi Zhang, Xuanjing Huang

SoMeLVLM: A Large Vision Language Model for Social Media Processing (2024.findings-acl)

Challenge:	Existing research on monological argumentation covers claims generation, argument structure prediction, and essay scoring.
Approach:	They propose to identify argument pairs from two posts with opposite stances to a certain topic.
Outcome:	The proposed framework outperforms competing models on a large-scale dataset . it also proves that it is useful for analyzing argument pairs from two posts .

Copied to clipboard

Xinnong Zhang, Haoyu Kuang, Xinyi Mou, Hanjia Lyu, Kun Wu, Siming Chen, Jiebo Luo, Xuanjing Huang, Zhongyu Wei

Bridging by Word: Image Grounded Vocabulary Construction for Visual Captioning (P19-1)

Challenge:	Genereal domain large models lack nuanced multimodal understanding of social media . general domain models focus more on text than other modalities, which is not consistent with real-world user habits.
Approach:	They propose a Large Vision Language Model for Social Media Processing that combines five key capabilities to understand and generate real social media behavior.
Outcome:	The proposed model achieves state-of-the-art performance in multiple social media tasks.

Copied to clipboard

Zhihao Fan, Zhongyu Wei, Siyuan Wang, Xuanjing Huang

K-Adapter: Infusing Knowledge into Pre-Trained Models with Adapters (2021.findings-acl)

Challenge:	Existing research on image captioning generates frequent n-grams with irrelevant words.
Approach:	They propose to construct an image-grounded vocabulary incorporating visual information and relations among words into the decoding process directly.
Outcome:	The proposed framework is compared with state-of-the-art models on MS COCO and Flickr30k and shows that it is more efficient than existing models.

Copied to clipboard

Ruize Wang, Duyu Tang, Nan Duan, Zhongyu Wei, Xuanjing Huang, Jianshu Ji, Guihong Cao, Daxin Jiang, Ming Zhou

Information Aggregation via Dynamic Routing for Sequence Encoding (C18-1)

Challenge:	Existing methods for injecting knowledge into pre-trained models are inconsistent and can flush out knowledge when multiple kinds of knowledge are injected.
Approach:	They propose a framework that retains the original parameters of pre-trained models fixed and supports the development of versatile knowledge-infused models.
Outcome:	The proposed framework retains the original parameters of the pre-trained model fixed and supports the development of versatile knowledge-infused models.

Copied to clipboard

Jingjing Gong, Xipeng Qiu, Shaojing Wang, Xuanjing Huang

ToolSword: Unveiling Safety Issues of Large Language Models in Tool Learning Across Three Stages (2024.acl-long)

Challenge:	Currently, little attention is paid to how to aggregate text sequences into fixed-size vectors.
Approach:	They propose an aggregation mechanism to obtain a fixed-size encoding with a dynamic routing policy.
Outcome:	The proposed method outperforms other aggregating methods on five text classification tasks.

Copied to clipboard

Junjie Ye, Sixian Li, Guanyu Li, Caishuang Huang, Songyang Gao, Yilong Wu, Qi Zhang, Tao Gui, Xuanjing Huang

Accelerating BERT Inference for Sequence Labeling via Early-Exit (2021.acl-long)

Challenge:	Existing research focuses on enhancing LLMs capabilities through tool utilization.
Approach:	They propose a framework to investigate safety issues in large language models in tool learning . they propose malicious queries and jailbreak attacks in the input stage .
Outcome:	The proposed framework investigates six safety scenarios for LLMs in tool learning . the data will be released upon acceptance of the proposed framework .

Copied to clipboard

Xiaonan Li, Yunfan Shao, Tianxiang Sun, Hang Yan, Xipeng Qiu, Xuanjing Huang

Robust Lottery Tickets for Pre-trained Language Models (2022.acl-long)

Challenge:	Existing early-exit mechanisms are designed for sequence-level tasks, rather than sequence labeling.
Approach:	They propose to extend sentence-level early-exit to accelerate inference of PTMs . they propose a token-level mechanism that allows partial tokens to exit early at different layers .
Outcome:	The proposed approach can save up to 66%75% inference cost with minimal performance degradation.

Copied to clipboard

Rui Zheng, Bao Rong, Yuhao Zhou, Di Liang, Sirui Wang, Wei Wu, Tao Gui, Qi Zhang, Xuanjing Huang

VehicleWorld: A Highly Integrated Multi-Device Environment for Intelligent Vehicle Interaction (2025.findings-emnlp)

Challenge:	Recent studies have shown that pre-trained language models contain smaller matching subnetworks that are not robust to adversarial examples.
Approach:	They propose a method to find robust tickets hidden in pre-trained language models by learning binary weight masks and an adversarial loss objective to guide the search.
Outcome:	The proposed method improves on previous work on adversarial robustness evaluation.

Copied to clipboard

Jie Yang, Jiajun Chen, Zhangyue Yin, Shuo Chen, Yuxin Wang, Yiran Guo, Yuan Li, Yining Zheng, Xuanjing Huang, Xipeng Qiu

Challenge:	Traditional Function Calling (FC) approaches operate statelessly, requiring multiple exploratory calls to build environmental awareness before execution, leading to inefficiency and limited error recovery.
Approach:	They propose a state-based function call approach that maintains explicit system state awareness and implements direct state transitions to achieve target conditions.
Outcome:	The proposed approach outperforms traditional function calling approaches, achieving superior execution accuracy and reduced latency.

Searching for Effective Neural Extractive Summarization: What Works and What’s Next (P19-1)

Copied to clipboard

Ming Zhong, Pengfei Liu, Danqing Wang, Xipeng Qiu, Xuanjing Huang

ARISE: An Adaptive Resolution-Aware Metric for Test-Time Scaling Evaluation in Large Reasoning Models (2026.findings-acl)

Challenge:	Recent years have seen success in the use of deep neural networks on text summarization, but there is no clear understanding of why they perform so well or how they might be improved.
Approach:	They propose to use different types of model architectures to improve extractive summarization systems.
Outcome:	The proposed framework achieves state-of-the-art on CNN/DailyMail by a large margin based on observations and analysis.

Copied to clipboard

Zhangyue Yin, Qiushi Sun, Zhiyuan Zeng, Zhiyuan Yu, Qipeng Guo, Xuanjing Huang, Xipeng Qiu

TableVLM: Multi-modal Pre-training for Table Structure Recognition (2023.acl-long)

Challenge:	Existing evaluation methods for test-time scaling are limited.
Approach:	They propose an adaptive resolution-aware scaling evaluation metric specifically designed to assess the test-time scaling effectiveness of large reasoning models.
Outcome:	The proposed metric provides a reliable and fine-grained measurement of test-time scaling capabilities, revealing significant variations in scaling efficiency across models.

Copied to clipboard

Leiyuan Chen, Chengsong Huang, Xiaoqing Zheng, Jinshu Lin, Xuanjing Huang

ORTicket: Let One Robust BERT Ticket Transfer across Different Tasks (2024.lrec-main)

Challenge:	Tables are useful for displaying data in an organized manner, but they are difficult to extract from images because of their structure, notation, and representation.
Approach:	They propose a multi-modal pre-training model for table structure recognition that captures table structure-related features by multiple unsupervised objectives inspired by masked visual-language modeling.
Outcome:	The proposed model improves tree-editing-distance-score on ComplexTable by 1.97% .

Copied to clipboard

Yuhao Zhou, Wenxiang Chen, Rui Zheng, Zhiheng Xi, Tao Gui, Qi Zhang, Xuanjing Huang

TextFusion: Privacy-Preserving Pre-trained Model Inference via Token Fusion (2022.emnlp-main)

Challenge:	Pretrained language models are susceptible to subtle perturbations and require multiple adversarial training during fine-tuning to improve their robustness.
Approach:	They propose a novel adversarial defense method ORTicket that fine-tunes a model for downstream tasks.
Outcome:	The proposed method achieves comparable robustness to other defense methods while maintaining the efficiency of fine-tuning.

Copied to clipboard

Xin Zhou, Jinzhu Lu, Tao Gui, Ruotian Ma, Zichu Fei, Yuran Wang, Yong Ding, Yibo Cheung, Qi Zhang, Xuanjing Huang

Toward Optimal LLM Alignments Using Two-Player Games (2025.findings-emnlp)

Challenge:	Existing methods to preserve inference privacy are available as cloud services . however, the risk of privacy leakage remains, according to recent studies .
Approach:	They propose a method to preserve inference privacy by fusing token representations in the cloud.
Outcome:	The proposed method preserves inference privacy without sacrificing performance on different scenarios.

Copied to clipboard

Rui Zheng, Hongyi Guo, Zhihan Liu, Xiaoying Zhang, Yuanshun Yao, Xiaojun Xu, Zhaoran Wang, Zhiheng Xi, Tao Gui, Qi Zhang, Xuanjing Huang, Yang Liu, Hang Li

LLM can Achieve Self-Regulation via Hyperparameter Aware Generation (2024.findings-acl)

Challenge:	Alignment of large language models (LLM) is a process that ensures the model’s responses to user prompts align with human intentions and social values.
Approach:	They propose an alignment method based on a two-agent game consisting of an adversarial agent and a defensive agent.
Outcome:	The proposed method improves on a two-agent game with an adversarial agent and a defensive agent.

Copied to clipboard

Siyin Wang, Shimin Li, Tianxiang Sun, Jinlan Fu, Qinyuan Cheng, Jiasheng Ye, Junjie Ye, Xipeng Qiu, Xuanjing Huang

Which Reasoning Trajectories Teach Students to Reason Better? A Simple Metric of Informative Alignment (2026.acl-long)

Challenge:	Existing decoding strategies and hyperparameters may not be optimal for each sample.
Approach:	They propose a model that auto-regulates decoding strategies and hyperparameters . this approach eliminates the need for extensive manual tuning, they argue .
Outcome:	The proposed model eliminates the need for extensive manual tuning, offering a more autonomous, self-regulate model behavior.

Copied to clipboard

Yuming Yang, Mingyoung Lai, Wanxu Zhao, Xiaoran Fan, Zhiheng Xi, Mingqi Wu, Chiyue Huang, Jun Zhao, Haijun Lv, Jian Tong, Yunhua Zhou, Yicheng Zou, Qipeng Guo, Tao Gui, Qi Zhang, Xuanjing Huang

Distantly Supervised Named Entity Recognition using Positive-Unlabeled Learning (P19-1)

Challenge:	Existing methods assess suitability primarily through student likelihood, favoring trajectories that align closely with the student model’s current behavior but overlooking more informative ones.
Approach:	They propose a Rank–Surprisal Ratio metric that captures both alignment and informativeness to assess the suitability of a reasoning trajectory.
Outcome:	The proposed metric captures both alignment and informativeness to assess the suitability of a reasoning trajectory.

Copied to clipboard

Minlong Peng, Xiaoyu Xing, Qi Zhang, Jinlan Fu, Xuanjing Huang

F-Eval: Asssessing Fundamental Abilities with Refined Evaluation Methods (2024.acl-long)

Challenge:	Empirical studies on four public NER datasets demonstrate the effectiveness of our proposed method.
Approach:	They propose a method to perform named entity recognition using unlabeled data and named entity dictionaries.
Outcome:	The proposed method can estimate task loss as if there is fully labeled data.

Copied to clipboard

Yu Sun, Keyuchen Keyuchen, Shujie Wang, Peiji Li, Qipeng Guo, Hang Yan, Xipeng Qiu, Xuanjing Huang, Dahua Lin

Challenge:	Large language models (LLMs) have been evaluated for their instruction-following capabilities but lack references to their fundamental abilities.
Approach:	They propose a bilingual evaluation benchmark to evaluate the fundamental abilities of large language models including expression, commonsense and logic.
Outcome:	The proposed evaluation methods show higher correlation coefficients and larger distinction than other evaluators.

Defense against Synonym Substitution-based Adversarial Attacks via Dirichlet Neighborhood Ensemble (2021.acl-long)

Copied to clipboard

Yi Zhou, Xiaoqing Zheng, Cho-Jui Hsieh, Kai-Wei Chang, Xuanjing Huang

Efficient Adversarial Training with Robust Early-Bird Tickets (2022.emnlp-main)

Challenge:	Recent studies show vulnerability of deep neural networks to adversarial examples that intentionally fool the networks.
Approach:	They propose a method for training a robust model to defense synonym substitution-based attacks by sampling embedding vectors for each word in an input sentence and augmenting them with the training data.
Outcome:	The proposed method outperforms other proposed defense methods by a significant margin across different network architectures and multiple data sets.

Copied to clipboard

Zhiheng Xi, Rui Zheng, Tao Gui, Qi Zhang, Xuanjing Huang

Locate Then Ask: Interpretable Stepwise Reasoning for Multi-hop Question Answering (2022.coling-1)

Challenge:	Existing methods to improve the robustness of pre-trained language models are expensive because of the need to generate adversarial examples via gradient descent.
Approach:	They propose an adversarial optimization method that searches for robust tickets with structured sparsity in the early stage and fine-tunes tickets in the remaining time.
Outcome:	The proposed method achieves up to 7 13 training speedups while maintaining comparable or even better robustness compared to the most competitive state-of-the-art methods.

Copied to clipboard

Siyuan Wang, Zhongyu Wei, Zhihao Fan, Qi Zhang, Xuanjing Huang

MM-Doc-R1: Training Agents for Long Document Visual Question Answering through Multi-turn Reinforcement Learning (2026.findings-acl)

Challenge:	Existing methods for multi-hop reasoning ignore grounding on supporting facts of each step, which tends to generate inaccurate decompositions.
Approach:	They propose an interpretable stepwise reasoning framework that incorporates supporting sentences and questions at each intermediate step and utilizes the inference of the current hop for the next until reasoning out the final result.
Outcome:	The proposed model can boost performance and yield a better interpretable reasoning process without decomposition supervision.

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

LLMEval-Med: A Real-world Clinical Benchmark for Medical LLMs with Physician Validation (2025.findings-emnlp)

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.

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

TL-Training: A Task-Feature-Based Framework for Training Large Language Models in Tool Use (2025.findings-emnlp)

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.

Copied to clipboard

Junjie Ye, Yilong Wu, Sixian Li, Yuming Yang, Zhiheng Xi, Tao Gui, Qi Zhang, Xuanjing Huang, Peng Wang, Zhongchao Shi, Jianping Fan, Zhengyin Du

UPLex: Fine-Grained Personality Control in Large Language Models via Unsupervised Lexical Modulation (2025.findings-emnlp)

Challenge:	a new approach to training large language models (LLMs) overlooks task-specific characteristics in tool use, leading to performance bottlenecks.
Approach:	They propose a task-feature-based framework that mitigates the effects of suboptimal training data . they use a dataset to train large-scale LLMs and a reward mechanism tailored to error categories .
Outcome:	The proposed framework matches or surpasses open- and closed-source LLMs in tool-use performance using only 1,217 training data points.

Copied to clipboard

Tianlong Li, Wenhao Liu, Muling Wu, Shihan Dou, Zhenghua Wang, Changze Lv, Xiaohua Wang, Xiaoqing Zheng, Xuanjing Huang

An Enhanced Knowledge Injection Model for Commonsense Generation (2020.coling-main)

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.

Copied to clipboard

Zhihao Fan, Yeyun Gong, Zhongyu Wei, Siyuan Wang, Yameng Huang, Jian Jiao, Xuanjing Huang, Nan Duan, Ruofei Zhang

ToolEyes: Fine-Grained Evaluation for Tool Learning Capabilities of Large Language Models in Real-world Scenarios (2025.coling-main)

Challenge:	a recent study shows that digging the relationship of concepts from scratch is non-trivial for commonsense generation tasks.
Approach:	They use a retrieve-and-edit framework to retrieve a prototype with these concepts . they use qt and qq to generate commonsense questions at scale .
Outcome:	The proposed method significantly improves the performance on commonsense generation tasks.

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

Style Transformer: Unpaired Text Style Transfer without Disentangled Latent Representation (P19-1)

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.

Copied to clipboard

Ning Dai, Jianze Liang, Xipeng Qiu, Xuanjing Huang

Activating Distributed Visual Region within LLMs for Efficient and Effective Vision-Language Training and Inference (2025.acl-long)

Challenge:	Disentangling the content and style in the latent space is prevalent in text style transfer . recurrent neural networks (RNN) based encoder and decoder cannot deal with the long-term dependency .
Approach:	They propose a style transformer which disentangles style information in latent space . they propose encoding and decoding methods that disentangle style information .
Outcome:	The proposed method can achieve better style transfer and better content preservation.

Copied to clipboard

Siyuan Wang, Dianyi Wang, Chengxing Zhou, Zejun Li, Zhihao Fan, Xuanjing Huang, Zhongyu Wei

Inverse-Q*: Token Level Reinforcement Learning for Aligning Large Language Models Without Preference Data (2024.findings-emnlp)

Challenge:	Existing Large Vision-Language Models (LVLMs) learn visual capacity through visual instruction tuning.
Approach:	They propose a method for LVLMs to be trained by selective layers tuning . they propose removing non-critical layers outside the visual region .
Outcome:	The proposed approach preserves nearly 99% of visual performance and improves textual task results while reducing training time.

Copied to clipboard

Han Xia, Songyang Gao, Qiming Ge, Zhiheng Xi, Qi Zhang, Xuanjing Huang

RIVAL: Reinforcement Learning with Iterative and Adversarial Optimization for Machine Translation (2025.findings-emnlp)

Challenge:	Reinforcement Learning from Human Feedback (RLHF) relies on complex methodologies like Proximal Policy Optimization (PPO) that require extensive hyper-parameter tuning and pose challenges in sample efficiency and stability.
Approach:	They propose an innovative framework that leverages direct preference optimization techniques but extends them by estimating the conditionally optimal policy directly from the model’s responses.
Outcome:	The proposed framework matches and exceeds the effectiveness of Proximal Policy Optimization (PPO) in terms of convergence speed and alignment of model responses with human preferences.

Copied to clipboard

Tianjiao Li, Mengran Yu, Chenyu Shi, Yanjun Zhao, Xiaojing Liu, Qi Zhang, Xuanjing Huang, Qiang Zhang, Jiayin Wang

Keep it Consistent: Topic-Aware Storytelling from an Image Stream via Iterative Multi-agent Communication (2020.coling-main)

Challenge:	Using reinforcement learning from human feedback, large language models perform poorly when applied to colloquial subtitle translation tasks.
Approach:	They propose an adversarial training framework that iteratively updates the offline reward model and the online LLM to improve training outcomes.
Outcome:	The proposed training framework significantly improves upon translation baselines.

Copied to clipboard

Ruize Wang, Zhongyu Wei, Ying Cheng, Piji Li, Haijun Shan, Ji Zhang, Qi Zhang, Xuanjing Huang

Cross-Domain Sentiment Classification with Target Domain Specific Information (P18-1)

Challenge:	Existing methods for visual storytelling construct text description independently for each image and roughly concatenate them as a story, which leads to the problem of generating semantically incoherent content.
Approach:	They propose a topic description task to detect the global semantic context of an image stream and a story is then constructed with the guidance of the topic description.
Outcome:	The proposed framework can generate stories with higher quality compared to state-of-the-art methods on a VIST dataset.

Copied to clipboard

Minlong Peng, Qi Zhang, Yu-gang Jiang, Xuanjing Huang

Extractive Summarization as Text Matching (2020.acl-main)

Challenge:	Existing methods for sentiment classification focus on learning domain-invariant representations . few of them pay attention to domain-specific information, which should also be informative.
Approach:	They propose a method to extract domain specific and invariant representations and train a classifier on each of them.
Outcome:	The proposed model can achieve better performance than state-of-the-art methods.

Copied to clipboard

Ming Zhong, Pengfei Liu, Yiran Chen, Danqing Wang, Xipeng Qiu, Xuanjing Huang

LifeSim: Long-Horizon User Life Simulator for Personalized Assistant Evaluation (2026.findings-acl)

Challenge:	Currently, most of the neural extractive summarization systems score and extract sentences individually and model the relationship between sentences.
Approach:	They propose to instantiate a neural extractive summarization task as a semantic text matching problem and use it to match a source document and candidate summaries in a semantic space.
Outcome:	The proposed framework is faster and more efficient than existing frameworks.

Copied to clipboard

Feiyu Duan, Xuanjing Huang, Zhongyu Wei

Challenge:	Existing benchmarks for personalized assistants fail to capture the complexity of external contexts and users’ cognitive states.
Approach:	They propose a user simulator that models user cognition through the Belief-Desire-Intention model within physical environments for coherent life trajectories generation and simulates intention-driven user interactive behaviors.
Outcome:	The proposed model can model user cognition through the Belief-Desire-Intention model within physical environments for coherent life trajectories generation and simulates intention-driven user interactive behaviors.

Advancing Parameter Efficiency in Fine-tuning via Representation Editing (2024.acl-long)

Copied to clipboard

Muling Wu, Wenhao Liu, Xiaohua Wang, Tianlong Li, Changze Lv, Zixuan Ling, Zhu JianHao, Cenyuan Zhang, Xiaoqing Zheng, Xuanjing Huang

LLMs Deceive Unintentionally: Emergent Misalignment in Dishonesty from Misaligned Samples to Biased Human-AI Interactions (2026.findings-acl)

Challenge:	Parameter Efficient Fine-Tuning (PEFT) has gained significant attention for its ability to achieve competitive results while updating only a small subset of trainable parameters.
Approach:	They propose a new approach to fine-tuning neural models that scales and biases the representation produced at each layer.
Outcome:	The proposed approach reduces the number of trainable parameters by a factor of 25,700 compared to full parameter fine-tuning and by . 32 compared with LoRA.

Copied to clipboard

Xuhao Hu, Wang Peng, Xiaoya Lu, Dongrui Liu, Xuanjing Huang, Jing Shao

Larger-Context Tagging: When and Why Does It Work? (2021.naacl-main)

Challenge:	Existing studies have shown that LLMs finetuned on incorrect completions can exhibit harmful behaviors, which is called emergent misalignment.
Approach:	They investigate whether LLMs finetuned on incorrect completions can exhibit harmful behaviors . they find that 1% of misalignment data is sufficient to decrease honest behavior .
Outcome:	The proposed model can be misaligned on errors within narrow domains to exhibit harmful behaviors . the proposed model is able to exhibit dishonest behavior with only 10% biased user population .

Copied to clipboard

Jinlan Fu, Liangjing Feng, Qi Zhang, Xuanjing Huang, Pengfei Liu

Parrot: A Training Pipeline Enhances Both Program CoT and Natural Language CoT for Reasoning (2025.emnlp-main)

Challenge:	Existing tagging systems that use sentence-level data are not well understood.
Approach:	They propose a larger-context approach to tagging tasks that incorporates contextual information into existing tapping systems.
Outcome:	The proposed aggregators improve on four tagging tasks and 13 datasets.

Copied to clipboard

Senjie Jin, Lu Chen, Zhiheng Xi, Yuhui Wang, Sirui Song, Yuhao Zhou, Xinbo Zhang, Peng Sun, Hong Lu, Tao Gui, Qi Zhang, Xuanjing Huang

RoTBench: A Multi-Level Benchmark for Evaluating the Robustness of Large Language Models in Tool Learning (2024.emnlp-main)

Challenge:	Existing work focuses on enabling models to generate natural language chain-of-thought rationales or leverage executable and verifiable code, such as Python.
Approach:	They propose a novel training pipeline that integrates sequential P-CoT and N-Co T generation and a subtask hybrid training strategy to facilitate natural language transferability.
Outcome:	The proposed training pipeline improves both N-CoT and P-Co T performance over the RL baseline.

Copied to clipboard

Junjie Ye, Yilong Wu, Songyang Gao, Caishuang Huang, Sixian Li, Guanyu Li, Xiaoran Fan, Qi Zhang, Tao Gui, Xuanjing Huang

A Closer Look at Data Bias in Neural Extractive Summarization Models (D19-54)

Challenge:	Current research emphasizes LLMs’ capacity to utilize tools in well-structured environments while overlooking their stability when confronted with the inevitable noise of the real world.
Approach:	They propose a multi-level benchmark to evaluate the robustness of large language models in tool learning by establishing five external environments with varying levels of noise.
Outcome:	The proposed model outperforms the GPT-4 model in tool learning in three critical phases: tool selection, parameter identification, and content filling.

Copied to clipboard

Ming Zhong, Danqing Wang, Pengfei Liu, Xipeng Qiu, Xuanjing Huang

Incorporating Argument-Level Interactions for Persuasion Comments Evaluation using Co-attention Model (C18-1)

Challenge:	In this paper, we examine the generalization behaviour of summarization models . we propose several properties of datasets that matter for generalization .
Approach:	They propose several properties of datasets which matter for generalization of summarization models.
Outcome:	The proposed approach improves the state-of-the-art model by rethinking the model design process on a typical dataset.

Copied to clipboard

Lu Ji, Zhongyu Wei, Xiangkun Hu, Yang Liu, Qi Zhang, Xuanjing Huang

Attending via both Fine-tuning and Compressing (2021.findings-acl)

Challenge:	Existing research explores different text features of reply comments on word level and ignores interactions between participants.
Approach:	They propose a co-attention mechanism based neural network to capture interactions between participants on argument level to better model dialogical argumentation.
Outcome:	The proposed model outperforms state-of-the-art methods on a publicly available dataset showing that it extracts interactive argument pairs from the original post and the reply.

Copied to clipboard

Jie Zhou, Yuanbin Wu, Qin Chen, Xuanjing Huang, Liang He

Explainability & Interpretability in NLP

Challenge:	Existing studies show that attention mechanisms can improve models' interpretation, but they are not explicable.
Approach:	They propose a framework consisting of a learner and a compressor to purify attention scores . they propose to fine-tune and compress the attention mechanism to obtain a more faithful explanation .
Outcome:	The proposed framework improves performance and interpretability on eight benchmark datasets.

CoLAKE: Contextualized Language and Knowledge Embedding (2020.coling-main)

Copied to clipboard

Tianxiang Sun, Yunfan Shao, Xipeng Qiu, Qipeng Guo, Yaru Hu, Xuanjing Huang, Zheng Zhang

Improving Abstractive Dialogue Summarization with Speaker-Aware Supervised Contrastive Learning (2022.coling-1)

Challenge:	Existing models for integrating factual knowledge into pre-trained language models are shallow, static, and separately pre-train entities.
Approach:	They propose a method which integrates knowledge contexts from large-scale knowledge bases into a unified data structure.
Outcome:	The proposed model outperforms existing models on knowledge-driven tasks and knowledge probing tasks.

Copied to clipboard

Zhichao Geng, Ming Zhong, Zhangyue Yin, Xipeng Qiu, Xuanjing Huang

DiffusionBERT: Improving Generative Masked Language Models with Diffusion Models (2023.acl-long)

Challenge:	Existing summarization systems based on pre-trained models cannot recognize the unique format of the speaker-utterance pair well in the dialogue.
Approach:	They propose three speaker-aware supervised contrastive learning tasks to solve the speaker identification problem in dialogue summarization task.
Outcome:	The proposed methods improve on two mainstream dialogue summarization datasets.

Copied to clipboard

Zhengfu He, Tianxiang Sun, Qiong Tang, Kuanning Wang, Xuanjing Huang, Xipeng Qiu

Lost in the Context: Insufficient and Distracted Attention to Contexts in Preference Modeling (2025.acl-long)

Challenge:	Existing generative masked language models have a shared training objective, i.e., denoising.
Approach:	They propose a noise schedule for the forward diffusion process that controls the degree of noise added at each step based on the information of each token.
Outcome:	The proposed model improves on existing models in terms of perplexity and BLEU score.

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

SongYang Gao, Shihan Dou, Qi Zhang, Xuanjing Huang, Jin Ma, Ying Shan

SpanNER: Named Entity Re-/Recognition as Span Prediction (2021.acl-long)

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.

Copied to clipboard

Jinlan Fu, Xuanjing Huang, Pengfei Liu

RoCoIns: Enhancing Robustness of Large Language Models through Code-Style Instructions (2024.lrec-main)

Challenge:	Recent years have seen the paradigm shift of Named Entity Recognition (NER) systems from sequence labeling to span prediction.
Approach:	They experimentally implement 154 named entity recognition models on 11 datasets and show that span prediction can serve as a system combiner to re-recognize named entities from different systems’ outputs.
Outcome:	The proposed model can be used to re-recognize named entities from different systems’ outputs.

Copied to clipboard

Yuansen Zhang, Xiao Wang, Zhiheng Xi, Han Xia, Tao Gui, Qi Zhang, Xuanjing Huang

A Knowledge-Aware Sequence-to-Tree Network for Math Word Problem Solving (2020.emnlp-main)

Challenge:	Large Language Models (LLMs) have shown remarkable capabilities in following human instructions and solving NLU tasks.
Approach:	They propose to use code style instructions to replace typically natural language instructions to provide more precise instructions and strengthen the robustness of LLMs.
Outcome:	The proposed method outperforms natural language models on eight robustness datasets and achieves an improvement of 5.68% in test set accuracy and a reduction of 5.66 points in Attack Success Rate (ASR).

Copied to clipboard

Qinzhuo Wu, Qi Zhang, Jinlan Fu, Xuanjing Huang

DELAN: Dual-Level Alignment for Vision-and-Language Navigation by Cross-Modal Contrastive Learning (2024.lrec-main)

Challenge:	Existing methods for solving math word problems ignore background common-sense knowledge . a novel knowledge-aware sequence-to-tree (KA-S2T) network incorporates external knowledge and global expression information.
Approach:	They propose a knowledge-aware sequence-to-tree network that incorporates external knowledge and global expression information into the problem.
Outcome:	The proposed model can achieve better performance than previous models on a Math23K dataset.

Copied to clipboard

Mengfei Du, Binhao Wu, Jiwen Zhang, Zhihao Fan, Zejun Li, Ruipu Luo, Xuanjing Huang, Zhongyu Wei

Self-Demos: Eliciting Out-of-Demonstration Generalizability in Large Language Models (2024.findings-naacl)

Challenge:	Existing studies focus on cross-modal attention at the fusion stage, but modality features generated by disparate uni-encoders reside in their own spaces, leading to a decline in the quality of cross-modulation and decision-making.
Approach:	They propose a framework to align navigation-related modalities before fusion by cross-modal contrastive learning.
Outcome:	The proposed framework integrates with the majority of existing models, resulting in improved navigation performance on various VLN benchmarks, including R2R, R4R, and CVDN.

Copied to clipboard

Wei He, Shichun Liu, Jun Zhao, Yiwen Ding, Yi Lu, Zhiheng Xi, Tao Gui, Qi Zhang, Xuanjing Huang

Making Harmful Behaviors Unlearnable for Large Language Models (2024.findings-acl)

Challenge:	Existing methods that rely on limited demos and out-of-demonstration (OOD) queries fail when faced with out- of-demotion queries.
Approach:	They propose a query-aware prompting method that elicits the inherent generalizability of large language models by query-based demo generation.
Outcome:	The proposed method outperforms state-of-the-art methods in the OOD setting and two public math benchmarks.

Copied to clipboard

Xin Zhou, Yi Lu, Ruotian Ma, Yujian Wei, Tao Gui, Qi Zhang, Xuanjing Huang

VCWE: Visual Character-Enhanced Word Embeddings (N19-1)

Challenge:	Large language models (LLMs) are often customized by fine-tuning for the requirements of different domains.
Approach:	They propose a controllable training framework to make undesired behaviors unlearnable during the fine-tuning process.
Outcome:	The proposed framework makes undesired behaviors unlearnable during the fine-tuning process while preserving the ability to learn other information.

Copied to clipboard

Chi Sun, Xipeng Qiu, Xuanjing Huang

Multitask Pre-training of Modular Prompt for Chinese Few-Shot Learning (2023.acl-long)

Challenge:	Currently, word embeddings are playing a pivotal role in many natural language processing tasks.
Approach:	They propose a model to learn Chinese word embeddings via three-level composition . they use convolutional neural network to extract intra-character compositionality from character shape .
Outcome:	The proposed model performs better on word similarity, sentiment analysis, named entity recognition and part-of-speech tagging tasks.

Copied to clipboard

Tianxiang Sun, Zhengfu He, Qin Zhu, Xipeng Qiu, Xuanjing Huang

SENT: Sentence-level Distant Relation Extraction via Negative Training (2021.acl-long)

Challenge:	Prompt tuning is a parameter-efficient approach to adapting pre-trained language models to downstream tasks.
Approach:	They propose to combine pre-trained modules with pre-trains to boost prompt tuning for few-shot learning.
Outcome:	The proposed model outperforms prompt tuning, full model tuning, and prior prompt pre-training methods in few-shot learning settings.

Copied to clipboard

Ruotian Ma, Tao Gui, Linyang Li, Qi Zhang, Xuanjing Huang, Yaqian Zhou

VIB-Probe: Detecting and Mitigating Hallucinations in Vision-Language Models via Variational Information Bottleneck (2026.acl-long)

Challenge:	Existing methods for relation extraction use bag labels, which introduce noise, to train the model.
Approach:	They propose to use negative training to train a model using complementary labels to separate the noisy data from the training data.
Outcome:	The proposed method improves on previous methods on sentence-level evaluation and de-noise effect.

Copied to clipboard

Feiran Zhang, Yixin Wu, Zhenghua Wang, Xiaohua Wang, Changze Lv, Xuanjing Huang, Xiaoqing Zheng

CoLo: A Contrastive Learning Based Re-ranking Framework for One-Stage Summarization (2022.coling-1)

Challenge:	Existing hallucination detection methods rely on external verification tools . however, entanglement of visual-linguistic syntax and noise makes it difficult to detect hallucis .
Approach:	They propose a hallucination detection framework that leverages the Variational Information Bottleneck theory to detect hallucinic heads and to infer hallucication mitigation strategies.
Outcome:	The proposed framework outperforms baselines in hallucinations and noise detection environments.

Copied to clipboard

Chenxin An, Ming Zhong, Zhiyong Wu, Qin Zhu, Xuanjing Huang, Xipeng Qiu

PASUM: A Pre-training Architecture for Social Media User Modeling Based on Text Graph (2024.lrec-main)

Challenge:	Existing methods for extractive and abstractive summarization use token-level or sentence-level training objectives.
Approach:	They propose a Contrastive Learning based re-ranking framework for one-stage summarization called CoLo.
Outcome:	The proposed framework boosts extractive and abstractive results on CNN/DailyMail benchmarks while maintaining inference efficiency.

Copied to clipboard

Kun Wu, Xinyi Mou, Lanqing Xue, Zhenzhe Ying, Weiqiang Wang, Qi Zhang, Xuanjing Huang, Zhongyu Wei

Generating Responses with a Specific Emotion in Dialog (P19-1)

Challenge:	Existing studies have incorporated different digital traces to better learn the representations of social media users, limited by overloaded text information and hard-to-collect social network information.
Approach:	They propose a Pre-training Architecture for Social Media User Modeling based on Text Graph and combine microblogs to represent social media users based upon the text graph model.
Outcome:	The proposed framework can represent users based on text even without social network information on microblogs.

Copied to clipboard

Zhenqiao Song, Xiaoqing Zheng, Lu Liu, Mu Xu, Xuanjing Huang

Debatrix: Multi-dimensional Debate Judge with Iterative Chronological Analysis Based on LLM (2024.findings-acl)

Challenge:	EmoDS can express emotions in both ways, but it is difficult to scale to large datasets.
Approach:	They propose an emotional dialog system that can express emotions in both ways . they use strong emotional words and neutral words to increase the intensity of emotions .
Outcome:	The proposed system performs better than baselines in BLEU, diversity and quality of emotional expression.

Copied to clipboard

Jingcong Liang, Rong Ye, Meng Han, Ruofei Lai, Xinyu Zhang, Xuanjing Huang, Zhongyu Wei

Challenge:	Recent studies have focused on short dialogues, but mainly on short debates.
Approach:	They propose to use Large Language Models to construct an automated debate judge to evaluate multi-turn debates.
Outcome:	The proposed system improves on the PanelBench benchmark, which compares its performance to actual debate outcomes.

GlossBERT: BERT for Word Sense Disambiguation with Gloss Knowledge (D19-1)

Copied to clipboard

Luyao Huang, Chi Sun, Xipeng Qiu, Xuanjing Huang

Unifying Cross-Lingual and Cross-Modal Modeling Towards Weakly Supervised Multilingual Vision-Language Pre-training (2023.acl-long)

Challenge:	Word Sense Disambiguation (WSD) aims to find the exact sense of an ambiguous word in a particular context.
Approach:	They propose to integrate gloss knowledge into supervised neural networks for Word Sense Disambiguation (WSD) this paper proposes to fine-tune a pre-trained BERT model and achieve new state-of-the-art results on WSD task.
Outcome:	The proposed model achieves state-of-the-art on the word Sense Disambiguation (WSD) task.

Copied to clipboard

Zejun Li, Zhihao Fan, Jingjing Chen, Qi Zhang, Xuanjing Huang, Zhongyu Wei

Multilinguality

Challenge:	Existing studies address the problem of translating English data into other languages, but they are limited in form and scale.
Approach:	They propose a framework to unify cross-lingual and cross-modal pre-training by using English data.
Outcome:	The proposed framework unifies cross-lingual and cross-modal pre-training on different data.

Towards Building More Robust NER datasets: An Empirical Study on NER Dataset Bias from a Dataset Difficulty View (2023.emnlp-main)

Copied to clipboard

Ruotian Ma, Xiaolei Wang, Xin Zhou, Qi Zhang, Xuanjing Huang

StepCoder: Improving Code Generation with Reinforcement Learning from Compiler Feedback (2024.acl-long)

Challenge:	Named Entity Recognition (NER) models rely on superficial entity patterns for predictions, without considering evidence from the context.
Approach:	They propose to de-bias NER datasets by altering entity-context distribution . they also validate the feasibility of the proposed de-bianking techniques .
Outcome:	The proposed methods can be applied to different models and improve existing models.

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

Programming Languages in NLP

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.

Simplify the Usage of Lexicon in Chinese NER (2020.acl-main)

Copied to clipboard

Ruotian Ma, Minlong Peng, Qi Zhang, Zhongyu Wei, Xuanjing Huang

EmbSpatial-Bench: Benchmarking Spatial Understanding for Embodied Tasks with Large Vision-Language Models (2024.acl-short)

Challenge:	Named entity recognition (NER) is concerned with the identification of named entities in unstructured text.
Approach:	They propose a method for incorporating word lexicon into character representations . experimental results show method can be easily incorporated with pre-trained models .
Outcome:	The proposed method achieves 6.15 times faster inference speed and better performance on four benchmark Chinese NER datasets.

Copied to clipboard

Mengfei Du, Binhao Wu, Zejun Li, Xuanjing Huang, Zhongyu Wei

UPPAM: A Unified Pre-training Architecture for Political Actor Modeling based on Language (2023.acl-long)

Challenge:	Recent studies have revealed significant deficiencies of LVLMs in understanding visual contents, leaving the gap between current embodied intelligence and large vision-language models (LVLM) .
Approach:	They propose to use a benchmark to evaluate LVLMs' spatial understanding of embodied environments to evaluate their ability to understand visual contents.
Outcome:	The proposed benchmark is derived from embodied scenes and covers 6 spatial relationships from an egocentric perspective.

Copied to clipboard

Xinyi Mou, Zhongyu Wei, Qi Zhang, Xuanjing Huang

Enhancing Unsupervised Semantic Parsing with Distributed Contextual Representations (2023.findings-acl)

Challenge:	Existing studies have incorporated contextual information to better learn the representation of political actors for specific tasks.
Approach:	They propose to use statements to represent political actors and learn mapping from languages to representations using social networks and behaviors as self-constructed supervision.
Outcome:	The proposed model can be generalized to political actors and solve downstream tasks.

Copied to clipboard

Zixuan Ling, Xiaoqing Zheng, Jianhan Xu, Jinshu Lin, Kai-Wei Chang, Cho-Jui Hsieh, Xuanjing Huang

Mitigating Position Bias in Transformers via Layer-Specific Positional Embedding Scaling (2026.findings-acl)

Challenge:	Existing methods to learn models on corpus of pairs of sentences require labor-intensive annotation.
Approach:	They propose to leverage distributed contextual word and phrase representations pre-trained on unlabelled texts to deal with homonymy and polysemy.
Outcome:	The proposed model achieves better accuracy on question-answering and relation extraction tasks.

Copied to clipboard

Zhenghua Wang, Yiran Ding, Changze Lv, Yixin Wu, Tianlong Li, Zhibo Xu, Muling Wu, Tianyuan Shi, Shizheng Li, Qi Qian, Xuanjing Huang, Xiaoqing Zheng

Revisiting Jailbreaking for Large Language Models: A Representation Engineering Perspective (2025.coling-main)

Challenge:	Existing methods to address the "lost-in-the-middle" problem suffer from high latency or suboptimal hand-crafted scaling strategies.
Approach:	They propose a layer-specific positional embedding scaling method that assigns distinct scaling factors to each layer.
Outcome:	Experiments show that the proposed method mitigates positional attention bias and delivers consistent improvements across multiple long-context benchmarks.

Copied to clipboard

Tianlong Li, Zhenghua Wang, Wenhao Liu, Muling Wu, Shihan Dou, Changze Lv, Xiaohua Wang, Xiaoqing Zheng, Xuanjing Huang

fastHan: A BERT-based Multi-Task Toolkit for Chinese NLP (2021.acl-demo)

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.

Copied to clipboard

Zhichao Geng, Hang Yan, Xipeng Qiu, Xuanjing Huang

RealBehavior: A Framework for Faithfully Characterizing Foundation Models’ Human-like Behavior Mechanisms (2023.findings-emnlp)

Challenge:	Recently, the need for Chinese natural language processing (NLP) has a dramatic increase for many downstream applications.
Approach:	They propose to use Chinese word segmentation (CWS), Part-of-Speech (POS) tagging, named entity recognition (NER), and dependency parsing to train a multi-task model based on a pruned BERT.
Outcome:	The proposed model performs better than popular segmentation tools on a non-training corpus.

Copied to clipboard

Enyu Zhou, Rui Zheng, Zhiheng Xi, Songyang Gao, Xiaoran Fan, Zichu Fei, Jingting Ye, Tao Gui, Qi Zhang, Xuanjing Huang

Challenge:	Existing studies on human-like behaviors in foundation models do not verify their faithfulness . a simple application of psychological tools cannot faithfully characterize all human-type behaviors .
Approach:	They propose a framework to characterize humanoid behaviors in foundation models . they argue that a simple application of psychological tools cannot faithfully characterize all human-like behaviors .
Outcome:	The proposed framework assesses the faithfulness of results based on reproducibility, internal consistency, and generalizability.

Are Structural Concepts Universal in Transformer Language Models? Towards Interpretable Cross-Lingual Generalization (2023.findings-emnlp)

Copied to clipboard

Ningyu Xu, Qi Zhang, Jingting Ye, Menghan Zhang, Xuanjing Huang

UTC-IE: A Unified Token-pair Classification Architecture for Information Extraction (2023.acl-long)

Challenge:	Large language models (LLMs) have implicitly transfer knowledge across languages, but not all languages have such generalization capabilities.
Approach:	They propose a meta-learning-based method to learn to align conceptual spaces of different languages to enhance cross-lingual generalization.
Outcome:	The proposed method achieves competitive results with state-of-the-art methods and narrows the performance gap between languages.

Copied to clipboard

Hang Yan, Yu Sun, Xiaonan Li, Yunhua Zhou, Xuanjing Huang, Xipeng Qiu

Analyzing the Effects of Supervised Fine-Tuning on Model Knowledge from Token and Parameter Levels (2025.emnlp-main)

Challenge:	Information Extraction (IE) tasks have been solved with different models because of their output structures.
Approach:	They propose a Unified Token-pair Classification architecture for Information Extraction that introduces Plusformer on top of the token-pear feature matrix.
Outcome:	The proposed approach outperforms task-specific and unified models on all tasks in 10 datasets and achieves better results on 2 joint IE datasets.

Copied to clipboard

Junjie Ye, Yuming Yang, Yang Nan, Shuo Li, Qi Zhang, Tao Gui, Xuanjing Huang, Peng Wang, Zhongchao Shi, Jianping Fan

VRPO: Rethinking Value Modeling for Robust RL under Noisy Supervision in LLM Post-Training (2026.acl-long)

Challenge:	Large language models (LLMs) acquire substantial world knowledge during pretraining, which is further shaped by post-training techniques such as supervised fine-tuning (SFT).
Approach:	They evaluate closed-book question answering (CBQA) performance across five LLMs from the LLaMA-2 and LLama-3 families and examine the impact of supervised fine-tuning on model knowledge.
Outcome:	The proposed model performance is 14% worse than models fine-tuned on 1,920 samples and 12% worse on 240 samples.

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

Hallucination Detection for Generative Large Language Models by Bayesian Sequential Estimation (2023.emnlp-main)

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.

Copied to clipboard

Xiaohua Wang, Yuliang Yan, Longtao Huang, Xiaoqing Zheng, Xuanjing Huang

Feedback-Driven Tool-Use Improvements in Large Language Models via Automated Build Environments (2026.findings-acl)

Challenge:	Existing methods for detecting hallucinations require large numbers of observations to be retrieved, increasing response times.
Approach:	They propose a framework that leverages Bayesian sequential analysis to optimize the trade-off between costs and benefits during the hallucination detection process.
Outcome:	The proposed framework surpasses existing methods in efficiency and precision of hallucination detection.

Copied to clipboard

Junjie Ye, Changhao Jiang, Zhengyin Du, Yufei Xu, Xuesong Yao, Zhiheng Xi, Xiaoran Fan, Qi Zhang, Tao Gui, Xuanjing Huang, Jiecao Chen

Late Prompt Tuning: A Late Prompt Could Be Better Than Many Prompts (2022.findings-emnlp)

Challenge:	Currently, there are no efficient reinforcement learning (RL) frameworks specifically designed for tool use.
Approach:	They propose an automated environment construction pipeline that incorporates scenario decomposition, document generation, function integration, complexity scaling, and localized deployment to enable high-quality training environments without external tools.
Outcome:	The proposed framework significantly improves the models’ tool-use performance without degrading their general capabilities.

Copied to clipboard

Xiangyang Liu, Tianxiang Sun, Xuanjing Huang, Xipeng Qiu

Watermarking PLMs on Classification Tasks by Combining Contrastive Learning with Weight Perturbation (2023.findings-emnlp)

Challenge:	Prompt tuning is parameter-efficient but lags behind other state-of-the-art methods.
Approach:	They propose a parameter-efficient tuning method that only optimizes a soft prompt to adapt PTMs to downstream tasks.
Outcome:	The proposed method is parameter-efficient but lags behind other state-of-the-art methods.

Copied to clipboard

Chenxi Gu, Xiaoqing Zheng, Jianhan Xu, Muling Wu, Cenyuan Zhang, Chengsong Huang, Hua Cai, Xuanjing Huang

Mitigating Hallucinations in VLMs: Enhancing Visual Attention via Head-Wise Perturbation (2026.findings-acl)

Challenge:	Large pre-trained language models (PLMs) are highly valuable intellectual property due to their expensive training costs.
Approach:	They propose to embed backdoors that can be triggered by specific inputs into models by model watermarking.
Outcome:	The proposed method can be used to protect the intellectual property of large pre-trained language models without knowledge about downstream tasks.

Copied to clipboard

Zhenghua Wang, Yixin Wu, Feiran Zhang, Qi Qian, Changze Lv, Xuanjing Huang, Xiaoqing Zheng

Leveraging Declarative Knowledge in Text and First-Order Logic for Fine-Grained Propaganda Detection (2020.emnlp-main)

Challenge:	Vision–Language Models (VLMs) have demonstrated strong capabilities in tasks that require joint understanding of text and images.
Approach:	They propose a strategy that incorporates head-wise attention perturbation via continuous multiplicative noise coupled with a visual-guided loss focused on vision-sensitive text tokens to promote a more balanced attention distribution.
Outcome:	The proposed approach outperforms baseline models on three benchmarks and consistently outperformed the baseline model.

Copied to clipboard

Ruize Wang, Duyu Tang, Nan Duan, Wanjun Zhong, Zhongyu Wei, Xuanjing Huang, Daxin Jiang, Ming Zhou

Challenge:	Existing methods for fine-grained propaganda detection are not based on input-output data, but instead use declarative knowledge to detect propagandistic text fragments.
Approach:	They propose a method to inject declarative knowledge of fine-grained propaganda techniques into training data to get better representations of propagandistic texts.
Outcome:	The proposed method achieves superior performance on a large dataset for propaganda detection.

Loose lips sink ships: Mitigating Length Bias in Reinforcement Learning from Human Feedback (2023.findings-emnlp)

Copied to clipboard

Wei Shen, Rui Zheng, Wenyu Zhan, Jun Zhao, Shihan Dou, Tao Gui, Qi Zhang, Xuanjing Huang

From Scores to Preferences: Redefining Evaluation Paradigm for Speech Quality Reward Modeling (2026.findings-acl)

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.

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

Template-free Prompt Tuning for Few-shot NER (2022.naacl-main)

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.

Copied to clipboard

Ruotian Ma, Xin Zhou, Tao Gui, Yiding Tan, Linyang Li, Qi Zhang, Xuanjing Huang

RethinkCWS: Is Chinese Word Segmentation a Solved Task? (2020.emnlp-main)

Challenge:	Prompt-based methods have been successfully applied in few-shot learning tasks . however, when applied to token-level labeling tasks, it would be time-consuming to enumerate the template queries over all potential entity spans.
Approach:	They propose a method to reformulate NER tasks as LM problems without templates.
Outcome:	The proposed method is 30.12 times faster than the template-based method under few-shot settings.

Copied to clipboard

Jinlan Fu, Pengfei Liu, Qi Zhang, Xuanjing Huang

Uncertainty-Aware Label Refinement for Sequence Labeling (2020.emnlp-main)

Challenge:	Recent years have seen remarkable success in the use of deep neural networks on Chinese word segmentation (CWS) however, the performance of CWS systems has gradually reached a plateau with the rapid development of deep networks.
Approach:	They propose a fine-grained evaluation for existing Chinese word segmentation systems that allows us to diagnose the strengths and weaknesses of existing models.
Outcome:	The proposed model can diagnose strengths and weaknesses of existing models and alleviate negative transfer problem when doing multi-criteria learning.

Copied to clipboard

Tao Gui, Jiacheng Ye, Qi Zhang, Zhengyan Li, Zichu Fei, Yeyun Gong, Xuanjing Huang

Green & Sustainable NLP

Challenge:	Conditional random fields (CRF) for label decoding have been a problem for many tasks.
Approach:	They propose a two-stage label decoding framework that model long-term label dependencies while being much more computationally efficient.
Outcome:	The proposed method outperforms the CRF-based methods and greatly accelerates the inference process.

ToolHop: A Query-Driven Benchmark for Evaluating Large Language Models in Multi-Hop Tool Use (2025.acl-long)

Copied to clipboard

Junjie Ye, Zhengyin Du, Xuesong Yao, Weijian Lin, Yufei Xu, Zehui Chen, Zaiyuan Wang, Sining Zhu, Zhiheng Xi, Siyu Yuan, Tao Gui, Qi Zhang, Xuanjing Huang, Jiecao Chen

A Lexicon-Based Supervised Attention Model for Neural Sentiment Analysis (C18-1)

Challenge:	Effective evaluation of multi-hop tool use is critical for analyzing the understanding, reasoning, and function-calling capabilities of large language models.
Approach:	They propose a dataset that provides rigorous evaluation of multi-hop tool use.
Outcome:	The proposed model achieves 49.04% accuracy across five model families.

Copied to clipboard

Yicheng Zou, Tao Gui, Qi Zhang, Xuanjing Huang

Mitigating Object Hallucinations in MLLMs via Multi-Frequency Perturbations (2025.findings-emnlp)

Challenge:	Existing attention models do not take full advantage of sentiment lexicons, which provide rich sentiment information and play a critical role in sentiment analysis.
Approach:	They propose a lexicon-based supervised attention model which allows a neural network to focus on the sentiment content, thus generating sentiment-informative representations.
Outcome:	The proposed model outperforms existing models on three large-scale sentiment classification datasets.

Copied to clipboard

Shuo Li, Jiajun Sun, Guodong Zheng, Xiaoran Fan, Yujiong Shen, Yi Lu, Zhiheng Xi, Yuming Yang, Wenming Tan, Tao Ji, Tao Gui, Qi Zhang, Xuanjing Huang

Towards Adversarially Robust Text Classifiers by Learning to Reweight Clean Examples (2022.findings-acl)

Challenge:	Multimodal large language models have demonstrated remarkable performance in visual-language tasks, but their authenticity is often compromised by object hallucinations.
Approach:	They propose a multi-frequency perturbation method that leverages both low-frequency and high-frequency features of images to perturb visual feature representations and explicitly suppress redundant frequency-domain features during inference.
Outcome:	The proposed method significantly mitigates object hallucinations across various model architectures.

Copied to clipboard

Jianhan Xu, Cenyuan Zhang, Xiaoqing Zheng, Linyang Li, Cho-Jui Hsieh, Kai-Wei Chang, Xuanjing Huang

Distill Visual Chart Reasoning Ability from LLMs to MLLMs (2025.findings-emnlp)

Challenge:	Existing defense methods improve the adversarial robustness by making models adapt to training set augmented with some adversarials.
Approach:	They propose to introduce a reweighting mechanism to calibrate the training distribution to obtain robust models.
Outcome:	The proposed method minimizes the loss of validation set mixed with clean examples and adversarial ones in an online learning manner.

Copied to clipboard

Wei He, Zhiheng Xi, Wanxu Zhao, Xiaoran Fan, Yiwen Ding, Zifei Shan, Tao Gui, Qi Zhang, Xuanjing Huang

KNSE: A Knowledge-aware Natural Language Inference Framework for Dialogue Symptom Status Recognition (2023.findings-acl)

Challenge:	a new method for generating chart annotations is proposed to improve visual reasoning in multimodal large language models.
Approach:	They propose a code-as-intermediary translation method for distilling visual reasoning abilities from LLMs to MLLMs.
Outcome:	The proposed method is cost-effective, efficient and scalable.

Copied to clipboard

Wei Chen, Shiqi Wei, Zhongyu Wei, Xuanjing Huang

Natural Language Interfaces

Challenge:	Existing frameworks for symptom status recognition in doctor-patient dialogues are inadequate.
Approach:	They propose a framework for symptom status recognition that formalizes a natural language inference task . they generate knowledge about the symptom and a hypothesis about its status for each symptom .
Outcome:	The proposed framework outperforms baselines and has advantages in cross-disease and cross-symptom scenarios.

Automatic Term Name Generation for Gene Ontology: Task and Dataset (2020.findings-emnlp)

Copied to clipboard

Yanjian Zhang, Qin Chen, Yiteng Zhang, Zhongyu Wei, Yixu Gao, Jiajie Peng, Zengfeng Huang, Weijian Sun, Xuanjing Huang

Measure Children’s Mindreading Ability with Machine Reading (2023.findings-emnlp)

Challenge:	Gene Ontology (GO) terms are used to describe gene function in biology and bio-medicine.
Approach:	They propose a task to generate term names for GO and build a large-scale benchmark dataset.
Outcome:	The proposed model outperforms baselines by incorporating the relations between genes, words and terms for term name generation.

Copied to clipboard

Yuliang Yan, Xiaohua Wang, Xiang Zhou, Xiaoqing Zheng, Xuanjing Huang

Automatic Essay Scoring Incorporating Rating Schema via Reinforcement Learning (D18-1)

Challenge:	Existing scoring models do not take the features of the stories and video clips into account when scoring, which will reduce the accuracy of the models.
Approach:	They propose to leverage the features extracted from stories and videos related to the questions being asked during the children’s mindreading evaluation.
Outcome:	The proposed framework agrees well with human experts on scores produced by the models.

Copied to clipboard

Yucheng Wang, Zhongyu Wei, Yaqian Zhou, Xuanjing Huang

Challenge:	Existing systems for automatic essay scoring are trained to predict the score of each essay at a time without considering rating schema.
Approach:	They propose a reinforcement learning framework that incorporates quadratic weighted kappa as guidance to optimize the scoring system.
Outcome:	Experiments on benchmark datasets show the proposed framework is effective.

Cross-Linguistic Syntactic Difference in Multilingual BERT: How Good is It and How Does It Affect Transfer? (2022.emnlp-main)

Copied to clipboard

Ningyu Xu, Tao Gui, Ruotian Ma, Qi Zhang, Jingting Ye, Menghan Zhang, Xuanjing Huang

Scaling Laws for Fact Memorization of Large Language Models (2024.findings-emnlp)

Challenge:	Multilingual BERT (mBERT) has demonstrated considerable cross-lingual syntactic ability, but it is not well understood what leads to this variation and whether it fairly reflects difference between languages.
Approach:	They propose to use multilingual BERT to enable zero-shot cross-lingual transfer of syntactic knowledge between different languages by generating grammatical relations in 24 different languages.
Outcome:	The results show that the distance between the distributions of different languages is highly consistent with the syntactic difference in terms of linguistic formalisms.

Copied to clipboard

Xingyu Lu, Xiaonan Li, Qinyuan Cheng, Kai Ding, Xuanjing Huang, Xipeng Qiu

EvoWiki: Evaluating LLMs on Evolving Knowledge (2025.acl-long)

Challenge:	Fact knowledge memorization is crucial for Large Language Models (LLMs) to generate factual and reliable responses.
Approach:	They analyze scaling laws for LLM’s fact knowledge and LLMs’ behaviors of memorizing different types of facts.
Outcome:	The proposed model can generalize on unseen facts and its scaling law is similar to general pre-training.

Copied to clipboard

Wei Tang, Yixin Cao, Yang Deng, Jiahao Ying, Bo Wang, Yizhe Yang, Yuyue Zhao, Qi Zhang, Xuanjing Huang, Yu-Gang Jiang, Yong Liao

Challenge:	Existing knowledge evolution benchmarks are static and fail to capture the evolving nature of LLMs and knowledge.
Approach:	They propose an evolving dataset that categorizes information into stable, evolved, and uncharted states.
Outcome:	The proposed dataset is auto-updatable and enables evaluation of continuously changing knowledge and newly released LLMs.

A Progressive Framework for Role-Aware Rumor Resolution (2022.coling-1)

Copied to clipboard

Lei Chen, Guanying Li, Zhongyu Wei, Yang Yang, Baohua Zhou, Qi Zhang, Xuanjing Huang

Challenge:	Existing methods for rumor resolution ignore intrinsic propagation mechanisms of rumors and present poor adaptive ability when unprecedented news emerges.
Approach:	They propose to identify triggering posts and exploit their characteristics to facilitate rumor verification.
Outcome:	The proposed model and scheme exploits rumor diffusion patterns and linguistic features to facilitate verification.

TextFlint: Unified Multilingual Robustness Evaluation Toolkit for Natural Language Processing (2021.acl-demo)

Copied to clipboard

Xiao Wang, Qin Liu, Tao Gui, Qi Zhang, Yicheng Zou, Xin Zhou, Jiacheng Ye, Yongxin Zhang, Rui Zheng, Zexiong Pang, Qinzhuo Wu, Zhengyan Li, Chong Zhang, Ruotian Ma, Zichu Fei, Ruijian Cai, Jun Zhao, Xingwu Hu, Zhiheng Yan, Yiding Tan, Yuan Hu, Qiyuan Bian, Zhihua Liu, Shan Qin, Bolin Zhu, Xiaoyu Xing, Jinlan Fu, Yue Zhang, Minlong Peng, Xiaoqing Zheng, Yaqian Zhou, Zhongyu Wei, Xipeng Qiu, Xuanjing Huang

Tasty Burgers, Soggy Fries: Probing Aspect Robustness in Aspect-Based Sentiment Analysis (2020.emnlp-main)

Challenge:	Existing approaches to textual robustness evaluation focus on slightly modifying the input data, which maintains the original meaning and results in a different prediction.
Approach:	They propose a multilingual robustness evaluation toolkit for NLP that integrates universal text transformations, task-specific transformations and adversarial attack.
Outcome:	The toolkit includes universal text transformation, task-specific transformation, adversarial attack, subpopulation, and their combinations to provide comprehensive robustness analyses.

Copied to clipboard

Xiaoyu Xing, Zhijing Jin, Di Jin, Bingning Wang, Qi Zhang, Xuanjing Huang

Sentiment Analysis

Challenge:	Existing ABSA test sets cannot be used to distinguish the sentiment of the target aspect from the non-target aspect.
Approach:	They propose a simple but effective approach to enrich ABSA test sets by disentangle the confounding sentiments of non-target aspects from the target aspect’s sentiment.
Outcome:	The proposed model can distinguish the sentiment of the non-target aspects from the target aspect’s sentiment by using the Aspect Robustness Test Set (ARTS).

Improving Continual Pre-training Through Seamless Data Packing (2025.findings-acl)

Copied to clipboard

Ruicheng Yin, Xuan Gao, Changze Lv, Xiaohua Wang, Xiaoqing Zheng, Xuanjing Huang

Aggregation of Reasoning: A Hierarchical Framework for Enhancing Answer Selection in Large Language Models (2024.lrec-main)

Challenge:	Empirical evaluations across various model architectures and corpus domains demonstrate the effectiveness of our method, outperforming baselines in 99% of all settings.
Approach:	They propose a method that uses a sliding window technique to pack data before continual pre-training to preserve contextual information and enhance model performance.
Outcome:	Empirical evaluations across various model architectures and corpus domains demonstrate the effectiveness of the proposed method outperforming baselines in 99% of settings.

Copied to clipboard

Zhangyue Yin, Qiushi Sun, Qipeng Guo, Zhiyuan Zeng, Xiaonan Li, Tianxiang Sun, Cheng Chang, Qinyuan Cheng, Ding Wang, Xiaofeng Mou, Xipeng Qiu, Xuanjing Huang

Case2Code: Scalable Synthetic Data for Code Generation (2025.coling-main)

Challenge:	Recent advances in Chain-of-Thought prompting have facilitated significant breakthroughs for Large Language Models (LLMs) in complex reasoning tasks.
Approach:	They propose a hierarchical reasoning aggregation framework to address this problem . they propose dynamic sampling to adjust the number of reasoning chains .
Outcome:	The proposed framework outperforms existing ensemble methods on complex reasoning tasks.

Copied to clipboard

Yunfan Shao, Linyang Li, Yichuan Ma, Peiji Li, Demin Song, Qinyuan Cheng, Shimin Li, Xiaonan Li, Pengyu Wang, Qipeng Guo, Hang Yan, Xipeng Qiu, Xuanjing Huang, Dahua Lin

Programming Languages in NLP

Challenge:	Large Language Models (LLMs) have shown outstanding breakthroughs in code generation.
Approach:	They propose a case-to-code induction task that exploits the expressiveness and correctness of programs by incorporating LLMs into their training.
Outcome:	The proposed task improves distribution case-to-code induction and various coding generation tasks.

LoRACoE: Improving Large Language Model via Composition-based LoRA Expert (2025.emnlp-main)

Copied to clipboard

Guanyu Li, Zhiheng Xi, Zhihao Zhang, Boyang Hong, Tao Gui, Qi Zhang, Xuanjing Huang

Beyond Scaling: Measuring and Predicting the Upper Bound of Knowledge Retention in Language Model Pre-Training (2026.acl-long)

Challenge:	Recent studies show that the Mixture of Experts architecture improves performance of large language models.
Approach:	They propose a method to build static experts using LoRA parameters . they propose to use rank-level parameters to build experts based on rank-based parameters based in LoRA module.
Outcome:	The proposed method improves task performance across a broader range of tasks.

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

Promoting Data and Model Privacy in Federated Learning through Quantized LoRA (2024.findings-emnlp)

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.

Copied to clipboard

Zhu JianHao, Changze Lv, Xiaohua Wang, Muling Wu, Wenhao Liu, Tianlong Li, Zixuan Ling, Cenyuan Zhang, Xiaoqing Zheng, Xuanjing Huang

CodeIE: Large Code Generation Models are Better Few-Shot Information Extractors (2023.acl-long)

Challenge:	Existing federated learning frameworks require substantial data and computational resources to develop large language models.
Approach:	They propose a method that distributes a quantized version of the model’s parameters during training and combine it with a popular fine-tuning method to significantly reduce communication costs.
Outcome:	The proposed method enables accurate estimations for parameter updates while preventing clients from accessing a model whose performance is comparable to the centrally hosted one.

Copied to clipboard

Peng Li, Tianxiang Sun, Qiong Tang, Hang Yan, Yuanbin Wu, Xuanjing Huang, Xipeng Qiu

TextObfuscator: Making Pre-trained Language Model a Privacy Protector via Obfuscating Word Representations (2023.findings-acl)

Challenge:	Large language models pre-trained on massive corpora have shown impressive few-shot learning ability on many NLP tasks.
Approach:	They propose to recast structured output in the form of code instead of natural language and use generative LLMs of code to perform IE tasks.
Outcome:	The proposed method outperforms fine-tuning moderate-size pre-trained models and prompting NL-LLMs under few-shot settings.

Copied to clipboard

Xin Zhou, Yi Lu, Ruotian Ma, Tao Gui, Yuran Wang, Yong Ding, Yibo Zhang, Qi Zhang, Xuanjing Huang

Navigating the OverKill in Large Language Models (2024.acl-long)

Challenge:	Existing inference services are plagued by privacy concerns, such as sharing sensitive data with service providers.
Approach:	They propose a framework for protecting inference privacy by applying random perturbations to clustered representations.
Outcome:	The proposed framework protects inference privacy by applying random perturbations to clustered representations.

Copied to clipboard

Chenyu Shi, Xiao Wang, Qiming Ge, Songyang Gao, Xianjun Yang, Tao Gui, Qi Zhang, Xuanjing Huang, Xun Zhao, Dahua Lin

FLAT: Chinese NER Using Flat-Lattice Transformer (2020.acl-main)

Challenge:	Recent studies have highlighted a tendency among large language models to refuse to answer benign queries.
Approach:	They propose a model-agnostic approach to reduce excessive attention to harmful words like ‘kill’ and a method to decode the next-token predictions by contrastive decoding.
Outcome:	The proposed approach reduces the refusal rate by 20% while having little impact on safety.

Copied to clipboard

Xiaonan Li, Hang Yan, Xipeng Qiu, Xuanjing Huang