Papers by Ding Zhao

DevEval: A Manually-Annotated Code Generation Benchmark Aligned with Real-World Code Repositories (2024.findings-acl)

Challenge:	Existing methods for intent classification fail to distinguish new intents due to intertwined centers . a novel framework that learns geometry-aware representations to maximally separate all intents is proposed .
Approach:	They propose a new intent discovery framework that learns geometry-aware representations to maximally separate all intents.
Outcome:	The proposed framework achieves a new state-of-the-art performance on three benchmarking datasets.

Copied to clipboard

Jia Li, Ge Li, Yunfei Zhao, Yongmin Li, Huanyu Liu, Hao Zhu, Lecheng Wang, Kaibo Liu, Zheng Fang, Lanshen Wang, Jiazheng Ding, Xuanming Zhang, Yuqi Zhu, Yihong Dong, Zhi Jin, Binhua Li, Fei Huang, Yongbin Li, Bin Gu, Mengfei Yang

OpenPrompt: An Open-source Framework for Prompt-learning (2022.acl-demo)

Challenge:	Existing benchmarks are poorly aligned with real-world code repositories and are insufficient to evaluate the coding abilities of Large Language Models (LLMs).
Approach:	They propose a repository-level benchmark named DevEval to evaluate LLMs' coding abilities in real-world code repositories.
Outcome:	The proposed benchmarks show that the LLMs perform better in real-world code repositories than existing benchmarks.

Copied to clipboard

Ning Ding, Shengding Hu, Weilin Zhao, Yulin Chen, Zhiyuan Liu, Haitao Zheng, Maosong Sun

Embodied Executable Policy Learning with Language-based Scene Summarization (2024.naacl-long)

Challenge:	Prompt-learning is a new paradigm in natural language processing, adapting pre-trained language models to cloze-style prediction, autoregressive modeling, or sequence to sequence generation.
Approach:	They propose a framework for prompt-learning that integrates pre-trained language models with a unified framework.
Outcome:	The proposed framework is easy to use and flexible enough to integrate with other frameworks.

Copied to clipboard

Jielin Qiu, Mengdi Xu, William Han, Seungwhan Moon, Ding Zhao

B-APO: Bias-Targeted Adversarial Preference Optimization for Debiasing Multimodal Large Language Models (2026.findings-acl)

Challenge:	Existing Large Language models with text inputs lack the capability to evolve with non-expert interactions with environments.
Approach:	They propose a novel learning paradigm that generates robots’ executable actions in the form of text, derived solely from visual observations.
Outcome:	The proposed learning paradigm surpasses baselines and can adapt to the target tasks effectively.

Copied to clipboard

Pinlong Zhao, Zike Ding, Zengshu Ye, Zhou Zhaoting

Robustness in NLP

Challenge:	Existing debiasing methods create biased responses by completely removing an entire modality, forming an extreme and static training environment.
Approach:	They propose a method to debiase multimodal large language models by masking one modality and then enlarge the margin between clean and adversarial responses.
Outcome:	The proposed method achieves superior debiasing performance while maintaining general capabilities.

CMR Scaling Law: Predicting Critical Mixture Ratios for Continual Pre-training of Language Models (2024.emnlp-main)

Copied to clipboard

Jiawei Gu, Zacc Yang, Chuanghao Ding, Rui Zhao, Fei Tan

Forging Multiple Training Objectives for Pre-trained Language Models via Meta-Learning (2022.findings-emnlp)

Challenge:	Large Language Models (LLMs) excel in diverse tasks but often underperform in specialized fields due to limited domain-specific or proprietary corpus.
Approach:	They propose a power-law relationship between loss, mixture ratio, and training tokens scale and formalize the trade-off between general and domain-specific capabilities.
Outcome:	The proposed model achieves the desired domain transfer while maintaining general ability and highest utilization of available resources.

Copied to clipboard

Hongqiu Wu, Ruixue Ding, Hai Zhao, Boli Chen, Pengjun Xie, Fei Huang, Min Zhang

EPiDA: An Easy Plug-in Data Augmentation Framework for High Performance Text Classification (2022.naacl-main)

Challenge:	Empirical studies show that learning multiple training objectives in a single model makes the learned language representation barely converge to the desired optimum.
Approach:	They propose a meta-learning-based adaptive sampler which learns latent sampling pattern on arbitrary pre-training objectives.
Outcome:	Empirical studies show that learning multiple objectives in a single model makes it difficult to achieve the desired optimum.

Copied to clipboard

Minyi Zhao, Lu Zhang, Yi Xu, Jiandong Ding, Jihong Guan, Shuigeng Zhou

Retrieving Multimodal Information for Augmented Generation: A Survey (2023.findings-emnlp)

Challenge:	Existing methods for data augmentation do not fully exploit the potential of DA in NLP.
Approach:	They propose an easy and plug-in framework for data augmentation to support effective text classification.
Outcome:	The proposed framework outperforms existing methods in most cases, but not using agent networks or pre-trained generation networks.

Copied to clipboard

Ruochen Zhao, Hailin Chen, Weishi Wang, Fangkai Jiao, Xuan Long Do, Chengwei Qin, Bosheng Ding, Xiaobao Guo, Minzhi Li, Xingxuan Li, Shafiq Joty

3DS: Medical Domain Adaptation of LLMs via Decomposed Difficulty-based Data Selection (2025.emnlp-main)

Challenge:	Large Language Models (LLMs) are increasingly using multimodality to augment their generation ability, but there is no unified perception of at which stage and how to incorporate different modalities.
Approach:	They propose to use multimodality to augment Large Language Models (LLMs) this will provide scholars with a deeper understanding of the methods' applications and encourage them to adapt existing techniques to the fast-growing field of LLMs.
Outcome:	The proposed methods improve factuality, reasoning, interpretability, and robustness of the generated content.

Copied to clipboard

Hongxin Ding, Yue Fang, Runchuan Zhu, Xinke Jiang, Jinyang Zhang, Yongxin Xu, Weibin Liao, Xu Chu, Junfeng Zhao, Yasha Wang

RAG-on-a-Diet: A Reinforcement Learning-Based Dynamic Resource Optimization Framework for RAG (2026.acl-long)

Challenge:	Effective domain adaptation typically involves supervised fine-tuning on carefully selected instruction-tuned data.
Approach:	They propose a model-centric data selection framework that aligns data selection with the model’s knowledge distribution to improve model performance.
Outcome:	The proposed framework outperforms existing methods by up to 2.97% accuracy in the healthcare domain.

Copied to clipboard

Hongwen Ding, Yizheng Zhao

Group-wise Contrastive Learning for Neural Dialogue Generation (2020.findings-emnlp)

Challenge:	Existing frameworks for knowledge-intensive multi-hop question answering do not adapt to how a trajectory unfolds.
Approach:	They propose a lightweight reinforcement-learning agent that treats each reasoning hop as an independent decision and selects the smallest model sufficient for it.
Outcome:	The proposed agent cuts Monetary Inference Cost by 60.07% against IRCoT with only a 3.7% F1 drop and matches Adaptive-RAG’s F1 at 37.30% lower cost.

Copied to clipboard

Hengyi Cai, Hongshen Chen, Yonghao Song, Zhuoye Ding, Yongjun Bao, Weipeng Yan, Xiaofang Zhao

Consolidation or Adaptation? PRISM: Disentangling SFT and RL Data via Gradient Concentration (2026.acl-long)

Challenge:	Existing approaches to training dialogue models have low diversity in open-domain contexts . prior art suggests that naive MLE objective is not effective enough .
Approach:	They propose to incorporate contrastive learning into dialogue generation by using a pretrained baseline model as a reference.
Outcome:	The proposed framework is suited for training a wide range of dialogue generation models with favorable performance over baseline training approaches.

Copied to clipboard

Yang Zhao, Yangou Ouyang, Xiao Ding, Hepeng Wang, Bibo Cai, Kai Xiong, Jinglong Gao, Zhouhao Sun, Li Du, Bing Qin, Ting Liu

AssistedDS: Benchmarking How External Domain Knowledge Assists LLMs in Automated Data Science (2025.findings-emnlp)

Challenge:	Existing data arbitration strategies for large language model training rely on surface-level heuristics that fail to diagnose intrinsic learning needs.
Approach:	They propose a framework that arbitrates data based on its degree of cognitive conflict with the model's existing knowledge.
Outcome:	Extensive experiments on WebShop and ALFWorld show that PRISM outperforms state-of-the-art hybrid methods while reducing computational costs by up to 3.22 .

Copied to clipboard

An Luo, Xun Xian, Jin Du, Fangqiao Tian, Ganghua Wang, Ming Zhong, Shengchun Zhao, Xuan Bi, Zirui Liu, Jiawei Zhou, Jayanth Srinivasa, Ashish Kundu, Charles Fleming, Mingyi Hong, Jie Ding

Robustness in NLP

Challenge:	Large language models (LLMs) have advanced the automation of data science workflows, yet it remains unclear whether they can critically leverage external domain knowledge as human data scientists do in practice.
Approach:	They propose a benchmark to evaluate how large language models handle external domain knowledge in tabular prediction tasks.
Outcome:	The proposed model evaluates whether it can critically leverage external domain knowledge as human data scientists do in practice.

OpenDelta: A Plug-and-play Library for Parameter-efficient Adaptation of Pre-trained Models (2023.acl-demo)

Copied to clipboard

Shengding Hu, Ning Ding, Weilin Zhao, Xingtai Lv, Zhen Zhang, Zhiyuan Liu, Maosong Sun

Yang Zhao, Li Du, Xiao Ding, Yangou Ouyang, Hepeng Wang, Kai Xiong, Jinglong Gao, Zhouhao Sun, Dongliang Xu, Qing Yang, Dongchen Li, Bing Qin, Ting Liu

Challenge:	Existing implementations that modify the code of the backbone PTMs and hard-code specific delta tuning methods for each PTM have limited the practicality and flexibility of delta tuning.
Approach:	They propose an open-source library that provides a plug-and-play implementation of delta tuning methods for pre-trained models.
Outcome:	The proposed methods eliminate the need to modify the backbone PTMs’ code, making OpenDelta compatible with different, even novel PTM.

Copied to clipboard

Threshold Differential Attention for Sink-Free, Ultra-Sparse, and Non-Dispersive Language Modeling (2026.acl-long)

Challenge:	Existing methods for selecting training data from general datasets fail to account for the joint distribution of instructions, resulting in inefficient learning and suboptimal knowledge transfer.
Approach:	They propose a method that constructs a mixed gradient-based instruction graph to capture the joint distribution and interdependencies among instructions.
Outcome:	The proposed method outperforms existing methods on domain adaptation tasks and in complex, data-scarce scenarios.

Copied to clipboard

Xingyue Huang, Xueying Ding, Mingxuan Ju, Yozen Liu, Neil Shah, Tong Zhao

Safety is Not Only About Refusal: Reasoning-Enhanced Fine-tuning for Interpretable LLM Safety (2025.findings-acl)

Challenge:	a strict sum-to-one constraint forces attention sinks on irrelevant tokens, while probability mass disperses as sequence lengths increase.
Approach:	They propose a sink-free attention mechanism that achieves ultra-sparsity and improved robustness at longer sequence lengths without the computational overhead of projection methods.
Outcome:	The proposed mechanism produces >99 % exact zeros and eliminates attention sinks while maintaining competitive performance on standard and long-context benchmarks.

Copied to clipboard

Yuyou Zhang, Miao Li, William Han, Yihang Yao, Zhepeng Cen, Ding Zhao

HyKGE: A Hypothesis Knowledge Graph Enhanced RAG Framework for Accurate and Reliable Medical LLMs Responses (2025.acl-long)

Challenge:	Large Language Models (LLMs) are vulnerable to jailbreak attacks that exploit weaknesses in traditional safety alignment.
Approach:	They propose a framework that trains models to engage in explicit safe reasoning before response . they propose RATIONAL, which allows models to reject harmful prompts while providing meaningful and context-aware responses.
Outcome:	The proposed framework fine-tunes models to reason about query intent, ethics, and potential harm.

Copied to clipboard

Xinke Jiang, Ruizhe Zhang, Yongxin Xu, Rihong Qiu, Yue Fang, Zhiyuan Wang, Jinyi Tang, Hongxin Ding, Xu Chu, Junfeng Zhao, Yasha Wang

Knowledge Representation

Challenge:	Recent approaches suffer from insufficient and repetitive knowledge retrieval, tedious and time-consuming query parsing, and monotonous knowledge utilization.
Approach:	They propose a retrieval-augmented generation framework which leverages LLMs’ powerful reasoning capacity to compensate for the incompleteness of user queries.
Outcome:	The proposed framework improves the accuracy and reliability of Large Language Models (LLMs) by combining the rich knowledge of LLMs with Hypothesis Outputs.

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

Copied to clipboard

Deming Ding, Shichun Liu, Enhui Yang, Jiahang Lin, Ziying Chen, Shihan Dou, Honglin Guo, Weiyu Cheng, Pengyu Zhao, Chengjun Xiao, Qunhong Zeng, Qi Zhang, Xuanjing Huang, Qidi Xu, Tao Gui

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.

A Novel Matching Paradigm: Unified Generative and Discriminative LLM with Prompt Compression for Relevance Learning (2026.acl-industry)

Copied to clipboard

Guoliang Zhao, Zixin Cui, Chao Ye, Dengwu He, Fei Huang, Yubo Liu, Shuanglong Li, Tzungren Kuo, Bin Ding, Shuang Zhang, null KunhongZhu, Zhi Guo, Liu Lin

DUAL RM: Beyond Rule-based Preference Reward Modeling via Meta-Reward (2026.acl-long)

Challenge:	Existing approaches to matching use Large Language Models as feature extractors, underutilizing their full modeling capabilities.
Approach:	They propose a matching paradigm that integrates two-tower, single-towing, and generative tasks within a unified LLM framework via attention-mask partitioning.
Outcome:	The proposed model achieves superior performance and strong practical value in an industrial search engine.

Copied to clipboard

Xiaobo Liang, Wanfu Wang, Qipeng Huang, Yuyang Ding, Zecheng Tang, Yixin Ji, Qianben Chen, Zhe Zhao, Kehai Chen, Juntao Li, Min Zhang

Challenge:	Existing preference-based reward modeling methods face a recursive dependency where each verifier requires a meta-verifier, leading to continuous and costly dependence on human annotation.
Approach:	They propose a dual RM that couples discriminative and generative reward models under a non-parametric meta-reward.
Outcome:	The proposed model achieves strong performance across major preference benchmarks and even when trained exclusively on language modality, it exhibits robust cross-modal transfer on Omni-RewardBench.

Parallelism and Generation Order in Masked Diffusion Language Models: Limits Today, Potential Tomorrow (2026.findings-acl)

Copied to clipboard

Yangyang Zhong, Yanmei Gu, Zhengqing Zang, Xiaomeng Li, Yuqi Ding, Xibei Jia, Yuting Shen, Zhenzhong Lan, Liwang Zhu, Weiping Liu, Junlin Zhou, Haisheng Liu, Zhong Xin Yu, Pengxin Luo, Donglian Qi, Yunfeng Yan, Junbo Zhao

ITAKE: Interactive Unstructured Text Annotation and Knowledge Extraction System with LLMs and ModelOps (2024.acl-demos)

Challenge:	Autoregressive (AR) language models dominate modern natural language processing due to strong likelihood-based training objectives and reliable left-to-right decoding.
Approach:	They characterize MDLM behavior along two dimensions: parallelism strength and generation order . authors propose a Generate-then-Edit paradigm that mitigates dependency loss .
Outcome:	The proposed model improves on tasks that require "backward information" the Generate-then-Edit paradigm improves parallel decoding efficiency while reducing dependency loss.

Copied to clipboard

Jiahe Song, Hongxin Ding, Zhiyuan Wang, Yongxin Xu, Yasha Wang, Junfeng Zhao

On LLMs-Driven Synthetic Data Generation, Curation, and Evaluation: A Survey (2024.findings-acl)

Challenge:	Unstructured text data contains a large amount of valuable knowledge, but there are many tools that do not meet the needs of actual business.
Approach:	They propose an unstructured text annotation and knowledge extraction system that integrates Large Language Models and ModelOps to improve model supervision and performance.
Outcome:	The proposed system integrates large language models and ModelOps to improve performance in low-resource contexts.

Copied to clipboard

Lin Long, Rui Wang, Ruixuan Xiao, Junbo Zhao, Xiao Ding, Gang Chen, Haobo Wang

Challenge:	Large Language Models (LLMs) provide a data-centric solution to alleviate limitations of real-world data with synthetic data generation.
Approach:	They propose a generic workflow for LLM-driven synthetic data generation.
Outcome:	The proposed workflows highlight gaps in existing research and outline avenues for future studies.

A Simple and Effective Method to Improve Zero-Shot Cross-Lingual Transfer Learning (2022.coling-1)

Copied to clipboard

Kunbo Ding, Weijie Liu, Yuejian Fang, Weiquan Mao, Zhe Zhao, Tao Zhu, Haoyan Liu, Rong Tian, Yiren Chen

Hierarchical Token Prepending: Enhancing Information Flow in Decoder-based LLM Embeddings (2026.acl-long)

Challenge:	Existing zero-shot cross-lingual transfer methods rely on parallel corpora or bilingual dictionaries . however, its effect is limited by the gap between embedding clusters of different languages .
Approach:	They propose Embedding-Push, Attention-Pull, and Robust targets to transfer English embeddings to virtual multilingual embedders without semantic loss.
Outcome:	Experimental results show that the proposed method outperforms existing methods on cross-lingual tasks and can achieve a better multilingual alignment.

Copied to clipboard

Xueying Ding, Xingyue Huang, Mingxuan Ju, Liam Collins, Yozen Liu, Leman Akoglu, Neil Shah, Tong Zhao

Fusing Highly Specialized Language Models for Comprehensive Expertise (2025.acl-long)

Challenge:	Large language models produce powerful text embeddings, but their causal attention mechanism restricts the flow of information from later to earlier tokens, harming performance.
Approach:	They propose a method that prepending a single summary token to reduce attention-level compression by partitioning the input into blocks and prepending blocks to subsequent blocks.
Outcome:	The proposed method achieves consistent performance gains across 11 retrieval datasets and 30 general embedding benchmarks.

Copied to clipboard

Ning Ding, Yulin Chen, Ganqu Cui, Xingtai Lv, Weilin Zhao, Kaiyan Zhang, Ruobing Xie, Bowen Zhou, Zhiyuan Liu, Maosong Sun

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

Challenge:	Existing models that focus on language, programming code, and mathematical symbols are not able to achieve mastery of all three domains simultaneously.
Approach:	They propose to fuse highly-specialized models that are already sufficiently trained on different domains to achieve a highly-specific model.
Outcome:	The proposed model could achieve mastery of the three crucial domains simultaneously.

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.

Remember Me, Refine Me: A Dynamic Procedural Memory Framework for Experience-Driven Agent Evolution (2026.findings-acl)

Copied to clipboard

Zouying Cao, Jiaji Deng, Li Yu, Weikang Zhou, Zhaoyang Liu, Bolin Ding, Hai Zhao

Green & Sustainable NLP

Challenge:	Existing frameworks treat memory as a static append-only archive . Existing systems focus on passive accumulation, resulting in a 'passive accumulation' of memory.
Approach:	They propose a framework for experience-driven agent evolution that integrates procedural memory with contextual information to create a high-quality experience pool.
Outcome:	Experiments on BFCL-V3 and AppWorld show that ReMe outperforms memoryless Qwen3-8B.

Your Language Model May Think Too Rigidly: Achieving Reasoning Consistency with Symmetry-Enhanced Training (2025.findings-acl)

Copied to clipboard

Yihang Yao, Zhepeng Cen, Miao Li, William Han, Yuyou Zhang, Emerson Liu, Zuxin Liu, Chuang Gan, Ding Zhao

MAESTRO: Meta-learning Adaptive Estimation of Scalarization Trade-offs for Reward Optimization (2026.acl-long)

Challenge:	Large Language Models (LLMs) have demonstrated strong reasoning capabilities across various tasks.
Approach:	They propose a data-centric approach that enhances LLMs’ awareness of symmetry in query variations and propose syMmetry-ENhanceD (MEND) data augmentation.
Outcome:	Extensive experiments on logical and arithmetic reasoning tasks show that the proposed approach improves model robustness at the knowledge extraction stage through query augmentation.

Copied to clipboard

Yang Zhao, Hepeng Wang, Xiao Ding, Yangou Ouyang, Bibo Cai, Kai Xiong, Jinglong Gao, Zhouhao Sun, Li Du, Bing Qin, Ting Liu

Green & Sustainable NLP

Challenge:	Group-Relative Policy Optimization (GRPO) has emerged as an efficient paradigm for aligning Large Language Models (LLMs), but its efficacy is confined to domains with verifiable ground truths.
Approach:	They propose a meta-cognitive orchestration layer that treats reward scalarization as a dynamic latent policy, leveraging the model’s terminal hidden states as 'a semantic bottleneck' . Across seven benchmarks, MAESTRO consistently outperforms single-reward and static multi-objective baselines while preserving the efficiency advantages of GRPO.
Outcome:	The proposed model outperforms single-reward and static multi-objective baselines while preserving efficiency advantages.

SCCS: Semantics-Consistent Cross-domain Summarization via Optimal Transport Alignment (2023.findings-acl)

Copied to clipboard

Jielin Qiu, Jiacheng Zhu, Mengdi Xu, Franck Dernoncourt, Trung Bui, Zhaowen Wang, Bo Li, Ding Zhao, Hailin Jin

AD-LLM: Benchmarking Large Language Models for Anomaly Detection (2025.findings-acl)

Challenge:	Existing methods for multimodal summarization ignore the structure and semantics of the whole video and article.
Approach:	They propose a semantic-consistent cross-domain summarization model that extracts features from video and article and uses fusion methods to select representative one.
Outcome:	The proposed model produces high-quality multimodal summaries on three MSMO datasets.

Copied to clipboard

Tiankai Yang, Yi Nian, Li Li, Ruiyao Xu, Yuangang Li, Jiaqi Li, Zhuo Xiao, Xiyang Hu, Ryan A. Rossi, Kaize Ding, Xia Hu, Yue Zhao

Mitigating Hallucinations in Multi-modal Large Language Models via Image Token Attention-Guided Decoding (2025.naacl-long)

Challenge:	Anomaly detection (AD) is an important machine learning task with many real-world uses, including fraud detection, medical diagnosis, and industrial monitoring.
Approach:	They propose a benchmark that evaluates how large language models (LLMs) can help with NLP anomaly detection.
Outcome:	The proposed model can perform zero-shot detection without tasks-specific training, data augmentation and model selection, and it can suggest unsupervised AD models.

Copied to clipboard

Xinhao Xu, Hui Chen, Mengyao Lyu, Sicheng Zhao, Yizhe Xiong, Zijia Lin, Jungong Han, Guiguang Ding

Exploring Concept Depth: How Large Language Models Acquire Knowledge and Concept at Different Layers? (2025.coling-main)

Challenge:	Multi-modal large language models (MLLMs) generate plausible but incorrect content, resulting in hallucinations . recent advances in MLLM technology have demonstrated their outstanding performance in a variety of visual tasks, such as object detection.
Approach:	They propose a plug-and-play method which leverages MLLMs’ internal representations to mitigate hallucinations by analyzing input and output tokens.
Outcome:	The proposed method exploits MLLMs’ internal representations to mitigate hallucinations.

Copied to clipboard

Mingyu Jin, Qinkai Yu, Jingyuan Huang, Qingcheng Zeng, Zhenting Wang, Wenyue Hua, Haiyan Zhao, Kai Mei, Yanda Meng, Kaize Ding, Fan Yang, Mengnan Du, Yongfeng Zhang

Curr-ReFT: Overcoming Training Bottlenecks in Small-scale Vision-Language Models via Curriculum Reinforcement Finetuning (2025.findings-emnlp)

Challenge:	Large language models have shown remarkable performances across a wide range of tasks, but mechanisms by which they encode tasks of varying complexity remain poorly understood.
Approach:	They propose to explore the possibility that LLMs process concepts in different layers . they propose to categorize concepts based on their level of abstraction .
Outcome:	The proposed model can process complex concepts in shallow layers, the authors show . the proposed model could be used to prob complex tasks in shallow ones .

Copied to clipboard

Huilin Deng, Ding Zou, Xinghao Zhao, Rui Ma, Yanming Guo, Yang Cao, Yu Kang

A Survey of Large Language Models for Text-Guided Molecular Discovery: From Molecule Generation to Optimization (2026.acl-long)

Challenge:	State-of-the-art vision-language models require massive scaling that limits practical deployment.
Approach:	They propose to use supervised fine-tuning to train small-scale vision-language models but face out-of-domain collapse when trained with traditional supervised learning (SFT).
Outcome:	Experiments show that curr-reFT achieves state-of-the-art performance across visual tasks in both in- and out-of domain settings and benchmarks.

Copied to clipboard

Ziqing Wang, Kexin Zhang, Zihan Zhao, Yibo Wen, Abhishek Pandey, Han Liu, Kaize Ding

Tailored Primitive Initialization is the Secret Key to Reinforcement Learning (2026.acl-long)

Challenge:	Large language models (LLMs) are introducing a paradigm shift in molecular discovery by enabling text-guided interaction with chemical spaces through natural language and symbolic notations.
Approach:	They analyze the current LLM learning paradigms to tackle four critical evaluation dimensions that have emerged as critical dimensions in recent studies.
Outcome:	The proposed models are able to interact with chemical spaces through natural language and symbolic notations, and have emerging extensions to incorporate multi-modal inputs.

Copied to clipboard

Yihang Yao, Guangtao Zeng, Raina Wu, Yang Zhang, Ding Zhao, Zhang-Wei Hong, Chuang Gan

Data Augmentation using LLMs: Data Perspectives, Learning Paradigms and Challenges (2024.findings-acl)

Challenge:	Reinforcement learning (RL) has emerged as a powerful paradigm for improving the reasoning capabilities of large language models.
Approach:	They propose a pipeline that automatically discovers thinking token patterns with reasoning primitives and curates SFT datasets to prepare LLMs for RL.
Outcome:	The proposed pipeline outperforms baseline methods on mathematical and logical reasoning benchmarks on RL tasks.

Copied to clipboard

Bosheng Ding, Chengwei Qin, Ruochen Zhao, Tianze Luo, Xinze Li, Guizhen Chen, Wenhan Xia, Junjie Hu, Anh Tuan Luu, Shafiq Joty

PharmMT: A Neural Machine Translation Approach to Simplify Prescription Directions (2020.findings-emnlp)

Challenge:	Data augmentation (DA) is a key technique for enhancing model performance by diversifying training examples without the need for additional data collection.
Approach:	They examine various strategies that utilize LLMs for data augmentation, including a novel exploration of learning paradigms where LLM-generated data is used for diverse forms of further training.
Outcome:	The proposed approach addresses the primary open challenges faced by LLMs in the field of large language models and aims to serve as a comprehensive guide for researchers and practitioners.

Copied to clipboard

Jiazhao Li, Corey Lester, Xinyan Zhao, Yuting Ding, Yun Jiang, V.G.Vinod Vydiswaran

Text Generation

Challenge:	a novel machine translation-based approach to simplify prescription directions is proposed . the language used by physicians and health professionals includes medical jargon and implicit directives .
Approach:	They propose a machine translation-based approach to automatically and reliably simplify prescription directions into patient-friendly language.
Outcome:	The proposed system achieves a BLEU score of 60.27 over 530K prescriptions from a large mail-order pharmacy.

SciRAG: Adaptive, Citation-Aware, and Outline-Guided Retrieval and Synthesis for Scientific Literature (2026.eacl-long)

Copied to clipboard

Hang Ding, Yilun Zhao, Tiansheng Hu, Zihang Wang, Manasi Patwardhan, Arman Cohan

Let’s Be Self-generated via Step by Step: A Curriculum Learning Approach to Automated Reasoning with Large Language Models (2025.findings-acl)

Challenge:	Existing retrieval-augmented generation methods overlook citation graph structure, adapt poorly to complex queries, and yield fragmented, hard-to-verify syntheses.
Approach:	They propose a retrieval-augmented generation framework that addresses these gaps by combining adaptive retrieval and symbolic reasoning.
Outcome:	Extensive experiments show that SciRAG outperforms prior systems in factual accuracy and synthesis quality.

Copied to clipboard

Kangyang Luo, Zichen Ding, Zhenmin Weng, Lingfeng Qiao, Meng Zhao, Xiang Li, Di Yin, Jinlong Shu

RU22Fact: Optimizing Evidence for Multilingual Explainable Fact-Checking on Russia-Ukraine Conflict (2024.lrec-main)

Challenge:	Existing efforts to improve CoT prompting have limitations that require extensive human effort or performance needs to be improved.
Approach:	They propose a prompt approach for automatic reasoning called LBS3 inspired by curriculum learning which better reflects human learning habits.
Outcome:	The proposed approach achieves strongly competitive performance compared to baselines in reasoning-intensive tasks with varying open- and closed-source LLMs.

Copied to clipboard

Yirong Zeng, Xiao Ding, Yi Zhao, Xiangyu Li, Jie Zhang, Chao Yao, Ting Liu, Bing Qin

FCM: A Fine-grained Comparison Model for Multi-turn Dialogue Reasoning (2021.findings-emnlp)

Challenge:	Existing methods to verify factuality of claims do not provide sufficient evidence for explainable fact-checking systems.
Approach:	They propose a method to automatically retrieve and summarize evidence from the Web and a novel multilingual explainable fact-checking dataset on the Russia-Ukraine conflict in 2022.
Outcome:	The proposed method can retrieve and summarize evidence from the Web and generate explanations in 16 languages.

Copied to clipboard

Xu Wang, Hainan Zhang, Shuai Zhao, Yanyan Zou, Hongshen Chen, Zhuoye Ding, Bo Cheng, Yanyan Lan

Natural Language Interfaces

Challenge:	Existing neural dialogue models only capture syntactic and semantic information, but fail to model the logical consistency between the dialogue history and the generated response.
Approach:	They propose a fine-grained comparison model to capture syntactic and semantic information and then compare each candidate's representation with the whole history to obtain a history consistency representation.
Outcome:	The proposed model obtains higher ranking scores than baseline models on two public dialogue datasets.

Multi-stage Distillation Framework for Cross-Lingual Semantic Similarity Matching (2022.findings-naacl)

Copied to clipboard

Kunbo Ding, Weijie Liu, Yuejian Fang, Zhe Zhao, Qi Ju, Xuefeng Yang, Rong Tian, Zhu Tao, Haoyan Liu, Han Guo, Xingyu Bai, Weiquan Mao, Yudong Li, Weigang Guo, Taiqiang Wu, Ningyuan Sun

Learning to Maximize Mutual Information for Chain-of-Thought Distillation (2024.findings-acl)

Challenge:	Existing studies have shown that cross-lingual knowledge distillation can improve the performance of pre-trained models for cross-linguistic similarity matching tasks.
Approach:	They propose a multi-stage distillation framework for constructing a small-size but high-performance cross-lingual model using contrastive learning, bottleneck, and parameter recurrent strategies.
Outcome:	The proposed model can compress the size of XLM-R and MiniLM by more than 50% while the performance is only reduced by about 1%.

Copied to clipboard

Xin Chen, Hanxian Huang, Yanjun Gao, Yi Wang, Jishen Zhao, Ke Ding

Paragraph-level Neural Question Generation with Maxout Pointer and Gated Self-attention Networks (D18-1)

Challenge:	Knowledge distillation is a technique of transferring knowledge from large, complex models to smaller ones.
Approach:	They propose a method utilizing chain-of-thought distillation to transfer knowledge from large, complex models to smaller ones by maximizing mutual information of the representation features of the two tasks.
Outcome:	The proposed method outperforms the state-of-the-art knowledge distillation method on four datasets.

Copied to clipboard

Yao Zhao, Xiaochuan Ni, Yuanyuan Ding, Qifa Ke

Bridging SFT and RL: Dynamic Policy Optimization for Robust Reasoning (2026.findings-acl)

Challenge:	Existing rule-based question generation models rely on one or two sentences as input, while long text has posed challenges for sequence to sequence neural models.
Approach:	They propose a maxout pointer mechanism with gated self-attention encoder to address the challenges of processing long text inputs for question generation.
Outcome:	The proposed model outperforms existing models with sentence-level or paragraph-level inputs pushing the state-of-the-art result from 13.9 to 16.3 (BLEU_4).

Copied to clipboard

Taojie Zhu, Dongyang Xu, Ding Zou, Sen Zhao, Qiaobo Hao, Zhiguo Yang, Yonghong He

GLaRA: Graph-based Labeling Rule Augmentation for Weakly Supervised Named Entity Recognition (2021.eacl-main)

Challenge:	Existing unified optimization strategies overlook the statistical conflict between these distinct gradient signals.
Approach:	They propose a framework to reduce bias-variance trade-offs in Large Language Models . they propose DYPO, which leverages intrinsic group dynamics to significantly reduce RL gradient variance .
Outcome:	The proposed framework outperforms traditional pipelines on reasoning benchmarks and out-of-distribution tasks.

Copied to clipboard

Xinyan Zhao, Haibo Ding, Zhe Feng

ViDoRAG: Visual Document Retrieval-Augmented Generation via Dynamic Iterative Reasoning Agents (2025.emnlp-main)

Challenge:	Named entity recognition (NER) models often need to be trained with many manual labels to perform well.
Approach:	They propose to train named entity recognition systems using heuristic labeling rules . they create a graph with nodes representing candidate rules extracted from unlabeled data .
Outcome:	The proposed method achieves an average improvement of +20% over the baseline on three NER datasets.

Copied to clipboard

Qiuchen Wang, Ruixue Ding, Zehui Chen, Weiqi Wu, Shihang Wang, Pengjun Xie, Feng Zhao

Parameter-efficient Continual Learning Framework in Industrial Real-time Text Classification System (2022.naacl-industry)

Challenge:	Existing benchmarks focus on image-based question answering (QA) but ignore the fundamental challenges of efficient retrieval, comprehension, and reasoning within dense visual documents.
Approach:	They propose a novel multi-agent RAG framework tailored for complex reasoning across visual documents that employs a Gaussian Mixture Model (GMM)-based hybrid strategy to handle multi-modal retrieval.
Outcome:	The proposed framework outperforms existing methods by over 10% on the competitive ViDoSeek benchmark.

Copied to clipboard

Tao Zhu, Zhe Zhao, Weijie Liu, Jiachi Liu, Yiren Chen, Weiquan Mao, Haoyan Liu, Kunbo Ding, Yudong Li, Xuefeng Yang

Natural Logic at the Core: Dynamic Rewards for Entailment Tree Generation (2025.findings-acl)

Challenge:	Existing continual learning methods use data replay, parameter isolation and regularization to mitigate catastrophic forgetting.
Approach:	They propose a parameter-efficient continual learning framework that updates parameters offline and then trains using an online regularization method.
Outcome:	The proposed framework reduces catastrophic forgetting and saves the model with the changed parameters instead of all parameters.

Copied to clipboard

Jihao Shi, Xiao Ding, Kai Xiong, Hengwei Zhao, Bing Qin, Ting Liu

Reasoning

Challenge:	Existing approaches to generating entailment trees lack logical consistency . static reward structures or intricate dependencies within multi-step reasoning are often ignored .
Approach:	They propose a method that integrates natural logic principles into reinforcement learning to guide entailment tree generation.
Outcome:	Experiments on EntailmentBank show that the proposed method improves interpretability and generalization.

Read Anywhere Pointed: Layout-aware GUI Screen Reading with Tree-of-Lens Grounding (2024.emnlp-main)

Copied to clipboard

Yue Fan, Lei Ding, Ching-Chen Kuo, Shan Jiang, Yang Zhao, Xinze Guan, Jie Yang, Yi Zhang, Xin Wang

Challenge:	Existing models for GUI understanding ignore a key GUI-referring task: screen reading based on user-indicated points.
Approach:	They propose a Tree-of-Lens agent that constructs a Hierarchical Layout Tree based on user input points and a GUI screenshot.
Outcome:	The proposed agent can interpret the Screen Point-and-Read task on mobile, web, and operating systems.

Do Emergent Abilities Exist in Quantized Large Language Models: An Empirical Study (2024.lrec-main)

Copied to clipboard

Peiyu Liu, Zikang Liu, Ze-Feng Gao, Dawei Gao, Wayne Xin Zhao, Yaliang Li, Bolin Ding, Ji-Rong Wen

Vulnerability of Text-to-Image Models to Prompt Template Stealing: A Differential Evolution Approach (2025.findings-acl)

Challenge:	Large Language Models (LLMs) require significant computational resources for deployment and use.
Approach:	They propose to use low-bit quantization methods to reduce memory footprint and increase inference rate to improve performance of Large Language Models.
Outcome:	The proposed methods can reduce the memory footprint and increase the inference rate of LLMs.

Copied to clipboard

Yurong Wu, Fangwen Mu, Qiuhong Zhang, Jinjing Zhao, Xinrun Xu, Lingrui Mei, Yang Wu, Lin Shi, Junjie Wang, Zhiming Ding, Yiwei Wang

Self-Evolving GPT: A Lifelong Autonomous Experiential Learner (2024.acl-long)

Challenge:	Prompt trading has emerged as a significant intellectual property concern in recent years, where vendors entice users by showcasing sample images before selling prompt templates that can generate similar images.
Approach:	They propose a prompt-stealing benchmark consisting of 50 templates and 450 images organized into Easy and Hard difficulty levels.
Outcome:	The proposed method outperforms baseline methods with an average improvement of over 10%.

Copied to clipboard

Jinglong Gao, Xiao Ding, Yiming Cui, Jianbai Zhao, Hepeng Wang, Ting Liu, Bing Qin

Don’t Be Misled by Style: A Style-Adaptive Reranker for Capturing Effective Knowledge in Retrieval-Augmented Generation (2026.acl-long)

Challenge:	Existing approaches to provide LLMs with textual task-solving experience rely on manual efforts to acquire and apply such experience for each task.
Approach:	They propose a lifelong autonomous experiential learning framework based on LLMs that learns and accumulates experience through experience transfer and induction.
Outcome:	The proposed framework performs reliably in each intermediate step and improves GPT-3.5 and GPT-4 on widely used NLP datasets.

Copied to clipboard

Ruwen Zhang, Bo Liu, Zhang Sheng Xiang, Yida Chen, Hantao Zhao, Ding Ding, Jiahui Jin, Jiuxin Cao

Meeseeks: A Feedback-Driven, Iterative Self-Correction Benchmark evaluating LLMs’ Instruction Following Capability (2026.findings-acl)

Challenge:	Existing rerankers are mainly trained on well-edited texts, but stylistic features can be misled by reranked models.
Approach:	They propose a style-augmented multi-task framework that prioritizes effective knowledge over stylistic perturbations by using an LLM to derive passage-level supervision on whether a passage helps or harms answer correctness.
Outcome:	Extensive experiments show that SARK improves generation performance across multiple LLMs under mixed-style conditions.

Copied to clipboard

Jiaming Wang, Yunke Zhao, Peng Ding, Jun Kuang, Yibin Shen, Zhe Tang, Yilin Jin, ZongYu Wang, Xiaoyu Li, Xuezhi Cao

HER: Human-like Reasoning and Reinforcement Learning for LLM Role-playing (2026.findings-acl)

Challenge:	Existing models lack the ability to adhere to instructions, resulting in suboptimal performance.
Approach:	They propose an automated iterative instruction-following benchmark with integrated feedback mechanism.
Outcome:	The proposed benchmark identifies erroneous components in model responses and provides feedback accurately.

Copied to clipboard

Chengyu Du, Xintao Wang, Aili Chen, Weiyuan Li, Rui Xu, Junteng Liu, Zishan Huang, Rong Tian, Zijun Sun, Yuhao Li, Liheng Feng, Deming Ding, Pengyu Zhao, Yanghua Xiao

Reasoning

Challenge:	Existing models for LLM role-playing lack high-quality datasets with explicit reasoning traces and reliable reward signals aligned with human preferences.
Approach:	They propose a unified framework for cognitive-level persona simulation that strictly distinguishes characters’ first-person thinking processes from LLMs’ third-person reasoning.
Outcome:	The proposed framework outperforms the Qwen3-32B baseline model and achieves a 30.26% and 14.97% performance on the minimax benchmarks.

Deciphering the Impact of Pretraining Data on Large Language Models through Machine Unlearning (2024.findings-acl)

Copied to clipboard

Yang Zhao, Li Du, Xiao Ding, Kai Xiong, Zhouhao Sun, Shi Jun, Ting Liu, Bing Qin

Analyzing the Rapid Generalization of SFT via the Perspective of Attention Head Activation Patterns (2025.acl-long)

Challenge:	Existing studies have suggested that the composition of the pretraining corpus exerts a significant impact upon the performance of LLMs.
Approach:	They analyze the impact of 48 datasets from 5 major categories of pretraining data of Large Language Models and measure their impacts on LLMs using benchmarks about nine major categories.
Outcome:	The proposed analysis provides insights into the organization of data to support more efficient pretraining of Large Language Models.

Copied to clipboard

Yang Zhao, Li Du, Xiao Ding, Kai Xiong, Ting Liu, Bing Qin

Green & Sustainable NLP

Challenge:	Currently, LLMs learn in a data-driven schema while the instructions about complex tasks are both scarce and hard to collect or construct.
Approach:	They employ a gradient-based method to dissect the process that the Supervised Fine-tuning Process (SFT) adapts LLMs to downstream tasks via the perspective of attention patterns.
Outcome:	The proposed method dissects the process that the SFT process adapts LLMs to downstream tasks via the perspective of attention patterns.

H-MAS: Hierarchical Multi-Agent Scheduling for Multi-Tenant LLM Serving (2026.findings-acl)

Copied to clipboard

Yuhan Liu, Cong Xu, Qi Jia, Yihua Wang, Feiyu Chen, Liang Jin, Lu Liu, Yaqian Zhao, Yuting Ding, Xiang Li

Challenge:	Multi-tenant Model-as-a-Service (MaaS) workloads exhibit non-stationarity across multiple time scales . existing request schedulers often rely on a fixed policy that remains unchanged at runtime .
Approach:	They propose a hierarchical multi-agent scheduler that operates in a layered closed loop . they propose to maintain 1.2–3.0 higher Goodput than SGLang and vLLM .
Outcome:	Experiments show that H-MAS achieves 1.2–3.0 higher Goodput than SGLang and vLLM . it maintains more stable QoS under diverse request lengths and heterogeneous SLO targets .

AEGIS: A Holistic Benchmark for Evaluating Forensic Analysis of AI-Generated Academic Images (2026.acl-long)

Copied to clipboard

Bo Zhang, Tzu-Yen Ma, Zichen Tang, Junpeng Ding, Zirui Wang, Yizhuo Zhao, Peilin Gao, Zijie Xi, Zixin Ding, Haiyang Sun, Haocheng Gao, Yuan Liu, Liangjia Wang, Yiling Huang, Yujie Wang, Yuyue Zhang, Ronghui Xi, Yuanze Li, Jiacheng Liu, Zhongjun Yang, Haihong E

A Thorough Examination on Zero-shot Dense Retrieval (2023.findings-emnlp)

Challenge:	AEGIS examines whether current models can effectively audit AI-generated images in academic papers.
Approach:	They propose a holistic benchmark for forensic analysis of AI-Generated academic ImageS that reveals limitations in academic image forensics.
Outcome:	AEGIS compared with existing benchmarks on seven academic categories and features key advances in forensic analysis.

Copied to clipboard

Ruiyang Ren, Yingqi Qu, Jing Liu, Xin Zhao, Qifei Wu, Yuchen Ding, Hua Wu, Haifeng Wang, Ji-Rong Wen

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

Challenge:	Recent advances in dense retrieval (DR) models have been shown to be not as competitive as traditional sparse retrieval models in a zero-shot retrieval setting.
Approach:	They propose to examine the zero-shot capability of DR models by analyzing key factors related to source training set and potential bias from target dataset.
Outcome:	The proposed model is not as competitive as sparse retrieval models in a zero-shot retrieval setting.

Copied to clipboard

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

CoAct: Co-Active LLM Preference Learning with Human-AI Synergy (2026.acl-long)

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

Ruiyao Xu, Mihir Parmar, Tiankai Yang, Zhengyu Hu, Yue Zhao, Kaize Ding

AdaTP: Attention-Debiased Token Pruning for Video Large Language Models (2025.findings-emnlp)

Challenge:	Existing methods to learn from preference-based feedback are expensive and scarce.
Approach:	They propose a framework that synergistically combines self-rewarding and active learning through human-AI collaboration.
Outcome:	The proposed framework outperforms existing methods on three reasoning benchmarks and achieves average improvements of +13.25% on GSM8K, +8.19% on MATH, and +13.16% on WebInstruct.

Copied to clipboard

Fengyuan Sun, Leqi Shen, Hui Chen, Sicheng Zhao, Jungong Han, Guiguang Ding

Decoding Scientific Experimental Images: The SPUR Benchmark for Perception, Understanding, and Reasoning (2026.acl-long)

Challenge:	Existing visual token compression methods rely on attention scores but have inherent biases . global and local attention biased scores cause excessive computational overhead .
Approach:	They propose a token pruning pipeline that targets global and local attention biases . the pipeline is designed to reduce computational overhead of Video Large Language Models based on visual tokens compiled from multiple video frames .
Outcome:	The proposed method significantly reduces the computational overhead of Video Large Language Models while retaining the performance of vanilla models.

Copied to clipboard

Junpeng Ding, Zichen Tang, Haihong E, Mengyuan Ji, Yang Liu, Haolin Tian, Haiyang Sun, Pengqi Sun, Yang Xu, Yichen Liu, Haocheng Gao, Zijie Xi, Ruomeng Jiang, Peizhi Zhao, Rongjin Li, Yuanze Li, Jiacheng Liu, Zhongjun Yang, Jintong Chen, Siying Lin

DFAMS: Dynamic-flow guided Federated Alignment based Multi-prototype Search (2026.acl-long)

Challenge:	Xu and Peng, 2025) . . SPUR is a comprehensive benchmark for scientific experimental image perception, understanding, and reasoning, comprising 4,264 question-answering (QA) pairs derived from 1,084 expert-curated images.
Approach:	They propose to use 4,264 question-answering (QA) pairs derived from 1,084 expert-curated images to evaluate the visual perception of multimodal large language models (MLLMs) . they also propose to utilize cross-panel relation understanding to evaluate MLLM’s ability to decipher intricate cross-panel relations.
Outcome:	The proposed model is based on 4,264 question-answering pairs derived from 1,084 expert-curated images.

Copied to clipboard

Zhibang Yang, Xinke Jiang, Rihong Qiu, Ruiqing Li, Yihang Zhang, Yue Fang, Yongxin Xu, Hongxin Ding, Xu Chu, Junfeng Zhao, Yasha Wang

LLMBox: A Comprehensive Library for Large Language Models (2024.acl-demos)

Challenge:	Existing methods for ambiguous queries struggle to retrieve high-quality documents . DFAMS outperforms advanced FR methods by 14.37% in knowledge classification accuracy .
Approach:	They propose a framework that leverages dynamic information flow to identify latent query intents and construct semantically aligned knowledge partitions for accurate retrieval across heterogeneous sources.
Outcome:	The proposed framework outperforms existing methods in classification accuracy and retrieval recall tests.

Copied to clipboard

Tianyi Tang, Hu Yiwen, Bingqian Li, Wenyang Luo, ZiJing Qin, Haoxiang Sun, Jiapeng Wang, Shiyi Xu, Xiaoxue Cheng, Geyang Guo, Han Peng, Bowen Zheng, Yiru Tang, Yingqian Min, Yushuo Chen, Jie Chen, Ranchi Zhao, Luran Ding, Yuhao Wang, Zican Dong, Xia Chunxuan, Junyi Li, Kun Zhou, Xin Zhao, Ji-Rong Wen

Large Language Models Are Still Misled by Simple Bias Ensembles (2026.findings-acl)

Challenge:	a library to facilitate the development, use, and evaluation of large language models (LLMs) is presented.
Approach:	They propose a unified library to facilitate the development, use and evaluation of large language models (LLMs).
Outcome:	The proposed library is based on extensive experiments in a variety of evaluation settings.

Copied to clipboard

Zhouhao Sun, Zhiyuan Kan, Xiao Ding, Li Du, Bibo Cai, Yang Zhao, Bing Qin, Ting Liu

Self-supervised Product Title Rewrite for Product Listing Ads (2022.naacl-industry)

Challenge:	Existing benchmarks for large language models are constrained to datasets where each sample is manually injected with only one type of bias.
Approach:	They propose a multi-bias benchmark where each sample contains multiple types of biases.
Outcome:	The proposed benchmark shows that existing LLMs and debiasing methods perform poorly on this benchmark, highlighting the challenge of eliminating compounded biases.

Copied to clipboard

Xue Zhao, Dayiheng Liu, Junwei Ding, Liang Yao, Mahone Yan, Huibo Wang, Wenqing Yao

GuideLLM: Exploring LLM-Guided Conversation with Applications in Autobiography Interviewing (2025.naacl-long)

Challenge:	Existing work has investigated the title optimization for Product Listing Ads (PLAs) however, little work has examined the effectiveness of this method.
Approach:	They propose a method to rewrite product listing ads titles without considering the fluency and information priority.
Outcome:	The proposed solution reduces the cost and improves CTR in the offline test and real-world online test by a large amount.

Copied to clipboard

Jinhao Duan, Xinyu Zhao, Zhuoxuan Zhang, Eunhye Grace Ko, Lily Boddy, Chenan Wang, Tianhao Li, Alexander Rasgon, Junyuan Hong, Min Kyung Lee, Chenxi Yuan, Qi Long, Ying Ding, Tianlong Chen, Kaidi Xu

Natural Language Interfaces

Challenge:	Large Language Models (LLMs) have demonstrated their effectiveness in human-guided dialogues, but tasks in the real world are more complex and require greater autonomy from LLMs.
Approach:	They propose to characterize LLM-guided conversation into three fundamental components: Goal Navigation, Context Management, Empathetic Engagement and implement an interviewing environment for the evaluation of LLMs.
Outcome:	The proposed LLM outperforms baseline LLMs in interviewing quality and autobiography generation quality.

Easy and Efficient Transformer: Scalable Inference Solution For Large NLP Model (2022.naacl-industry)

Copied to clipboard

Gongzheng Li, Yadong Xi, Jingzhen Ding, Duan Wang, Ziyang Luo, Rongsheng Zhang, Bai Liu, Changjie Fan, Xiaoxi Mao, Zeng Zhao

Transfer Knowledge from Natural Language to Electrocardiography: Can We Detect Cardiovascular Disease Through Language Models? (2023.findings-eacl)

Challenge:	Recent studies show that transformer-based models are effective over many tasks, but they are expensive to deploy in the industrial application.
Approach:	They propose a transformer-based inference solution that optimizes kernels for long inputs and large hidden sizes and a flexible CUDA memory manager to reduce the memory footprint when deploying a large model.
Outcome:	The proposed solution achieves an average speedup of 1.40-4.20x on the transformer decoder layer with an A100 GPU.

Copied to clipboard

Jielin Qiu, William Han, Jiacheng Zhu, Mengdi Xu, Michael Rosenberg, Emerson Liu, Douglas Weber, Ding Zhao

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

Challenge:	Recent advances in Large Language Models (LLMs) have shown powerful ability in various downstream applications.
Approach:	They propose an approach for cardiovascular disease diagnosis and automatic ECG diagnosis report generation.
Outcome:	The proposed approach generates high-quality cardiac diagnosis reports and achieves competitive zero-shot classification performance even compared with supervised baselines.

Copied to clipboard

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

ProMed: Shapley Information Gain Guided Reinforcement Learning for Proactive Medical LLMs (2026.acl-long)

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

Hongxin Ding, Baixiang Huang, Yue Fang, Weibin Liao, Xinke Jiang, Jinyang Zhang, Yinghao Zhu, Zheng Li, Liantao Ma, Junfeng Zhao, Yasha Wang

Challenge:	Existing medical Large Language Models (LLMs) follow a reactive paradigm, risking diagnostic errors by answering before seeking sufficient details.
Approach:	They propose a reinforcement learning framework that transitions LLMs toward a proactive paradigm, enabling them to ask clinically valuable questions before decision-making.
Outcome:	Experiments on partial-information medical benchmarks show that ProMed outperforms state-of-the-art methods by 6.29% on average and delivers a 54.45% gain over the reactive paradigm.

Inside Out: Evolving User-Centric Core Memory Trees for Long-Term Personalized Dialogue Systems (2026.acl-long)

Copied to clipboard

Jihao Zhao, Ding Chen, Zhaoxin Fan, Kerun Xu, Mengting Hu, Bo Tang, Feiyu Xiong, Zhiyu li

Natural Language Interfaces

Challenge:	Existing personalized dialogue systems struggle to reconcile unbounded interactions with finite context constraints.
Approach:	They propose a framework that utilizes a globally maintained PersonaTree as the carrier of long-term user profiling.
Outcome:	The proposed framework outperforms existing systems in suppressing contextual noise and persona inconsistency.

Can Brain Signals Reveal Inner Alignment with Human Languages? (2023.findings-emnlp)

Copied to clipboard

Jielin Qiu, William Han, Jiacheng Zhu, Mengdi Xu, Douglas Weber, Bo Li, Ding Zhao

Sentiment Analysis

Challenge:	Brain Signals, such as Electroencephalography, and human languages have been explored independently for many downstream tasks, however, the connection between them has not been well explored.
Approach:	They introduce a multimodal transformer alignment model to observe coordinated representations between EEG and language.
Outcome:	The proposed method achieved an F1-score improvement of 1.7% on ZuCo and 9.3% on Zuco datasets for sentiment analysis, and 7.4% on ZuCO for relation detection.

RocketQA: An Optimized Training Approach to Dense Passage Retrieval for Open-Domain Question Answering (2021.naacl-main)

Copied to clipboard

Yingqi Qu, Yuchen Ding, Jing Liu, Kai Liu, Ruiyang Ren, Wayne Xin Zhao, Daxiang Dong, Hua Wu, Haifeng Wang

OmniAlign-V: Towards Enhanced Alignment of MLLMs with Human Preference (2025.acl-long)

Challenge:	Open-domain question answering uses dense passage retrieval to find answers . however, it is difficult to effectively train a dual-encoder due to discrepancy between training and inference .
Approach:	They propose an optimized training approach to improve dense passage retrieval using RocketQA . they propose cross-batch negatives, denoised hard negatives and data augmentation .
Outcome:	The proposed approach outperforms state-of-the-art models on both MSMARCO and Natural Questions.

Copied to clipboard

Xiangyu Zhao, Shengyuan Ding, Zicheng Zhang, Haian Huang, Maosongcao Maosongcao, Jiaqi Wang, Weiyun Wang, Xinyu Fang, Wenhai Wang, Guangtao Zhai, Hua Yang, Haodong Duan, Kai Chen