Papers by Lu Cao
Monitoring Decoding: Mitigating Hallucination via Evaluating the Factuality of Partial Response during Generation (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods to mitiga hallucinations rely on sampling multiple full-length generations, which introduces significant response latency and becomes ineffective when the model consistently produces hallucines. |
| Approach: | They propose a framework that dynamically monitors the generation process and selectively applies in-process interventions to revise hallucination-prone tokens. |
| Outcome: | The proposed framework outperforms self-consistency-based approaches in both effectiveness and efficiency, achieving higher factual accuracy while significantly reducing computational overhead. |
Thread: A Logic-Based Data Organization Paradigm for How-To Question Answering with Retrieval Augmented Generation (2025.emnlp-main)
Copied to clipboard
Kaikai An, Fangkai Yang, Liqun Li, Junting Lu, Sitao Cheng, Shuzheng Si, Lu Wang, Pu Zhao, Lele Cao, Qingwei Lin, Saravan Rajmohan, Dongmei Zhang, Baobao Chang
| Challenge: | Recent advances in retrieval-augmented generation (RAG) have substantially improved question-answering systems, particularly for factoid ‘5Ws’ questions. |
| Approach: | They propose a data organization paradigm where large language models transform documents into more structured and loosely interconnected LUs. |
| Outcome: | Experiments in open-domain and industrial settings show that the proposed paradigm outperforms existing paradigms and shows high adaptability across diverse document formats. |
ScaleBox: Enabling High-Fidelity and Scalable Code Verification for Large Language Models (2026.acl-demo)
Copied to clipboard
Jiasheng Zheng, Xin Zheng, Boxi Cao, Pengbo Wang, Zhengzhao Ma, Qiming Zhu, Jiazhen Jiang, Yaojie Lu, Hongyu Lin, Xianpei Han, Le Sun
| Challenge: | Existing code sandboxes fail to provide accurate verification and efficiency under high-concurrency workloads. |
| Approach: | They propose a high-fidelity code verification system that provides sandbox feedback for RL training and evaluation. |
| Outcome: | The proposed system outperforms heuristic-matching baselines on LiveCodeBench and training stability on high-concurrency workloads. |
All Languages Matter: Understanding and Mitigating Language Bias in Multilingual RAG (2026.acl-long)
Copied to clipboard
Dan Wang, Guozhao Mo, Yafei Shi, Cheng Zhang, Bo Zheng, Boxi Cao, Xuanang Chen, Yaojie Lu, Hongyu Lin, Ben He, Xianpei Han, Le Sun
| Challenge: | Existing mRAG systems suffer from a language bias during reranking, systematically favoring English and the query’s native language. |
| Approach: | They propose a language-agnostic utility-driven reranker alignment technique to mitigate language bias during re-ranking. |
| Outcome: | The proposed approach mitigates language bias and consistently improves mRAG performance across languages. |
LLMs Assist NLP Researchers: Critique Paper (Meta-)Reviewing (2024.emnlp-main)
Copied to clipboard
Jiangshu Du, Yibo Wang, Wenting Zhao, Zhongfen Deng, Shuaiqi Liu, Renze Lou, Henry Zou, Pranav Narayanan Venkit, Nan Zhang, Mukund Srinath, Haoran Zhang, Vipul Gupta, Yinghui Li, Tao Li, Fei Wang, Qin Liu, Tianlin Liu, Pengzhi Gao, Congying Xia, Chen Xing, Cheng Jiayang, Zhaowei Wang, Ying Su, Raj Shah, Ruohao Guo, Jing Gu, Haoran Li, Kangda Wei, Zihao Wang, Lu Cheng, Surangika Ranathunga, Meng Fang, Jie Fu, Fei Liu, Ruihong Huang, Eduardo Blanco, Yixin Cao, Rui Zhang, Philip Yu, Wenpeng Yin
| Challenge: | a comparative analysis of paper (meta-)reviews by large language models (LLMs) aims to identify and distinguish LLMs from human activities . |
| Approach: | They present a comparative analysis to identify and distinguish LLM activities from human activities. |
| Outcome: | The proposed analysis aims to improve recognition of instances when someone implicitly uses LLMs for reviewing activities. |
Learning In-context Learning for Named Entity Recognition (2023.acl-long)
Copied to clipboard
Jiawei Chen, Yaojie Lu, Hongyu Lin, Jie Lou, Wei Jia, Dai Dai, Hua Wu, Boxi Cao, Xianpei Han, Le Sun
| Challenge: | Existing methods to recognize entities in text are limited by the diversity of entity types and the lack of high-quality annotations. |
| Approach: | They propose an in-context learning-based NER approach that can inject in-const NER ability into PLMs and recognize entities of novel types on-the-fly using only a few demonstrative instances. |
| Outcome: | The proposed method outperforms the PLMs+fine-tuning counterparts on 4 few-shot NER datasets and significantly outperformed the Plms+initialized extractors. |
CoE-SQL: In-Context Learning for Multi-Turn Text-to-SQL with Chain-of-Editions (2024.naacl-long)
Copied to clipboard
| Challenge: | Recent studies have demonstrated that Large Language Models (LLMs) have impressive capabilities in a variety of domains and tasks. |
| Approach: | They propose a method which prompts LLMs to generate SQL queries based on the previously generated SQL query with an edition chain. |
| Outcome: | The proposed method outperforms different in-context learning baselines and achieves state-of-the-art performance on two benchmarks SParC and CoSQL using LLMs. |
Unsupervised Dual Paraphrasing for Two-stage Semantic Parsing (2020.acl-main)
Copied to clipboard
| Challenge: | Existing semantic parsing frameworks rely on nontrivial human labor to generate canonical utterances. |
| Approach: | They propose a framework that uses an unsupervised paraphrase model to parse canonical utterances. |
| Outcome: | The proposed framework is effective and compatible with supervised training. |
A Chain-of-Task Framework for Instruction Tuning of LLMs Based on Chinese Grammatical Error Correction (2025.coling-main)
Copied to clipboard
| Challenge: | Existing approaches to address Grammatical Error Correction (GEC) tasks are based on large scale labeled data, which leads to extremely high data annotation costs. |
| Approach: | They propose a Chain-of-Task framework to reduce over-correction in large language models . they propose supervised fine-tuning strategy and an algorithm for automatic dataset annotation . |
| Outcome: | The proposed framework achieves state-of-the-art on both FCGEC (in-domain) and NaCGEC (out-of domain) test sets. |
Jailbreak Open-Sourced Large Language Models via Enforced Decoding (2024.acl-long)
Copied to clipboard
Hangfan Zhang, Zhimeng Guo, Huaisheng Zhu, Bochuan Cao, Lu Lin, Jinyuan Jia, Jinghui Chen, Dinghao Wu
| Challenge: | Existing studies show that Large Language Models can be misused to generate undesired content. |
| Approach: | They propose to use large language models to manipulate the generation process to generate undesired content without heavy computations or prompt designs. |
| Outcome: | The proposed method shows that open-sourced large language models could be misused to generate undesired content without heavy computations or prompt designs. |
AWESOME: GPU Memory-constrained Long Document Summarization using Memory Mechanism and Global Salient Content (2024.naacl-long)
Copied to clipboard
| Challenge: | Existing solutions focus on efficient attentions or divide-and-conquer strategies, but these methods sacrifice global context, leading to incoherent and uninformative summaries. |
| Approach: | They propose to leverage the memory-efficient nature of divide-and-conquer methods while preserving global context. |
| Outcome: | The proposed framework improves informativeness, faithfulness, and coherence over baselines on government reports, meeting transcripts, screenplays, scientific papers, and novels. |
HIBRIDS: Attention with Hierarchical Biases for Structure-aware Long Document Summarization (2022.acl-long)
Copied to clipboard
| Challenge: | Document structure is critical for efficient information consumption, but it is difficult to encode it efficiently into the modern Transformer architecture. |
| Approach: | They propose a task which injects Hierarchical Biases foR Incorporating Document Structure into attention score calculation. |
| Outcome: | The proposed model produces better question-summary hierarchies than comparisons on hierarchy quality and content coverage, the authors show . |
Exploring Schema Generalizability of Text-to-SQL (2023.findings-acl)
Copied to clipboard
| Challenge: | Existing text-to-SQL models are limited in their generalizability, despite their performance being over-estimated. |
| Approach: | They propose a framework to generate novel text-to-SQL data via automatic and synchronous (DS, SQL) pair altering. |
| Outcome: | The proposed framework generates text-to-SQL data via automatic and synchronous (DS, SQL) pair altering. |
CSS: A Large-scale Cross-schema Chinese Text-to-SQL Medical Dataset (2023.findings-acl)
Copied to clipboard
| Challenge: | a cross-domain text-to-SQL task aims to parse user questions into SQL on complete unseen databases . a single-domain task evaluates the performance on identical databases based on the same domain . |
| Approach: | They propose a cross-domain text-to-SQL task that parses user questions into SQL on unseen databases. |
| Outcome: | The proposed system can parse user questions into SQL on complete unseen databases. |
CLIFF: Contrastive Learning for Improving Faithfulness and Factuality in Abstractive Summarization (2021.emnlp-main)
Copied to clipboard
| Challenge: | Existing methods for generating abstractive summarization are inconsistent and rely on heuristically created data for error handling. |
| Approach: | They propose a contrastive learning formulation that leverages both positive and negative summaries to train summarization systems that are better at distinguishing between them. |
| Outcome: | The proposed learning framework produces more factual summaries than strong comparisons with post error correction, entailment-based reranking, and unlikelihood training. |
Efficient Attentions for Long Document Summarization (2021.naacl-main)
Copied to clipboard
| Challenge: | Existing models that use full attentions have quadratic computational and memory complexities, and are too costly for long documents. |
| Approach: | They propose an efficient encoder-decoder attention with head-wise positional strides to effectively pinpoint salient information from the source. |
| Outcome: | The proposed model can process ten times more tokens than current models that use full attentions. |
LGESQL: Line Graph Enhanced Text-to-SQL Model with Mixed Local and Non-Local Relations (2021.acl-long)
Copied to clipboard
| Challenge: | Existing methods to encode text-to-SQL data are node-centric and ignore semantics embedded in the topological structure of edges. |
| Approach: | They propose a Line Graph Enhanced Text-to-SQL model to mine relational features without constructing meta-paths. |
| Outcome: | The proposed model achieves state-of-the-art on the cross-domain text-to-SQL benchmark Spider at the time of writing. |
MinerU2.5: A Decoupled Vision-Language Model for Efficient High-Resolution Document Parsing (2026.acl-industry)
Copied to clipboard
Junbo Niu, Zheng Liu, Zhuangcheng Gu, Bin Wang, Linke Ouyang, Zhiyuan Zhao, Tao Chu, Tianyao He, Fan Wu, Qintong Zhang, Zhenjiang Jin, Guang Liang, Rui Zhang, Wenzheng Zhang, Yuan Qu, Zhifei Ren, Yuefeng Sun, Zirui Tang, Boyu Niu, Yuanhong Zheng, Dongsheng Ma, Ziyang Miao, Hejun Dong, Siyi Qian, Junyuan Zhang, Fangdong Wang, Jingzhou Chen, Xiaomeng Zhao, Liqun Wei, Wei Li, Shasha Wang, RuiLiang Xu, Yuanyuan Cao, Lu Chen, Qianqian Wu, Huaiyu Gu, Lindong Lu, Dechen Lin, null Shenguanlin, Xuanhe Zhou, Linfeng Zhang, Yuhang Zang, Xiaoyi Dong, Jiaqi Wang, Bo Zhang, Lei Bai, Pei Chu, Weijia Li, Jiang Wu, Lijun Wu, Zhenxiang Li, Guangyu Wang, Zhongying Tu, Chao Xu, Kai Chen, Bowen Zhou, Dahua Lin, Wentao Zhang, Conghui He
| Challenge: | Document images are characterized by higher resolutions, denser content, and more complex structural layouts. |
| Approach: | They propose a 1.2B-parameter document parsing vision-language model that decouples layout analysis from local content recognition. |
| Outcome: | The proposed model surpasses general-purpose and domain-specific models on multiple benchmarks while maintaining significantly lower computational overhead. |
Quantifying and Mitigating Unimodal Biases in Multimodal Large Language Models: A Causal Perspective (2024.findings-emnlp)
Copied to clipboard
| Challenge: | Recent advances in Large Language Models have facilitated the development of Multimodal LLMs. |
| Approach: | They propose a causal framework to interpret unimodal biases in visual question answering problems and a framework to integrate information from different modalities and mitigate biase. |
| Outcome: | The proposed framework analyzes visual question answering (VQA) problems to assess their impact on predictions. |
Multilingual Knowledge Graph Completion with Self-Supervised Adaptive Graph Alignment (2022.acl-long)
Copied to clipboard
Zijie Huang, Zheng Li, Haoming Jiang, Tianyu Cao, Hanqing Lu, Bing Yin, Karthik Subbian, Yizhou Sun, Wei Wang
| Challenge: | Existing methods to predict missing facts in knowledge graphs are limited in language alignment . SS-AGA uses seed alignment as an edge type to fuses all KGs as a whole graph . |
| Approach: | They propose a self-supervised adaptive graph alignment method that fuses all KGs as a whole graph by regarding alignment as 'a new edge type' they propose SS-AGA method that uses relation-aware attention weights to capture potential alignment pairs in a new paradigm. |
| Outcome: | The proposed method can predict missing facts in a knowledge graph (KG) but language alignment is scarce and new alignment identification is noisy. |
Putting Captions to the Test: Evaluating Video Caption Quality through Multiple-Choice Question Answering (2026.acl-long)
Copied to clipboard
| Challenge: | Existing metrics for video captioning are based on text-based comparisons with ground-truth references. |
| Approach: | They propose a reference-free benchmark that assesses video captions based on their utility . they will release the benchmark to facilitate reproducible research . |
| Outcome: | The proposed benchmark improves on human-verified, fine-grained questions . it correlates significantly better with human judgments than existing metrics . |
TIE: Topological Information Enhanced Structural Reading Comprehension on Web Pages (2022.naacl-main)
Copied to clipboard
| Challenge: | Existing models for structural reading comprehension (SRC) only focus on comprehension of plain text, tables, tables or knowledge bases. |
| Approach: | They propose a topological information enhanced model which transforms a token-level task into a tag-level one by introducing a two-stage process. |
| Outcome: | The proposed model outperforms baselines and achieves state-of-the-art performance on the web-based SRC benchmark WebSRC at the time of writing. |
Memorizing is Not Enough: Deep Knowledge Injection Through Reasoning (2025.acl-long)
Copied to clipboard
Ruoxi Xu, Yunjie Ji, Boxi Cao, Yaojie Lu, Hongyu Lin, Xianpei Han, Ben He, Yingfei Sun, Xiangang Li, Le Sun
| Challenge: | Existing knowledge injection frameworks focus on knowledge memorization and retrieval, but static nature of large language models leads to outdated information as the real world evolves or when adapting to domain-specific knowledge. |
| Approach: | They propose a four-tier knowledge injection framework that defines the levels of knowledge injection: memorization, retrieval, reasoning, and association. |
| Outcome: | The proposed framework defines the levels of knowledge injection: memorization, retrieval, reasoning, and association. |
ShadowGNN: Graph Projection Neural Network for Text-to-SQL Parser (2021.naacl-main)
Copied to clipboard
| Challenge: | Existing semantic parsing models struggle to adapt to unseen database schemas . a new architecture, ShadowGNN, processes schemas at abstract and semantic levels . |
| Approach: | They propose a new architecture which processes schemas at abstract and semantic levels. |
| Outcome: | The proposed architecture outperforms state-of-the-art models on a text-to-sql benchmark . it uses domain-independent representations to extract logical linking between question and schema . |
D-Artemis: A Deliberative Cognitive Framework for Mobile GUI Multi-Agents (2026.findings-acl)
Copied to clipboard
Hongze Mi, Yibo Feng, WenJie Lu, Yuqi Wang, Jinyuan Li, Song Cao, He Cui, Tengfei Tian, Xuelin Zhang, Haotian Luo, Di Sun, Jun Fang, Hua Chai, Naiqiang Tan, Gang Pan
| Challenge: | Graphical User Interface (GUI) agents aim to automate a wide spectrum of human tasks by emulating user interaction. |
| Approach: | They propose a deliberative framework that leverages a fine-grained tip retrieval mechanism to inform its decision-making process. |
| Outcome: | The proposed framework achieves SOTA among open-source general models on AndroidWorld and ScreenSpot-V2 . it leverages a fine-grained, app-specific tip retrieval mechanism to inform its decision-making process . |
ParaCook: On Time-Efficient Planning for Multi-Agent Systems (2026.findings-acl)
Copied to clipboard
Shiqi Zhang, Xinbei Ma, Yunqing Xu, Zouying Cao, Pengrui Lu, Haobo Yuan, Tiancheng Shen, Zhuosheng Zhang, Hai Zhao, Ming-Hsuan Yang
| Challenge: | Existing agent benchmarks focus on task completion while neglecting time efficiency in parallel and asynchronous operations. |
| Approach: | They propose a framework for large language models that allows agents to plan long-horizon tasks in a scalable way. |
| Outcome: | The proposed framework is based on the Overcooked game and can be used to evaluate time efficiency-aware multi-agent planning. |
Critic-CoT: Boosting the Reasoning Abilities of Large Language Model via Chain-of-Thought Critic (2025.findings-acl)
Copied to clipboard
Xin Zheng, Jie Lou, Boxi Cao, Xueru Wen, Yuqiu Ji, Hongyu Lin, Yaojie Lu, Xianpei Han, Debing Zhang, Le Sun
| Challenge: | Existing approaches to improve the reasoning performance of large language models rely on intuitive instance-level feedback, which limits the reasoning capabilities. |
| Approach: | They propose a framework that pushes LLMs toward System-2-like critic capability by using a step-wise CoT reasoning paradigm and automatic construction of weak-supervision data without human annotation. |
| Outcome: | The proposed model significantly improves task-solving performance by filtering out invalid solutions or iterative refinement. |
JoPA: Explaining Large Language Model’s Generation via Joint Prompt Attribution (2025.acl-long)
Copied to clipboard
| Challenge: | Existing attempts to explain the entire language generation often treat input prompt texts independently, ignoring their combinatorial effects on the follow-up generation. |
| Approach: | They propose a framework for explaining how a few prompt texts collaboratively influences the LLM's complete generation. |
| Outcome: | The proposed explanations demonstrate faithfulness and efficiency of the proposed framework. |
Retaining Key Information under High Compression Ratios: Query-Guided Compressor for LLMs (2024.acl-long)
Copied to clipboard
| Challenge: | Existing methods to compress long contexts have degraded dramatically as compression ratios increase, sometimes even falling to the closed-book level. |
| Approach: | They propose a query-guided compression method that preserves key information within the compressed context. |
| Outcome: | The proposed method can consistently perform well even at high compression ratios, and offers significant benefits in terms of inference cost and throughput. |
A Large Scale Speech Sentiment Corpus (2020.lrec-1)
Copied to clipboard
| Challenge: | Existing corpus for sentiment analysis uses text inputs, but voice inputs are becoming more important as smart assistants and mobile voice control become more prevalent. |
| Approach: | They propose to extend the Switchboard-1 Telephone Speech Corpus by adding sentiment labels from 3 different human annotators for every transcript segment. |
| Outcome: | The proposed corpus contains 49500 labeled speech segments covering 140 hours of audio. |
BUMP: A Benchmark of Unfaithful Minimal Pairs for Meta-Evaluation of Faithfulness Metrics (2023.acl-long)
Copied to clipboard
Liang Ma, Shuyang Cao, Robert L Logan IV, Di Lu, Shihao Ran, Ke Zhang, Joel Tetreault, Alejandro Jaimes
| Challenge: | Existing benchmarks measure the correlation with human judgements of faithfulness on model-generated summaries, but they are insufficient for diagnosing whether metrics are consistent, effective on human-written texts, and sensitive to different error types. |
| Approach: | They propose to use unfaithful minimal pairs to measure the consistency of automatic faithfulness metrics by comparing human-written summary pairs with a dataset of 889 human-writing, minimally different summary pairs. |
| Outcome: | The proposed benchmarks show that the most discriminative metrics tend not to be the most consistent, and that the best performing metrics are sensitive to errors. |
Which Sense Dominates Multisensory Semantic Understanding? A Brain Decoding Study (2024.lrec-main)
Copied to clipboard
| Challenge: | Decoding semantic meanings from brain activity is open to multisensory stimulation, as word meanings can be delivered by both auditory and visual inputs. |
| Approach: | They aim to develop a computational model to probing what information from the act of language understanding is represented in human brain. |
| Outcome: | The proposed model dissociates multisensory integration of word understanding into written text, spoken text and image perception respectively, exploring the decoding efficiency and reliability of unisensory information in the brain representation. |
STORM-BORN: A Challenging Mathematical Derivations Dataset Curated via a Human-in-the-Loop Multi-Agent Framework (2025.findings-acl)
Copied to clipboard
Wenhao Liu, Zhenyi Lu, Xinyu Hu, Jerry Zhang, Dailin Li, Jiacheng Cen, Huilin Cao, Haiteng Wang, Yuhan Li, Xie Kun, Dandan Li, Pei Zhang, Chengbo Zhang, Yuxiang Ren, Xiaohong Huang, Yan Ma
| Challenge: | Existing datasets suffer from outdated and insufficient challenging content, neglecting human-like reasoning, and limited reliability due to single-LLM generation. |
| Approach: | They propose a human-in-the-loop, multi-agent data generation framework that integrates reasoning-dense filters, multiagent collaboration, and human mathematicians’ evaluations to ensure the reliability and quality of the dataset. |
| Outcome: | The proposed framework improves accuracy and quality of the 2,000-synthesized datasets by integrating reasoning-dense filters, multi-agent collaboration, and human mathematicians’ evaluations. |
WordGame: Efficient & Effective LLM Jailbreak via Simultaneous Obfuscation in Query and Response (2025.findings-naacl)
Copied to clipboard
| Challenge: | Recent advances in large language models have raised concerns about their susceptibility to jailbreaking attacks, which leads to harmful content inadvertently. |
| Approach: | They propose to exploit the safety alignment patterns of LLMs by simultaneous obfuscation in queries and responses to break down adversarial intent of query. |
| Outcome: | The proposed attack breaks down adversarial intent of query and encourages benign content regarding the games to precede anticipated harmful content in the response. |
Mulan: A Multi-Level Alignment Model for Video Question Answering (2023.findings-emnlp)
Copied to clipboard
| Challenge: | Existing methods focus on visual-language alignment at the video level, but they do not account for fine-grained semantic interaction between video and text. |
| Approach: | They propose a multi-level Alignment Model for Video Question Answering that establishes alignment between visual and textual modalities at the object-level, frame-level and video-level. |
| Outcome: | The proposed model outperforms state-of-the-art methods even with a small amount of extra visual-language pre-training data and a reduced number of trainable parameters. |
Line Graph Enhanced AMR-to-Text Generation with Mix-Order Graph Attention Networks (2020.acl-main)
Copied to clipboard
| Challenge: | Existing graph-to-sequence approaches use graph neural networks as encoders, but they lack the structure information needed to translate AMR into the graph-based data. |
| Approach: | They propose a graph-to-sequence task which aims to recover natural language from Abstract Meaning Representations (AMR) they adopt graph attention networks with higher-order neighborhood information to explore the edge relations in AMR graphs. |
| Outcome: | The proposed framework achieves state-of-the-art performance on English AMR benchmark datasets and is able to translate the AMR semantics into the natural language. |
Across Programming Language Silos: A Study on Cross-Lingual Retrieval-Augmented Code Generation (2026.findings-acl)
Copied to clipboard
Qiming Zhu, Jialun Cao, Xuanang Chen, Weili Zhang, Yaojie Lu, Hongyu Lin, Xianpei Han, Le Sun, Shing-Chi Cheung
| Challenge: | Current research on large language models with retrieval-augmented code generation (RACG) has focused on single-language settings, leaving their cross-lingual effectiveness underexplored. |
| Approach: | They construct a dataset covering 13 PLs with nearly 14K instances to study cross-lingual code knowledge transfer in RACG. |
| Outcome: | The proposed model shows unequal cross-lingual knowledge transfer even with direct injection and shows limited reliance on natural language information embedded in code when equipped with a code-specific retriever. |
CLUE: A Chinese Language Understanding Evaluation Benchmark (2020.coling-main)
Copied to clipboard
Liang Xu, Hai Hu, Xuanwei Zhang, Lu Li, Chenjie Cao, Yudong Li, Yechen Xu, Kai Sun, Dian Yu, Cong Yu, Yin Tian, Qianqian Dong, Weitang Liu, Bo Shi, Yiming Cui, Junyi Li, Jun Zeng, Rongzhao Wang, Weijian Xie, Yanting Li, Yina Patterson, Zuoyu Tian, Yiwen Zhang, He Zhou, Shaoweihua Liu, Zhe Zhao, Qipeng Zhao, Cong Yue, Xinrui Zhang, Zhengliang Yang, Kyle Richardson, Zhenzhong Lan
| Challenge: | Existing language evaluation benchmarks for English are limited to English . lack of such benchmarks makes it difficult to replicate success in other languages . |
| Approach: | They introduce a large-scale Chinese language understanding evaluation benchmark . the benchmark uses a set of current state-of-the-art pre-trained Chinese models . |
| Outcome: | The first large-scale Chinese Language Understanding Evaluation (CLUE) benchmark is released . the benchmark evaluates models across a wide range of tasks on original Chinese text . existing language evaluation benchmarks are mostly limited to English . |
ACT-SQL: In-Context Learning for Text-to-SQL with Automatically-Generated Chain-of-Thought (2023.findings-emnlp)
Copied to clipboard
| Challenge: | Recent studies have focused on the development of semantic parsers within the framework of cross-domain analysis. |
| Approach: | They propose a method to generate auto-CoT exemplars using ACT-SQL and extend it to multi-turn text-to-Sql tasks. |
| Outcome: | The proposed method achieves SOTA performance on the Spider dev set among existing in-context learning approaches. |
Visual Prompt Tuning for Few-Shot Text Classification (2022.coling-1)
Copied to clipboard
| Challenge: | Existing work on pretraining models for text classification uses image encoders instead of visual prompts. |
| Approach: | They propose a method to deploy large-scale pre-trained models in the prompt-tuning paradigm in few-shot learning. |
| Outcome: | The proposed method outperforms the most recent prompt-tuning methods on five public text classification datasets. |
RSVP: Reasoning Segmentation via Visual Prompting and Multi-modal Chain-of-Thought (2025.acl-long)
Copied to clipboard
Yi Lu, Jiawang Cao, Yongliang Wu, Bozheng Li, Licheng Tang, Yangguang Ji, Chong Wu, Jay Wu, Wenbo Zhu
| Challenge: | Recent advances in multi-modal learning have enhanced MLLMs' ability to reason about visual content. |
| Approach: | They propose a framework that unifies multi-step multimodal reasoning with grounded visual understanding. |
| Outcome: | The proposed framework surpasses state-of-the-art methods by +6.5 gIoU and +9.2 cIou on ReasonSeg and achieves 49.7 mAP on SegInW under zero-shot settings. |
Controllable Open-ended Question Generation with A New Question Type Ontology (2021.acl-long)
Copied to clipboard
| Challenge: | Existing question types are limited to generating multiple-sense questions . we present a question type-aware question generation framework to generate open-ended questions based on multiple-phrase questions - a task that is less explored . |
| Approach: | They propose a question type-aware question generation framework which predicts question focuses and produces the question. |
| Outcome: | The proposed model improves question quality over competitive comparisons on large-scale datasets. |
When Models Outthink Their Safety: Unveiling and Mitigating Self-Jailbreak in Large Reasoning Models (2026.findings-acl)
Copied to clipboard
Yingzhi Mao, Chunkang Zhang, Junxiang Wang, Xinyan Guan, Boxi Cao, Yaojie Lu, Hongyu Lin, Xianpei Han, Le Sun
| Challenge: | Existing methods often apply coarse-grained constraints over entire reasoning trajectories . Existing approaches often apply unsafe constraints, causing unsafe outputs . |
| Approach: | They propose a trajectory-level training framework that mitigates Self-Jailbreak . they propose 'chain-of-guardrail' to mitigate self-jailbreak by targeting step-level interventions . |
| Outcome: | The proposed framework mitigates Self-Jailbreak by targeting step-level interventions while maintaining reasoning ability. |
NeuSym-RAG: Hybrid Neural Symbolic Retrieval with Multiview Structuring for PDF Question Answering (2025.acl-long)
Copied to clipboard
Ruisheng Cao, Hanchong Zhang, Tiancheng Huang, Zhangyi Kang, Yuxin Zhang, Liangtai Sun, Hanqi Li, Yuxun Miao, Shuai Fan, Lu Chen, Kai Yu
| Challenge: | Existing approaches to retrieval augmented generation neglect PDF structure and layout . individual PDFs often exceed prompt limits and user queries may span multiple documents. |
| Approach: | They propose a hybrid neural symbolic retrieval framework which combines both paradigms in an interactive process. |
| Outcome: | The proposed framework organizes semi-structured PDF content into relational database and vectorstore . it defeats both RAG and structured baselines on three PDF-based QA datasets . |
HisDoc-OCR: Restoring Visual Grounding in MLLMs for Chinese Historical Document OCR (2026.findings-acl)
Copied to clipboard
| Challenge: | Despite multimodal large language models' strong performance on modern document OCR, their application to historical Chinese texts suffers from severe hallucinations, character fabrication, uncontrolled repetition, and semantic drift. |
| Approach: | They propose a multimodal large language model which restores visual grounding through three synergistic strategies: Layout Injection, First-Occurrence Boost, Self-Distilled Attention Focusing and HisDoc-OCR. |
| Outcome: | The proposed model outperforms general-purpose and OCR-specific models on Chinese historical documents. |
The Linguistic Connectivities Within Large Language Models (2025.findings-acl)
Copied to clipboard
Dan Wang, Boxi Cao, Ning Bian, Xuanang Chen, Yaojie Lu, Hongyu Lin, Jia Zheng, Le Sun, Shanshan Jiang, Bin Dong, Xianpei Han
| Challenge: | Recent studies have discovered notable disparities in their performance across different languages. |
| Approach: | They conduct a systematic investigation into the behaviors of large language models across 27 different languages on 3 different scenarios and reveals a Linguistic Map correlates with the richness of available resources and linguistic family relations. |
| Outcome: | The proposed model demonstrates that there are significant disparities in performance across languages across 27 different languages on 3 different scenarios. |
MMAU: A Holistic Benchmark of Agent Capabilities Across Diverse Domains (2025.findings-naacl)
Copied to clipboard
Guoli Yin, Haoping Bai, Shuang Ma, Feng Nan, Yanchao Sun, Zhaoyang Xu, Shen Ma, Jiarui Lu, Xiang Kong, Aonan Zhang, Dian Ang Yap, Yizhe Zhang, Karsten Ahnert, Vik Kamath, Mathias Berglund, Dominic Walsh, Tobias Gindele, Juergen Wiest, Zhengfeng Lai, Xiaoming Simon Wang, Jiulong Shan, Meng Cao, Ruoming Pang, Zirui Wang
| Challenge: | Existing benchmarks focus on specific application scenarios, emphasizing task completion but failing to dissect the underlying skills that drive these outcomes. |
| Approach: | They propose a Massive Multitask Agent Understanding benchmark that evaluates LLMs across five domains and offline tasks. |
| Outcome: | The Massive Multitask Agent Understanding (MMAU) benchmark evaluates models across five domains including Tool-use, Directed Acyclic Graph (DAG) QA, Data Science and Machine Learning coding, Contest-level programming and Mathematics. |
Direct Large Language Model Alignment Through Self-Rewarding Contrastive Prompt Distillation (2024.acl-long)
Copied to clipboard
| Challenge: | Existing methods to evaluate preference data without human annotations are difficult . et al., 2022b) is effective for aligning large language models with human expectations . |
| Approach: | They propose a method to evaluate the response preference using output probabilities under contrastive prompts. |
| Outcome: | The proposed method could surpass the RLHF method without human-annotated preference data. |
Attention Head Masking for Inference Time Content Selection in Abstractive Summarization (2021.naacl-main)
Copied to clipboard
| Challenge: | Existing studies show that multi-heads attentions at the same layer collectively guide the summarization. |
| Approach: | They propose an inference-time attention head masking mechanism that works on encoder-decoder attentions to pinpoint salient content at inference time. |
| Outcome: | The proposed technique outperforms state-of-the-art models on CNN/DailyMail and New York Times datasets and is data-efficient. |
Decoupled Dialogue Modeling and Semantic Parsing for Multi-Turn Text-to-SQL (2021.findings-acl)
Copied to clipboard
| Challenge: | Recent work on Text-to-SQL for multi-turn dialogue has attracted great interest . current approaches mostly employ end-to end models and face data sparsity problems . |
| Approach: | They propose a decoupled multi-turn text-to-SQL framework where dialogue context is explicitly solved by an utterance rewrite model and a single-turn Text-toSQl parser are proposed. |
| Outcome: | The proposed method outperforms existing models on SParC and CoSQL datasets without annotated in-domain data. |
Verifiable Generation with Subsentence-Level Fine-Grained Citations (2024.findings-acl)
Copied to clipboard
| Challenge: | Existing work mainly targets the generation of sentence-level citations, lacking specificity about which parts of a sentence are backed by the cited sources. |
| Approach: | They propose to use subsentence-level fine-grained citations to generate more precise location of generated content supported by the cited sources. |
| Outcome: | The proposed model improves the accuracy and trustworthiness of large language models by allowing users to trace the information back to its source and verify its correctness. |
MoleculeQA: A Dataset to Evaluate Factual Accuracy in Molecular Comprehension (2024.findings-emnlp)
Copied to clipboard
| Challenge: | Existing models generate erroneous information and evaluations fail to assess factual correctness of models. |
| Approach: | They propose to use MoleculeQA to evaluate molecular factual correctness in large language models by organizing molecules into a taxonomy and building QA pairs through human and LLM efforts. |
| Outcome: | The proposed model improves the factual correctness of generated information and enables the development of new models. |
MultiFinBen: Benchmarking Large Language Models for Multilingual and Multimodal Financial Application (2026.acl-long)
Copied to clipboard
Xueqing Peng, Lingfei Qian, Yan Wang, Ruoyu Xiang, Yueru He, Yang Ren, Mingyang Jiang, Vincent Jim Zhang, Yuqing Guo, Jeff Zhao, Huan He, Yi Han, Yun Feng, Yuechen Jiang, Yupeng Cao, Haohang Li, Yangyang Yu, Xiaoyu Wang, Penglei Gao, Shengyuan Lin, Keyi Wang, Shanshan Yang, Yilun Zhao, Zhiwei Liu, Peng Lu, Jerry Huang, Suyuchen Wang, Triantafillos Papadopoulos, Polydoros Giannouris, Efstathia Soufleri, Nuo Chen, Zhiyang Deng, Heming Fu, Yijia Zhao, Mingquan Lin, Meikang Qiu, Kaleb E Smith, Arman Cohan, Xiao-Yang Liu, Jimin Huang, Guojun Xiong, Alejandro Lopez-Lira, Xi Chen, Junichi Tsujii, Jian-Yun Nie, Sophia Ananiadou, Qianqian Xie
| Challenge: | Existing evaluations of LLMs in finance are text-only, monolingual, and largely saturated by current models. |
| Approach: | They propose a multilingual and multimodal benchmark for evaluating LLMs in real financial contexts. |
| Outcome: | The first expert-annotated multilingual and multimodal benchmark is released . it evaluates 21 leading LLMs and shows they perform better in multilingual settings . |
Defending Against Alignment-Breaking Attacks via Robustly Aligned LLM (2024.acl-long)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have made significant advancements but can be misused to generate harmful content. |
| Approach: | They propose a Robustly Aligned LLM to defend against alignment-breaking attacks by retraining existing LLMs and using adversarial or handcrafted jailbreaking prompts. |
| Outcome: | The proposed model reduces attack success rates from nearly 100% to around 10% or less. |
Adam’s Law: Textual Frequency Law on Large Language Models (2026.acl-long)
Copied to clipboard
| Challenge: | Textual frequency is a topic of understudied research, but its relevance to Large Language Models is not well understood. |
| Approach: | They propose a framework to estimate textual data frequency using a paraphraser and a textual distillation method to refine LLMs. |
| Outcome: | The proposed framework can be used to estimate sentence-level frequency with word-level frequencies. |
Task-Related In-Context Learning (2026.findings-acl)
Copied to clipboard
| Challenge: | Standard in-context learning assumes identical output spaces between test and retrieval datasets . however, in practice, these datasets can be fully aligned, partially alignes, or fully disjoint in label space . |
| Approach: | They propose a framework for in-context learning under output-space mismatch . they identify demonstrations relevant to the test label space via a Bayesian probabilistic criterion . |
| Outcome: | The proposed framework achieves state-of-the-art results across three LLMs, three task types, and four datasets. |
RedOne 2.0: Rethinking Domain-specific LLM Post-Training in Social Networking Services (2026.acl-industry)
Copied to clipboard
Fei zhao, Chonggang Lu, Haofu Qian, Fangcheng Shi, Zijie Meng, Jianzhao Huang, Zheyong Xie, Shaosheng Cao
| Challenge: | Social networking services (SNS) are critical infrastructure for global interaction . supervised fine-tuning (SFT) can improve in-domain performance, but it often induces a ”seesaw” trade-off with out-of-domain robustness . |
| Approach: | They propose an SNS-oriented LLM with a progressive, RL-prioritized post-training paradigm for fast and stable adaptation. |
| Outcome: | The proposed model improves over the previous 7B model by 2.41 on average . it also yields an 8.74 average gain over its Qwen3-4B base . |
Inference Time Style Control for Summarization (2021.naacl-main)
Copied to clipboard
| Challenge: | Existing methods to generate summaries of different styles without training separate models are lacking parallel data and expensive (re)training. |
| Approach: | They propose two methods that can be deployed during summary decoding on any pre-trained Transformer-based summarization model. |
| Outcome: | The proposed methods generate news headlines with various ideological leanings while still informative. |
AscendKernelGen: LLM-Driven Kernel Generation for NPUs (2026.findings-acl)
Copied to clipboard
Xinzi Cao, Jianyang Zhai, Pengfei Li, Zhiheng Hu, Cen Yan, null Mubingxu, Guanghuan Fang, Bin She, Jiayu Li, Yihan Su, Dongyang Tao, Feidiao Yang, Chang-Dong Wang, Yutong Lu, Weicheng Xue, Bin Zhou, Yonghong Tian
| Challenge: | Neural Processing Units (NPUs) are critical for AI infrastructure, but their development remains a bottleneck due to vendor-specific Domain-Specific Languages (DSLs). |
| Approach: | They propose a framework for NPU kernel development that bridges the gap in hardware-specific coding . compiler success on complex Level-2 kernels improves from 0% to 95.5%, they say . |
| Outcome: | The proposed framework bridges the gap in hardware-specific coding, showing a near-zero success rate on complex kernels. |
CROSSAGENTIE: Cross-Type and Cross-Task Multi-Agent LLM Collaboration for Zero-Shot Information Extraction (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models struggle with producing structured output while maintaining accuracy in zero-shot information extraction (IE) |
| Approach: | They propose a multi-agent framework that enhances zero-shot IE through multi-task collaboration. |
| Outcome: | CROSSAGENTIE outperforms state-of-the-art models in structured prediction . the framework significantly reduces inference cost while preserving accuracy . |
Phi: Preference Hijacking in Multi-modal Large Language Models at Inference Time (2025.emnlp-main)
Copied to clipboard
| Challenge: | Recent advances in Multimodal Large Language Models have raised serious safety concerns. |
| Approach: | They propose a method for manipulating the output preference of MLLMs using a preference hijacked image. |
| Outcome: | The proposed method works at inference time and requires no model modifications. |
SYNC: A Synthetic Long-Context Understanding Benchmark for Controlled Comparisons of Model Capabilities (2025.emnlp-main)
Copied to clipboard
| Challenge: | Existing synthetic tasks target narrow skill sets, limiting their ability to comprehensively assess model capabilities. |
| Approach: | They propose a new evaluation suite of synthetic tasks spanning domains including graph understanding and translation that test a wide range of capabilities. |
| Outcome: | The evaluation suite of synthetic tasks spanning domains including graph understanding and translation shows that the tasks perform significantly better on more challenging tasks. |
RedOne: Revealing Domain-specific LLM Post-Training in Social Networking Services (2025.emnlp-industry)
Copied to clipboard
Fei Zhao, Chonggang Lu, null Wangyue, Zheyong Xie, Ziyan Liu, Haofu Qian, Jianzhao Huang, Fangcheng Shi, Zijie Meng, Hongcheng Guo, Mingqian He, Xinze Lyu, Zheyu Ye, Weiting Liu, Boyang Wang, Shaosheng Cao
| Challenge: | Social networking services (SNS) have experienced rapid growth, which has proposed significant challenges for platform content management and interaction quality improvement. |
| Approach: | They propose a domain-specific LLM to break the performance bottleneck of single-task baselines and establish a comprehensive foundation for social networking services. |
| Outcome: | The proposed model achieves an average improvement of 14.02% across 8 major tasks and 7.56% in bilingual evaluation benchmark, compared with baseline models. |
Multi-View Source Ablation for Faithful Summarization (2023.findings-eacl)
Copied to clipboard
| Challenge: | MuFaSSa is a metric for evaluating faithfulness of abstractive summaries . it uses different strategies to remove information from source document to form multiple ablated views . |
| Approach: | They propose a metric for evaluating faithfulness of abstractive summaries using multiple ablated views. |
| Outcome: | The proposed metric outperforms existing models on summarization tasks and human-annotated faithfulness labels. |
Runaway is Ashamed, But Helpful: On the Early-Exit Behavior of Large Language Model-based Agents in Embodied Environments (2025.findings-emnlp)
Copied to clipboard
| Challenge: | Experiments with 4 different LLMs across 5 embodied environments show significant efficiency improvements, with only minor drops in agent performance. |
| Approach: | They propose an intrinsic method that injects exit instructions during generation and an extransic system that verifies task completion to determine when to halt an agent’s trial. |
| Outcome: | The proposed method injects exit instructions during generation and an exit method verifies task completion to determine when to halt an agent’s trial. |
Goal-oriented Vision-and-Dialog Navigation via Reinforcement Learning (2022.findings-emnlp)
Copied to clipboard
| Challenge: | Existing methods for vision-and-dialog navigation are limited and do not adapt to domain changes. |
| Approach: | They propose a problem where an agent computes dialog-navigation policies from trial and error. |
| Outcome: | The proposed agent outperforms baselines in success rate in photo-realistic simulations. |
TEBNER: Domain Specific Named Entity Recognition with Type Expanded Boundary-aware Network (2021.emnlp-main)
Copied to clipboard
| Challenge: | Existing methods to label data and identify entities require large amounts of manually annotated texts for training supervised models. |
| Approach: | They propose a dictionary extension method which extracts new entities through the type expanded model. |
| Outcome: | The proposed method outperforms state-of-the-art supervised systems on different types of datasets and surpasses supervised models. |
From Informal to Formal – Incorporating and Evaluating LLMs on Natural Language Requirements to Verifiable Formal Proofs (2025.acl-long)
Copied to clipboard
Jialun Cao, Yaojie Lu, Meiziniu Li, Haoyang Ma, Haokun Li, Mengda He, Cheng Wen, Le Sun, Hongyu Zhang, Shengchao Qin, Shing-Chi Cheung, Cong Tian
| Challenge: | Recent studies in formal mathematical reasoning have shown an unstoppable growth trend. |
| Approach: | They constructed 18k high-quality instruction-response pairs across five mainstream formal specification languages and evaluated them against ten open-sourced LLMs. |
| Outcome: | The proposed model compared instruction-response pairs across five formal specification languages and found that the LLMs were good at writing proof segments when given either the code, or the detailed description of proof steps. |
GeoEdit: Geometric Knowledge Editing for Large Language Models (2025.emnlp-main)
Copied to clipboard
Yujie Feng, Li-Ming Zhan, Zexin Lu, Yongxin Xu, Xu Chu, Yasha Wang, Jiannong Cao, Philip S. Yu, Xiao-Ming Wu
| Challenge: | Existing training-based model editing methods struggle to incorporate new knowledge while preserving unrelated general knowledge. |
| Approach: | They propose a framework that uses geometric relationships to differentiate between neurons associated with new knowledge updates and those related to general knowledge perturbations. |
| Outcome: | The proposed framework avoids updating neurons with directions approximately orthogonal to existing knowledge, thus preserving the model’s generalization ability. |
Time-aware Prompting for Text Generation (2022.findings-emnlp)
Copied to clipboard
| Challenge: | a new study investigates the effects of incorporating timestamps into generation systems . textual prompts focus more on non-temporal information and are less sensitive to given timestams . |
| Approach: | They propose a data-to-text generation dataset that includes chronologically ordered revisions of biographical articles from English Wikipedia. |
| Outcome: | The proposed models improve the quality of the data-to-text generation dataset TempWikiBio . the proposed models are more sensitive to time-aware prompts than textual prompts . |
iPET: An Interactive Emotional Companion Dialogue System with LLM-Powered Virtual Pet World Simulation (2025.acl-demo)
Copied to clipboard
Zheyong Xie, Shaosheng Cao, Zuozhu Liu, Zheyu Ye, Zihan Niu, Chonggang Lu, Tong Xu, Enhong Chen, Zhe Xu, Yao Hu, Wei Lu
| Challenge: | Existing approaches to role-playing emotional companion products lack sustained personalization and contextual adaptability, limiting their effectiveness in real-world settings. |
| Approach: | They propose a virtual pet agent that can enhance user engagement through rich, dynamic pet behaviors and interactions tailored to individual preferences. |
| Outcome: | The proposed system has been deployed in a real-world, non-commercial product for 200 days and has demonstrated its effectiveness in practical applications. |
CRUXEVAL-X: A Benchmark for Multilingual Code Reasoning, Understanding and Execution (2025.acl-long)
Copied to clipboard
Ruiyang Xu, Jialun Cao, Yaojie Lu, Ming Wen, Hongyu Lin, Xianpei Han, Ben He, Shing-Chi Cheung, Le Sun
| Challenge: | Existing code benchmarks focus on code generation, while those for code reasoning are insufficient. |
| Approach: | They propose a multi-lingual code reasoning benchmark that contains 19 programming languages and at least 600 subjects for each language. |
| Outcome: | The proposed model trains on Python and achieves 34.4% Pass@1 in other languages, revealing the cross-language generalization of LLMs. |
InstructMol: Multi-Modal Integration for Building a Versatile and Reliable Molecular Assistant in Drug Discovery (2025.coling-main)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) can attain professional-level proficiency in specific domains through fine-tuning. |
| Approach: | They propose a multi-modal LLM that aligns molecular structures with natural language via an instruction-tuning approach. |
| Outcome: | InstructMol surpasses existing models and reduces the gap with specialists in drug discovery tasks. |