Papers by Li Zeng
Verified Critical Step Optimization for LLM Agents (2026.findings-acl)
Copied to clipboard
| Challenge: | Critical Step Optimization (CSO) focuses preference learning on verified critical steps where alternative actions demonstrably flip task outcomes from failure to success. |
| Approach: | They propose a method which focuses preference learning on verified critical steps where alternative actions demonstrably flip task outcomes from failure to success. |
| Outcome: | The proposed method outperforms the existing methods on GAIA-Text-103 and XBench-DeepSearch while requiring supervision at only 16% of trajectory steps. |
POSITION BIAS MITIGATES POSITION BIAS: Mitigate Position Bias Through Inter-Position Knowledge Distillation (2025.emnlp-main)
Copied to clipboard
| Challenge: | Positional bias (PB) manifests as non-uniform sensitivity across contextual locations . previous studies have addressed PB by modifying the underlying architectures or employing extensive contextual awareness training. |
| Approach: | They propose a position-to-position knowledge distillation framework that leverages position-induced disparities to counteract PB. |
| Outcome: | The proposed framework reduces positional bias and improves performance on retrieval and reasoning tasks. |
AdaMix: Adaptive Mixing for Short and Long Reasoning Adapters (2026.acl-long)
Copied to clipboard
| Challenge: | Existing methods for large reasoning models have improved efficiency but still face limitations such as conflicting objectives and limited adaptability. |
| Approach: | They propose an adaptive reasoning framework that applies a uniform, computation-intensive deep reasoning strategy to all problems. |
| Outcome: | The proposed framework reduces the average response length of DeepSeek-R1-Distill-Qwen-7B by 54.9% while improving accuracy by up to 4.8% on five mathematical datasets. |
OmniCharacter: Towards Immersive Role-Playing Agents with Seamless Speech-Language Personality Interaction (2025.acl-long)
Copied to clipboard
Haonan Zhang, Run Luo, Xiong Liu, Yuchuan Wu, Ting-En Lin, Pengpeng Zeng, Qiang Qu, Feiteng Fang, Min Yang, Lianli Gao, Jingkuan Song, Fei Huang, Yongbin Li
| Challenge: | Existing methods focus on replicating dialogues in textual form, neglecting the role’s voice traits as a crucial effect in interaction, which tends to be more immersive experiences in realistic scenarios. |
| Approach: | They propose a first seamless speech-language personality interaction model to achieve immersive RPAs with low latency. |
| Outcome: | The proposed model exhibits role-specific personality traits and vocal traits throughout the interaction, enabling a mixture of speech and language responses. |
Searching for an Effective Defender: Benchmarking Defense against Adversarial Word Substitution (2021.emnlp-main)
Copied to clipboard
Zongyi Li, Jianhan Xu, Jiehang Zeng, Linyang Li, Xiaoqing Zheng, Qi Zhang, Kai-Wei Chang, Cho-Jui Hsieh
| Challenge: | Existing methods to defend against adversarial word-substitution attacks have not been evaluated or compared in a systematic manner. |
| Approach: | They propose to compare different defense methods under representative adversarial attacks . they propose a method that improves the robustness of neural text classifiers against such attacks a . |
| Outcome: | The proposed method improves robustness of neural text classifiers against such attacks by a significant margin. |
A Survey in Automatic Irony Processing: Linguistic, Cognitive, and Multi-X Perspectives (2022.coling-1)
Copied to clipboard
| Challenge: | figurative language research has focused on sarcasm and irony, but there is still a gap in the field. |
| Approach: | They propose to review computational irony, cognitive science, and neural models of irony processing . they aim to encourage a balanced and equal research environment in figurative languages . |
| Outcome: | The proposed multi-X irony processing perspectives will provide an overview of computational irony, insights from linguisic theory and cognitive science, and interactions with downstream NLP tasks. |
Exploring Large Language Models for Multi-Modal Out-of-Distribution Detection (2023.findings-emnlp)
Copied to clipboard
| Challenge: | Out-of-distribution (OOD) detection is essential for reliable and trustworthy machine learning. |
| Approach: | They propose to apply world knowledge to enhance OOD detection performance through selective generation from large language models (LLMs) they propose to extract visual objects from each image to fully capitalize on the aforementioned world knowledge. |
| Outcome: | The proposed method outperforms the state-of-the-art on visual OOD detection on in-distribution (ID) samples. |
DentalGPT: Incentivizing Multimodal Reasoning in Dentistry (2026.findings-acl)
Copied to clipboard
Zhenyang Cai, Jiaming Zhang, Junjie Zhao, Ziyi Zeng, Yanchao Li, Liang Jingyi, Junying Chen, Yunjin Yang, Jiajun You, Shuzhi Deng, null Xieruiqiii, Yuanting Chen, Xiangyi Feng, Jianquan Li, Liangyi Chen, Junwen Wang, Shan Jiang, Benyou Wang
| Challenge: | Current multimodal large language models (MLLMs) show limited understanding of dental images. |
| Approach: | They propose a dental-specialized multimodal large language model trained via staged multimodal alignment and reinforcement learning. |
| Outcome: | The proposed model outperforms state-of-the-art models on disease classification and dental VQA tasks. |
Cross-media Structured Common Space for Multimedia Event Extraction (2020.acl-main)
Copied to clipboard
| Challenge: | We propose a new task to extract events and their arguments from multimedia documents . traditional methods target text, images or videos, but multimedia content is distributed via multimedia . |
| Approach: | They propose a method that encodes structured representations of semantic information from textual and visual data into a common embedding space. |
| Outcome: | The proposed method achieves 4.0% and 9.8% absolute gains on text event argument role labeling and visual event extraction. |
Beyond Single-Event Extraction: Towards Efficient Document-Level Multi-Event Argument Extraction (2024.findings-acl)
Copied to clipboard
Wanlong Liu, Li Zhou, DingYi Zeng, Yichen Xiao, Shaohuan Cheng, Chen Zhang, Grandee Lee, Malu Zhang, Wenyu Chen
| Challenge: | mainstream event argument extraction methods process each event in isolation, resulting in inefficient inference and ignoring correlations among multiple events. |
| Approach: | They propose a multi-event argument argument extraction model which extracts arguments from all events simultaneously. |
| Outcome: | The proposed model performs better on four public datasets while saving time. |
Human-Inspired Obfuscation for Model Unlearning: Local and Global Strategies with Hyperbolic Representations (2025.findings-emnlp)
Copied to clipboard
| Challenge: | Existing methods for unlearning large language models struggle to balance effective forgetting with maintaining model utility. |
| Approach: | They propose a human-inspired unlearning framework that simulates forgetting on fuzzy data and represents them in hyperbolic and Euclidean spaces. |
| Outcome: | The proposed framework is able to forget sensitive content while maintaining the model’s language understanding, fluency, and benchmark performance. |
Efficient Inference for Large Vision-Language Models: Bottlenecks, Techniques, and Prospects (2026.findings-acl)
Copied to clipboard
Jun Zhang, Yicheng Ji, Feiyang Ren, Yihang Li, Bowen Zeng, Zonghao Chen, Ke Chen, Lidan Shou, Gang Chen, Huan Li
| Challenge: | Large Vision-Language Models are hindered by a systemic efficiency barrier known as visual token dominance. |
| Approach: | They propose a systematic taxonomy of efficiency techniques structured around the inference lifecycle . they examine visual encoding, prefilling, and decoding to understand bottlenecks . |
| Outcome: | The proposed techniques reveal how upstream decisions dictate downstream bottlenecks . the proposed techniques include hybrid compression and modality-aware decoding . |
PrivacyRestore: Privacy-Preserving Inference in Large Language Models via Privacy Removal and Restoration (2025.acl-long)
Copied to clipboard
| Challenge: | Existing privacy protection methods for large language models suffer from performance degradation or large inference time overhead. |
| Approach: | They propose a plug-and-play method to protect the privacy of user inputs during LLM inference . they use offline restoration vectors to train restoration vector for each privacy span type . |
| Outcome: | The proposed method can prevent the linear growth of the privacy budget. |
Entropy Scheduling in Reinforcement Learning for Large Language Models (2026.findings-acl)
Copied to clipboard
| Challenge: | entropy in reinforcement learning functions analogously to the learning rate in LLMs. |
| Approach: | They propose an entropy scheduling system that optimizes different pre-set goals by controlling and scheduling entropicy at each step of the RL process. |
| Outcome: | The proposed method improves AIME2024 from 50.9 to 54.9 within 40 training steps. |
ReviewRL: Towards Automated Scientific Review with RL (2025.emnlp-main)
Copied to clipboard
Sihang Zeng, Kai Tian, Kaiyan Zhang, Yuru Wang, Junqi Gao, Runze Liu, Sa Yang, Jingxuan Li, Xinwei Long, Jiaheng Ma, Biqing Qi, Bowen Zhou
| Challenge: | Existing automated review systems struggle with factual accuracy, rating consistency, and analytical depth. |
| Approach: | They propose a framework for generating comprehensive and factually grounded scientific paper reviews using supervised fine-tuning and reinforcement learning. |
| Outcome: | The proposed framework outperforms existing methods on ICLR 2025 papers. |
An Empirical Study of Position Bias in Modern Information Retrieval (2025.findings-emnlp)
Copied to clipboard
| Challenge: | a new evaluation framework is used to assess the extent and impact of position bias in information retrieval. |
| Approach: | They introduce a position-aware retrieval benchmark and a diagnostic metric to quantify position bias . they compare models with BM25, dense embedding models, ColBERT-style late-interaction models . |
| Outcome: | The proposed framework evaluates retrieval models for position bias from a worst-case perspective. |
From Charts to Code: A Hierarchical Benchmark for Multimodal Models (2026.acl-long)
Copied to clipboard
Jiahao Tang, Henry Hengyuan Zhao, Lijian Wu, Zijian Zhang, Yifei Tao, Dongxing Mao, Yang Wan, Jingru Tan, Min Zeng, Min Li, Alex Jinpeng Wang
| Challenge: | Chart2Code is a new benchmark for evaluating the natural language to chart code generation capabilities of large multimodal models. |
| Approach: | They introduce Chart2Code, a new benchmark for evaluating the natural language to chart code generation capabilities of large multimodal models. |
| Outcome: | The proposed benchmark is the first to scale task complexity while capturing diverse scenarios. |
LDM2: A Large Decision Model Imitating Human Cognition with Dynamic Memory Enhancement (2023.findings-emnlp)
Copied to clipboard
| Challenge: | Extensive experiments conducted in two interactive environments have shown that our LDM2 outperforms the baselines in terms of both score and success rate. |
| Approach: | They propose a large decision model with memory that leverages a dynamic memory mechanism to construct dynamic prompts, guiding the LLMs in making proper decisions according to the faced state. |
| Outcome: | The proposed model outperforms baseline models in two interactive environments in terms of score and success rate. |
Temporal Knowledge Graph Reasoning Based on N-tuple Modeling (2023.findings-emnlp)
Copied to clipboard
| Challenge: | Existing Temporal Knowledge Graphs (TKGs) only contain their core entities and form them as quadruples. |
| Approach: | They propose to describe a temporal fact more accurately as an n-tuple . they propose to use a neural network to learn evolutional representations of entities . |
| Outcome: | The proposed model oversimplifies and causes information loss on two datasets. |
OmniEvent: A Comprehensive, Fair, and Easy-to-Use Toolkit for Event Understanding (2023.emnlp-demo)
Copied to clipboard
| Challenge: | Event understanding is fundamental for humans to understand the world. |
| Approach: | They propose an event understanding toolkit called OmniEvent that is comprehensive and fair . it supports mainstream modeling paradigms and the processing of 15 widely-used datasets . |
| Outcome: | The toolkit supports mainstream modeling paradigms and the processing of 15 widely-used English and Chinese datasets. |
Towards Knowledge Checking in Retrieval-augmented Generation: A Representation Perspective (2025.naacl-long)
Copied to clipboard
Shenglai Zeng, Jiankun Zhang, Bingheng Li, Yuping Lin, Tianqi Zheng, Dante Everaert, Hanqing Lu, Hui Liu, Hui Liu, Yue Xing, Monica Xiao Cheng, Jiliang Tang
| Challenge: | Existing studies have shown that LLMs struggle to identify the boundaries of their own knowledge and tend to prioritize external information over internal knowledge learned during pre-training. |
| Approach: | They conduct a comprehensive analysis of LLM representation behaviors and demonstrate the significance of using representations in knowledge checking. |
| Outcome: | The proposed classifiers improve performance even when dealing with noisy knowledge databases. |
Data Quality Enhancement on the Basis of Diversity with Large Language Models for Text Classification: Uncovered, Difficult, and Noisy (2025.coling-main)
Copied to clipboard
| Challenge: | Existing methods for text classification based on large language models are difficult to apply directly to solve. |
| Approach: | They propose a data quality enhancement method to improve LLMs' performance in classification tasks by using a greedy algorithm to select data and then performing fine-tuning. |
| Outcome: | The proposed method improves the performance of large language models in text classification tasks and significantly improves training efficiency, saving nearly half of the training time. |
Investigating the Multilingual Calibration Effects of Language Model Instruction Tuning (2026.eacl-short)
Copied to clipboard
Jerry Huang, Peng Lu, Qiuhao Zeng, Yusuke Iwasawa, Yutaka Matsuo, Sarath Chandar, Edison Marrese-Taylor, Irene Li
| Challenge: | despite advances in foundation model research, the relationship between large language models and their calibration remains an open area of research. |
| Approach: | They examine a gap in the calibration of large language models within multilingual settings to better understand how data scarcity can potentially lead to different calibration effects. |
| Outcome: | The proposed calibration gap is found in two multilingual benchmarks over 29 and 42 languages. |
FAME: Towards Factual Multi-Task Model Editing (2024.emnlp-main)
Copied to clipboard
| Challenge: | Large language models embed extensive knowledge and perform exceptionally well across tasks. outdated knowledge or factual errors within LLMs can lead to misleading or incorrect responses. |
| Approach: | They propose to use a dataset to enhance the practicality of model editing to correct inaccurate information within LLMs. |
| Outcome: | The proposed method performs excellently across tasks and scenarios, confirming its practicality. |
Aligning Large Language Models with Implicit Preferences from User-Generated Content (2025.acl-long)
Copied to clipboard
Zhaoxuan Tan, Zheng Li, Tianyi Liu, Haodong Wang, Hyokun Yun, Ming Zeng, Pei Chen, Zhihan Zhang, Yifan Gao, Ruijie Wang, Priyanka Nigam, Bing Yin, Meng Jiang
| Challenge: | Existing preference learning methods rely heavily on curated data from humans or advanced LLMs, which is costly and difficult to scale. |
| Approach: | They propose a framework that leverages implicit preferences in unlabeled user-generated content to generate preference data. |
| Outcome: | The proposed framework transforms user-generated content into user queries and generates responses from the policy model. |
Multiple Knowledge-Enhanced Interactive Graph Network for Multimodal Conversational Emotion Recognition (2024.findings-emnlp)
Copied to clipboard
| Challenge: | Multimodal Emotion Recognition in Conversations models struggle due to lack of Common Sense Knowledge (CSK). |
| Approach: | They propose a multimodal approach to integrate multiple knowledge into the edge representations by integrating textual and visual CSK. |
| Outcome: | The proposed model outperforms state-of-the-art methods on two popular datasets. |
A Compressive Memory-based Retrieval Approach for Event Argument Extraction (2025.coling-main)
Copied to clipboard
| Challenge: | Existing retrieval-based EAE methods have input length constraints and the gap between the retriever and the inference model. |
| Approach: | They propose a retrieval-based retrieval mechanism that overcomes input length constraints . they use compressive memory to cache retrieved information and support continuous updates . |
| Outcome: | The proposed method outperforms retrieval-based methods on three public datasets. |
It’s Not Bragging If You Can Back It Up: Can LLMs Understand Braggings? (2025.acl-long)
Copied to clipboard
| Challenge: | Bragging is a pervasive social-linguistic phenomenon that reflects complex human interaction patterns. |
| Approach: | They propose to use bragging recognition, bragging explanation, and bragging generation tasks to examine bragging in large language models (LLMs) . |
| Outcome: | The proposed models can identify bragging intent, social appropriateness, and account for context sensitivity and provide new insights into how LLMs process bragging. |
MAVEN-ARG: Completing the Puzzle of All-in-One Event Understanding Dataset with Event Argument Annotation (2024.acl-long)
Copied to clipboard
Xiaozhi Wang, Hao Peng, Yong Guan, Kaisheng Zeng, Jianhui Chen, Lei Hou, Xu Han, Yankai Lin, Zhiyuan Liu, Ruobing Xie, Jie Zhou, Juanzi Li
| Challenge: | Existing datasets for event understanding have limited coverage due to complexity of tasks. |
| Approach: | They propose a dataset that augments MAVEN datasets with event argument annotations . they propose 98,591 events and 290,613 arguments obtained with laborious human annotation . |
| Outcome: | The proposed dataset is the first all-in-one dataset supporting event detection, event argument extraction, and event relation extraction. |
Cross-layer Attention Sharing for Pre-trained Large Language Models (2026.tacl-1)
Copied to clipboard
Yongyu Mu, Yuzhang Wu, Yuchun Fan, Chenglong Wang, Hengyu Li, Jiali Zeng, Qiaozhi He, Murun Yang, Fandong Meng, Jie Zhou, Tong Xiao, Jingbo Zhu
| Challenge: | Existing studies focus on compressing the Key-Value cache or grouping attention heads, while overlooking redundancy between layers. |
| Approach: | They propose a lightweight substitute for self-attention in well-trained LLMs that uses feed-forward networks to align attention heads between adjacent layers and low-rank matrices to approximate differences in layer-wise attention weights. |
| Outcome: | The proposed model reduces redundancy by sharing weights across layers while maintaining high response quality while reducing redundant calculations within 53% 84% of the total layers. |
Double Graph Based Reasoning for Document-level Relation Extraction (2020.emnlp-main)
Copied to clipboard
| Challenge: | Existing methods for document-level relation extraction fail to recognize relations between entities across sentences. |
| Approach: | They propose a method to recognize relations for long paragraphs by a Graph Aggregation-and-Inference Network (GAIN) they propose to use a heterogeneous mention-level graph and an entity-level EG graph to analyze the relationships. |
| Outcome: | The proposed method achieves a significant performance improvement (2.85 on F1) over the previous state-of-the-art. |
Mitigating Lost in Multi-turn Conversation via Curriculum RL with Verifiable Accuracy and Abstention Rewards (2026.acl-long)
Copied to clipboard
Ming Li, Pei Chen, Zhenhao Zhang, Tao Yang, Xinyang Zhang, Han Li, Tianyu Cao, Ming Zeng, Zhuofeng Wu, Meng Jiang, Huasheng Li, Lihong Li, Bing Yin
| Challenge: | Large Language Models exhibit strong capabilities in single-turn instruction following but suffer from Lost-in-Conversation (LiC) when instructions are revealed progressively in multi-turn settings, models get "Lost in Conversation" |
| Approach: | They propose a framework that encourages models to generate correct answers and judge solvability in multi-turn conversations. |
| Outcome: | The proposed framework improves models' ability to balance problem-solving with abstention . it reduces premature answering behaviors that cause lost-in-conversation (LiC) |
Conditioned Masked Language and Image Modeling for Image-Text Dense Retrieval (2022.findings-emnlp)
Copied to clipboard
| Challenge: | Large-scale two-stream pre-trained models like CLIP have achieved tremendous success in image-text retrieval. |
| Approach: | They propose a cross-modal framework for image-text retrieval using two-stream pre-trained models . they embed images and texts into instance representations with two separate encoders . experimental results on MSCOCO and Flickr30k reveal the effectiveness of their framework . |
| Outcome: | The proposed framework improves image-text retrieval performance on two popular cross-modal retrieval benchmarks. |
A Web Scale Entity Extraction System (2021.findings-emnlp)
Copied to clipboard
| Challenge: | Existing systems for large-scale entity extraction are limited by the scale and variety of data available on internet platforms. |
| Approach: | They propose to build an entity extraction system for multiple document types at large scale using multi-modal Transformers. |
| Outcome: | The proposed system extracts multiple types of entities from multiple document types at large scale using multi-modal Transformers. |
DRAMA: Dynamic Multi-Granularity Graph Estimate Retrieval over Tabular and Textual Question Answering (2024.lrec-main)
Copied to clipboard
| Challenge: | TableTextQA tasks require tabular and textual data, gaining increasing attention . however, row-based approaches suffer from limitations such as lack of interaction between rows . |
| Approach: | They propose a method that incorporates an interaction mechanism among multiple rows . Empirical results demonstrate that the proposed method is effective . |
| Outcome: | Empirical results show that the proposed model is effective on tabFact and HybridQA datasets. |
Preserving Knowledge Invariance: Rethinking Robustness Evaluation of Open Information Extraction (2023.emnlp-main)
Copied to clipboard
| Challenge: | Existing evaluation benchmarks focus on pairwise matching, ignoring robustness . current models exhibit frustrating degradation, with a maximum drop of 23.43 F1 score . |
| Approach: | They propose a benchmark that simulates the evaluation of open information extraction models in the real world . they perform experiments on typical models published in the last decade and a representative large language model . |
| Outcome: | The proposed model is rated robust on a knowledge-invariant clique with different syntactic and expressive forms. |
Focus on Your Question! Interpreting and Mitigating Toxic CoT Problems in Commonsense Reasoning (2024.acl-long)
Copied to clipboard
| Challenge: | Large language models exhibit high-level commonsense reasoning abilities, especially with enhancement methods like Chain-of-Thought (CoT). |
| Approach: | They propose a chain-of-thought-like method to elicit models' potential abilities to generate rationales and answers that are based on attribution tracing and causal tracers to probe the internal working mechanism of the LLM. |
| Outcome: | The proposed method eliminates Toxic CoT problems and improves the model’s overall commonsense reasoning performance by 5.5%. |
Context Reasoner: Incentivizing Reasoning Capability for Contextualized Privacy and Safety Compliance via Reinforcement Learning (2025.emnlp-main)
Copied to clipboard
Wenbin Hu, Haoran Li, Huihao Jing, Qi Hu, Ziqian Zeng, Sirui Han, Xu Heli, Tianshu Chu, Peizhao Hu, Yangqiu Song
| Challenge: | Current mitigation strategies fail to preserve contextual reasoning capabilities in risky scenarios, leading to systemic risks for legal compliance. |
| Approach: | They propose to use reinforcement learning with a rule-based reward to incentivize contextual reasoning capabilities while enhancing compliance with safety and privacy norms. |
| Outcome: | The proposed model outperforms Qwen2.5-7B-Instruct model in safety and privacy benchmarks and achieves +8.58% accuracy improvement. |
Type-Driven Multi-Turn Corrections for Grammatical Error Correction (2022.findings-acl)
Copied to clipboard
| Challenge: | Existing studies focus on data augmentation to combat exposure bias . but data augmented models lack the ability to recognize the procedure of gradual corrections . |
| Approach: | They propose a type-driven multi-turn corrections approach that uses multiple training instances to train dominant models. |
| Outcome: | The proposed model achieves state-of-the-art single-model performance on English GEC benchmarks. |
GTA: Supervised-Guided Reinforcement Learning for Text Classification with Large Language Models (2025.findings-emnlp)
Copied to clipboard
| Challenge: | Reinforcement learning fine-tuning methods suffer from inefficient exploration and slow convergence . supervised fine- tuning methods have limited performance ceiling and less solid theoretical foundation . |
| Approach: | They propose a Guess-Think-Answer framework that combines supervised and supervised learning in a unified training paradigm. |
| Outcome: | The proposed framework outperforms both standalone SFT and RL training models on three text classification benchmarks. |
Data Efficient RLVR via Off-Policy Influence Guidance (2026.acl-long)
Copied to clipboard
Erle Zhu, Dazhi Jiang, Yuan Wang, Xujun Li, Jiale Cheng, Yuxian Gu, Yilin Niu, Aohan Zeng, Jie Tang, Minlie Huang, Hongning Wang
| Challenge: | Existing data selection methods for RLVR are heuristic-based, lacking theoretical guarantees and generalizability. |
| Approach: | They propose an off-policy influence estimation method that approximates data influence using offline trajectories. |
| Outcome: | The proposed method reduces the computational cost of policy rollouts and improves storage and computation efficiency. |
Xiaomingbot: A Multilingual Robot News Reporter (2020.acl-demos)
Copied to clipboard
Runxin Xu, Jun Cao, Mingxuan Wang, Jiaze Chen, Hao Zhou, Ying Zeng, Yuping Wang, Li Chen, Xiang Yin, Xijin Zhang, Songcheng Jiang, Yuxuan Wang, Lei Li
| Challenge: | Xiaomingbot is a multilingual and multimodal software robot with four capabilities: news generation, news translation, news reading and avatar animation. |
| Approach: | They propose to build a multilingual and multimodal software robot with four inte- gal capabilities: news generation, news translation, news reading and avatar animation. |
| Outcome: | The proposed system generates Chinese news, then reads it in multiple languages and generates an animated avatar reading it. |
MMEvol: Empowering Multimodal Large Language Models with Evol-Instruct (2025.findings-acl)
Copied to clipboard
Run Luo, Haonan Zhang, Longze Chen, Ting-En Lin, Xiong Liu, Yuchuan Wu, Min Yang, Yongbin Li, Minzheng Wang, Pengpeng Zeng, Lianli Gao, Heng Tao Shen, Yunshui Li, Hamid Alinejad-Rokny, Xiaobo Xia, Jingkuan Song, Fei Huang
| Challenge: | a new framework for image-text instruction data evolution improves MLLM performance . lack of high-quality instruction data remains a major bottleneck in ML modeling . |
| Approach: | They propose a multimodal instruction data evolution framework that iteratively enhances data quality through fine-grained perception, cognitive reasoning, and interaction evolution. |
| Outcome: | The proposed approach improves MLLM performance in nine vision-language tasks while using significantly less data. |
Automatic Generation of Personalized Comment Based on User Profile (P19-2)
Copied to clipboard
| Challenge: | Experimental results show that our model can generate natural, human-like and personalized comments. |
| Approach: | They propose a model that takes user profile into account when generating comments on social media and integrates it with a gated memory. |
| Outcome: | The proposed model can generate natural, human-like and personalized comments on social media. |
Automated Peer Reviewing in Paper SEA: Standardization, Evaluation, and Analysis (2024.findings-emnlp)
Copied to clipboard
Jianxiang Yu, Zichen Ding, Jiaqi Tan, Kangyang Luo, Zhenmin Weng, Chenghua Gong, Long Zeng, RenJing Cui, Chengcheng Han, Qiushi Sun, Zhiyong Wu, Yunshi Lan, Xiang Li
| Challenge: | Existing approaches to review scientific papers are limited by their content or quality . SEA is a framework for automated scientific review, but its contents are generic or partial. |
| Approach: | They propose a framework for automated scientific review using large language models . they propose to use a standardized review dataset to fine-tune an LLM to generate high-quality reviews. |
| Outcome: | The proposed framework can generate high-quality reviews from standardized datasets and improves on the existing feedback mechanisms. |
Recurrent Attention for Neural Machine Translation (2021.emnlp-main)
Copied to clipboard
| Challenge: | Recent research questions the importance of dot-product self-attention in Transformer models and shows that most attention heads learn simple positional patterns. |
| Approach: | They propose a novel mechanism to replace dot-product self-attention with a recurrent atteNtion mechanism that directly learns attention weights without token-to-token interaction. |
| Outcome: | The proposed model outperforms the Transformer model on translation tasks with fewer parameters and inference time. |
DeepSeekMoE: Towards Ultimate Expert Specialization in Mixture-of-Experts Language Models (2024.acl-long)
Copied to clipboard
Damai Dai, Chengqi Deng, Chenggang Zhao, R.x. Xu, Huazuo Gao, Deli Chen, Jiashi Li, Wangding Zeng, Xingkai Yu, Y. Wu, Zhenda Xie, Y.k. Li, Panpan Huang, Fuli Luo, Chong Ruan, Zhifang Sui, Wenfeng Liang
| Challenge: | Mixture-of-Experts (MoE) architectures face challenges in ensuring expert specialization . despite the promising performance, scaling language models to an extremely large scale is associated with exceedingly high computational costs. |
| Approach: | They propose an architecture that allows for ultimate expert specialization by segmenting experts into mN ones and activating mK from them. |
| Outcome: | The proposed architecture achieves comparable performance with GShard with 2B parameters and computation. |
Instant Personalized Large Language Model Adaptation via Hypernetwork (2026.acl-long)
Copied to clipboard
Zhaoxuan Tan, Zixuan Zhang, Haoyang Wen, Zheng Li, Rongzhi Zhang, Pei Chen, Fengran Mo, Zheyuan Liu, Qingkai Zeng, Qingyu Yin, Meng Jiang
| Challenge: | Existing parameter-efficient fine-tuning methods require training a separate adapter for each user, making them computationally expensive and impractical for real-time updates. |
| Approach: | They propose a scalable framework that maps a user's profile directly to a full set of adapter parameters. |
| Outcome: | The proposed framework outperforms prompt-based personalization and OPPU while using substantially fewer computational resources at deployment. |
Judge as A Judge: Improving the Evaluation of Retrieval-Augmented Generation through the Judge-Consistency of Large Language Models (2025.findings-acl)
Copied to clipboard
Shuliang Liu, Xinze Li, Zhenghao Liu, Yukun Yan, Cheng Yang, Zheni Zeng, Zhiyuan Liu, Maosong Sun, Ge Yu
| Challenge: | Existing evaluation metrics cannot fairly evaluate the outputs of RAG models during training and evaluation. |
| Approach: | They propose a method which prompts LLMs to generate different judgments based on various combinations of judgment dimensions and utilizes the judge-consistency to evaluate these judgments. |
| Outcome: | The proposed method generates more accurate evaluations for RAG models across different RAG model and datasets. |
Turn Waste into Worth: Rectifying Top-k Router of MoE (2024.emnlp-main)
Copied to clipboard
Zhiyuan Zeng, Qipeng Guo, Zhaoye Fei, Zhangyue Yin, Yunhua Zhou, Linyang Li, Tianxiang Sun, Hang Yan, Dahua Lin, Xipeng Qiu
| Challenge: | Top-k router suffers from redundancy computation and memory costs due to unbalanced routing . some experts are overflow, where exceeding tokens are dropped, while others are empty, which are padded with zeros, negatively impacting model performance. |
| Approach: | They propose a top-k router that is unbalanced and uses a multi-gPU system to handle dropped tokens and padding. |
| Outcome: | The proposed model surpasses the top-1 router by 4.7% in terms of performance . the top-k router suffers from redundancy computation and memory costs . |
Dynamic Online Conversation Recommendation (2020.acl-main)
Copied to clipboard
| Challenge: | Existing models that assume static user interests are unable to capture the temporal aspects of user interactions and interest changes over time. |
| Approach: | They propose a neural architecture to exploit changes of user interactions and interests over time to predict which discussions they are likely to enter. |
| Outcome: | The proposed model outperforms state-of-the-art models that assume static user interests and handle future conversations that are unseen during training time. |
DocMEdit: Towards Document-Level Model Editing (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing models only output short phrases or sentences, raising doubts about their practical usability. |
| Approach: | They propose a dataset focused on document-level model editing that aims to correct errors and outdated knowledge in Large language models (LLMs) they propose to use document-based model editing to improve model capabilities in real-world scenarios. |
| Outcome: | The proposed model editing task improves model capabilities in real-world scenarios and reduces the cost of retraining. |
PsFuture: A Pseudo-Future-based Zero-Shot Adaptive Policy for Simultaneous Machine Translation (2024.emnlp-main)
Copied to clipboard
| Challenge: | Simultaneous machine translation (SiMT) requires target tokens to be generated in real-time as streaming source tokens are consumed. |
| Approach: | They propose a zero-shot adaptive read/write policy for siMT that generates target tokens concurrently as streaming source tokens are consumed. |
| Outcome: | The proposed policy achieves performance on par with strong baselines and the P2F method can further enhance performance. |
Leveraging Social Context for Humor Recognition and Sense of Humor Evaluation in Social Media with a New Chinese Humor Corpus - HumorWB (2024.lrec-main)
Copied to clipboard
| Challenge: | Existing humor computing research focuses on content while neglecting interaction relationships in social media. |
| Approach: | They propose a dataset which introduces social context information from social media . they propose 'humor recognition' task and 'horror evaluation task' |
| Outcome: | The proposed model incorporates social context information from social media . it shows that it is efficient and can be used to evaluate humor in real life . |
Weakly Supervised Text Classification using Supervision Signals from a Language Model (2022.findings-naacl)
Copied to clipboard
| Challenge: | Existing weakly supervised text classification methods require a large number of annotated data and human annotations are expensive. |
| Approach: | They propose to query a masked language model with cloze style prompts to obtain supervision signals. |
| Outcome: | The proposed method outperforms baseline methods on three datasets by 2%, 4%, and 3%. |
Logical Consistency as a Bridge: Improving LLM Hallucination Detection via Label Constraint Modeling between Responses and Self-Judgments (2026.acl-long)
Copied to clipboard
Hao Mi, Qiang Sheng, Shaofei Wang, Beizhe Hu, Yifan Sun, Zhengjia Wang, Hengqi Zeng, Yang Li, Danding Wang, Juan Cao
| Challenge: | Existing methods for hallucination detection focus on implicit neural uncertainty or explicit symbolic reasoning, ignoring factual hallucinosities. |
| Approach: | They propose a framework that bridges neural features and symbolic judgments for hallucination detection by leveraging a "meta-judgment" process to map symbolic labels back into the feature space. |
| Outcome: | Extensive experiments on 4 public datasets, across 4 LLMs, against 8 baselines demonstrate the superiority of LaaB. |
Toward Building General Foundation Models for Language, Vision, and Vision-Language Understanding Tasks (2023.findings-emnlp)
Copied to clipboard
| Challenge: | Existing foundation models can only perform the best in one type of understanding tasks. |
| Approach: | They propose a method for training a general foundation model, X-FM, using text, image, and image-text data. |
| Outcome: | The proposed method outperforms existing foundation models on language, vision, and vision-language understanding tasks. |
Tool Zero: Training Tool-Augmented LLMs via Pure RL from Scratch (2025.findings-emnlp)
Copied to clipboard
Yirong Zeng, Xiao Ding, Yutai Hou, Yuxian Wang, Li Du, Juyi Dai, Qiuyang Ding, Duyu Tang, Dandan Tu, Weiwen Liu, Bing Qin, Ting Liu
| Challenge: | Experimental results demonstrate that our models achieve over 7% performance improvement compared to both SFT and RL-with-SFT models under the same experimental settings. |
| Approach: | They propose a dynamic generalization-guided reward design for rule-based RL that shifts rewards from exploratory to exploitative tool-use patterns. |
| Outcome: | The proposed model achieves over 7% performance improvement compared to SFT and RL-with-SFT models under the same experimental settings. |
The Pragmatic Mind of Machines: Tracing the Emergence of Pragmatic Competence in Large Language Models (2026.eacl-long)
Copied to clipboard
| Challenge: | Current large language models (LLMs) have demonstrated emerging capabilities in social intelligence tasks, including implicature resolution and theory-of-mind reasoning. |
| Approach: | They introduce a dataset grounded in the pragmatic concept of alternatives to evaluate whether large language models can accurately infer nuanced speaker intentions. |
| Outcome: | The proposed model can infer nuanced speaker intentions by inferring the speaker’s intended meaning and explaining when and why a speaker would choose one utterance over its alternative. |
Evaluating Text Coherence at Sentence and Paragraph Levels (2020.lrec-1)
Copied to clipboard
| Challenge: | Existing text ordering models have been used to test coherence in NLP for a long time. |
| Approach: | They propose to perform paragraph ordering task and sentence ordering by using four corpora from different domains. |
| Outcome: | The proposed model performs better under certain extreme conditions than the most prevalent metric used before. |
KACE: Generating Knowledge Aware Contrastive Explanations for Natural Language Inference (2021.acl-long)
Copied to clipboard
| Challenge: | Existing approaches in NLP focus on “WHY A” rather than contrastive “WHA NOT B” Experimental results show that contrastive explanations are beneficial to fit the scenarios by clarifying the difference between the predicted answer and other possible wrong ones. |
| Approach: | They propose to generate contrastive explanations with counterfactual examples in NLI by identifying key phrases from input sentences and using them as key perturbations to generate counterfacts. |
| Outcome: | The proposed framework improves on SNLI and ETPA models by 91.9%. |
Which Side Are You On? A Multi-task Dataset for End-to-End Argument Summarisation and Evaluation (2024.findings-acl)
Copied to clipboard
Hao Li, Yuping Wu, Viktor Schlegel, Riza Batista-Navarro, Tharindu Madusanka, Iqra Zahid, Jiayan Zeng, Xiaochi Wang, Xinran He, Yizhi Li, Goran Nenadic
| Challenge: | Recent advances in large language models (LLMs) have made it difficult to build an automated debate system that helps people to synthesise persuasive arguments. |
| Approach: | They propose to use an argument mining dataset to capture the end-to-end process of preparing an argumentative essay for a debate. |
| Outcome: | The proposed dataset shows that it performs better on individual tasks than on human-centred evaluations. |
DCP: Dual-Cue Pruning for Efficient Large Vision-Language Models (2025.emnlp-main)
Copied to clipboard
| Challenge: | Existing pruning methods for large vision language models use visual tokens to prune . existing methods fail to balance efficiency and semantic alignment due to large number of visual token. |
| Approach: | They propose a cross-modal pruning framework that considers textual semantics and visual self-attention to combine them to achieve efficient inference acceleration. |
| Outcome: | The proposed pruning framework can retain only 25% of the visual tokens, with a minimal performance degradation of only 0.063% on LLaVA-1.5-13B. |
Corrupted but Not Broken: Understanding and Mitigating the Negative Impacts of Corrupted Data in Visual Instruction Tuning (2025.emnlp-main)
Copied to clipboard
Yunhao Gou, Hansi Yang, Zhili Liu, Kai Chen, Yihan Zeng, Lanqing Hong, Zhenguo Li, Qun Liu, Bo Han, James Kwok, Yu Zhang
| Challenge: | Visual Instruction Tuning (VIT) aims to enhance Multimodal Large Language Models (MLLMs), but its effectiveness is often compromised by corrupted datasets with issues such as hallucinated content and poor OCR quality. |
| Approach: | They propose a corruption-robust training paradigm that surpasses existing strategies for mitigating the effects of corrupted data. |
| Outcome: | The proposed training paradigm surpasses existing strategies for mitigating the effects of corrupted data. |
Counterfactual Generator: A Weakly-Supervised Method for Named Entity Recognition (2020.emnlp-main)
Copied to clipboard
| Challenge: | Using labeled data, named entity recognition is labor-intensive, time-consuming and expensive. |
| Approach: | They propose a method which decomposes named entity into two parts: entity and context. |
| Outcome: | The proposed method improves the generalization ability of models under limited observational examples. |
HyperEdit: Unlocking Instruction-based Text Editing in LLMs via Hypernetworks (2026.findings-acl)
Copied to clipboard
Yiming Zeng, Jinghan Cao, Zexin Li, Wanhao Yu, Zhankai Ye, Dawei Xiang, Ting Hua, Xin Liu, Shangqian Gao, Tingting Yu
| Challenge: | Existing approaches treat instruction-based text editing as a generic text generation problem. Existing methods either over-edit or fail to apply modifications consistently. |
| Approach: | They propose a framework that processes each editing request to best align with it. |
| Outcome: | The proposed framework achieves 9% improvement over the state-of-the-art model. |
Do Deep Neural Nets Display Human-like Attention in Short Answer Scoring? (2022.naacl-main)
Copied to clipboard
| Challenge: | DL-based graders often lack the ability to explain and justify how a prediction is made, which decreases their trustworthiness and hinders educators from embracing them in practice. |
| Approach: | They conducted a user study to determine whether DL-based graders align with human grader . they also ran a randomized controlled experiment to explore the impact of highlighting important words detected by DL grader. |
| Outcome: | The proposed method enables human graders to identify important words when marking short answer questions. |
Stepwise Perplexity-Guided Refinement for Efficient Chain-of-Thought Reasoning in Large Language Models (2025.findings-acl)
Copied to clipboard
Yingqian Cui, Pengfei He, Jingying Zeng, Hui Liu, Xianfeng Tang, Zhenwei Dai, Yan Han, Chen Luo, Jing Huang, Zhen Li, Suhang Wang, Yue Xing, Jiliang Tang, Qi He
| Challenge: | Chain-of-Thought (CoT) reasoning has improved the performance of large language models (LLMs) however, the detailed reasoning process in CoT often incurs long generation times and high computational costs due to the inclusion of unnecessary steps. |
| Approach: | They propose a method to identify critical reasoning steps using perplexity as a measure of their importance. |
| Outcome: | The proposed method achieves a better balance between reasoning accuracy and efficiency of CoT. |
SIFT: Grounding LLM Reasoning in Contexts via Stickers (2025.findings-emnlp)
Copied to clipboard
| Challenge: | Using a new approach, we can improve the pass@1 accuracy of LLM reasoning in large language models. |
| Approach: | They propose a method that leverages increasing inference-time compute to ground LLM reasoning in contexts. |
| Outcome: | The proposed approach improves pass@1 accuracy of DeepSeek-R1 on AIME2024 from 78.33% to **85.67%** and that on Aime2025 from 69.8% to **77.33%**. |
CogEvolve: A Multimodal Benchmark for Evaluating Relational Reasoning in Semantic Extension (2026.acl-long)
Copied to clipboard
| Challenge: | a gap exists between human embodied logic and machine statistical learning . authors: models internalize statistical patterns or mimic static recognition . |
| Approach: | They propose a cognitive linguistic benchmark to test whether large language models internalize statistical logic or not . they find that models function as "Super-Associators" expert at static recognition yet fail at causal reasoning . |
| Outcome: | The proposed model fails at causal reasoning and has a high-fidelity concept representation but lacks transformational operators essential for true relational understanding. |
Marco-Bench-MIF: On Multilingual Instruction-Following Capability of Large Language (2025.acl-long)
Copied to clipboard
Bo Zeng, Chenyang Lyu, Sinuo Liu, Mingyan Zeng, Minghao Wu, Xuanfan Ni, Tianqi Shi, Yu Zhao, Yefeng Liu, Chenyu Zhu, Ruizhe Li, Jiahui Geng, Qing Li, Yu Tong, Longyue Wang, Weihua Luo, Kaifu Zhang
| Challenge: | Existing datasets for instruction-following are monolingual and centered on English . existing data are unable to capture linguistic and cultural subtle differences . |
| Approach: | They propose an extension of IFEval to a localized multilingual version called Marco-Bench-MIF . their benchmark addresses linguistic constraints and cultural references via translation and verification . |
| Outcome: | The proposed extension of IFEval to a localized multilingual version covers 30 languages with varying levels of localization. |
PSSAT: A Perturbed Semantic Structure Awareness Transferring Method for Perturbation-Robust Slot Filling (2022.coling-1)
Copied to clipboard
Guanting Dong, Daichi Guo, Liwen Wang, Xuefeng Li, Zechen Wang, Chen Zeng, Keqing He, Jinzheng Zhao, Hao Lei, Xinyue Cui, Yi Huang, Junlan Feng, Weiran Xu
| Challenge: | Existing slot filling models memorize inherent patterns of entities and contexts from training data. |
| Approach: | They propose a perturbed semantic structure awareness transferring method for slot filling models . they use two MLM-based training strategies to learn contextual semantic structure and word distribution . |
| Outcome: | The proposed method outperforms existing methods and gains strong generalization while preventing model from memorizing inherent patterns of entities and contexts. |
DIGAT: Modeling News Recommendation with Dual-Graph Interaction (2022.findings-emnlp)
Copied to clipboard
| Challenge: | Existing news recommendation methods lack effective news-user feature interaction. |
| Approach: | They propose to use news-graph and user-graph channels to enhance news encodings . they also propose to perform effective feature interaction between news and user graphs based on semantic-augmented graphs. |
| Outcome: | The proposed graph attention networks outperform existing NR methods on the benchmark dataset MIND. |
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 . |
KnowCoder: Coding Structured Knowledge into LLMs for Universal Information Extraction (2024.acl-long)
Copied to clipboard
Zixuan Li, Yutao Zeng, Yuxin Zuo, Weicheng Ren, Wenxuan Liu, Miao Su, Yucan Guo, Yantao Liu, Lixiang Lixiang, Zhilei Hu, Long Bai, Wei Li, Yidan Liu, Pan Yang, Xiaolong Jin, Jiafeng Guo, Xueqi Cheng
| Challenge: | None. None.. None! |
| Approach: | None. None.. None! |
| Outcome: | None. None. No. : |
Dr. Assistant: Enhancing Clinical Diagnostic Inquiry via Structured Diagnostic Reasoning Data and Reinforcement Learning (2026.findings-acl)
Copied to clipboard
Yue Guo, Fanfu Wang, Jianwei Lv, Xincheng Shi, Yuchen Li, Youya Wang, Yunsheng Zeng, Yujing Liu, Yunhao Qiao, Gen Li, Junfeng Wang, Bo Yuan
| Challenge: | Clinical Decision Support Systems (CDSSs) provide reasoning and inquiry guidance for physicians, yet they face high maintenance costs and low generalization capability. |
| Approach: | They propose a clinical diagnostic model with clinical reasoning and inquiry skills, the Dr. Assistant, and a pipeline to capture abstract reasoning logic. |
| Outcome: | The proposed model outperforms open-source models and achieves competitive performance to closed-source model. |
HybridKV: Hybrid KV Cache Compression for Efficient Multimodal Large Language Model Inference (2026.acl-long)
Copied to clipboard
| Challenge: | Multimodal Large Language Models (MLLMs) are hindered by the rapid growth of key–value (KV) caches. |
| Approach: | They propose a hybrid KV cache compression framework that reduces KV memory by 7.9 and speeds up decoding by 1.52. |
| Outcome: | Experiments on 11 multimodal benchmarks show that HYBRIDKV cuts KV cache memory by 7.9 and speeds up decoding by 1.52. |
RESIN-11: Schema-guided Event Prediction for 11 Newsworthy Scenarios (2022.naacl-demo)
Copied to clipboard
Xinya Du, Zixuan Zhang, Sha Li, Pengfei Yu, Hongwei Wang, Tuan Lai, Xudong Lin, Ziqi Wang, Iris Liu, Ben Zhou, Haoyang Wen, Manling Li, Darryl Hannan, Jie Lei, Hyounghun Kim, Rotem Dror, Haoyu Wang, Michael Regan, Qi Zeng, Qing Lyu, Charles Yu, Carl Edwards, Xiaomeng Jin, Yizhu Jiao, Ghazaleh Kazeminejad, Zhenhailong Wang, Chris Callison-Burch, Mohit Bansal, Carl Vondrick, Jiawei Han, Dan Roth, Shih-Fu Chang, Martha Palmer, Heng Ji
| Challenge: | Existing methods for event prediction are incomplete and noisy. |
| Approach: | They propose to use news-related event schemas to extract newsworthy events . they build a demo website and include a video demonstrating the framework . |
| Outcome: | The proposed framework can be applied to a wide variety of newsworthy scenarios. |
Improving Consistency for Text Summarization with Energy Functions (2023.findings-emnlp)
Copied to clipboard
Qi Zeng, Qingyu Yin, Zheng Li, Yifan Gao, Sreyashi Nag, Zhengyang Wang, Bing Yin, Heng Ji, Chao Zhang
| Challenge: | Current abstractive summarization models generate inconsistent content due to the inherently noisy dataset and the discrepancy between maximum likelihood estimation based training objectives and consistency measurements. |
| Approach: | They propose a new consistency taxonomy that categorizes inconsistent content into faithfulness, factuality, and self-supportiveness. |
| Outcome: | Experiments on XSUM and CNN/DM datasets show that EnergySum mitigates the trade-off between accuracy and consistency. |
Exploring Memorization in Fine-tuned Language Models (2024.acl-long)
Copied to clipboard
Shenglai Zeng, Yaxin Li, Jie Ren, Yiding Liu, Han Xu, Pengfei He, Yue Xing, Shuaiqiang Wang, Jiliang Tang, Dawei Yin
| Challenge: | Existing studies have shown that pre-trained langauge models tend to memorize and regenerate segments of their pre-training corpus when prompted appropriately. |
| Approach: | They conduct the first comprehensive analysis to explore language models’ memorization during fine-tuning across tasks. |
| Outcome: | The proposed analysis shows that memorization presents a strong disparity among different fine-tuning tasks. |
Reinforcement Learning on Pre-Training Data (2026.acl-long)
Copied to clipboard
Siheng Li, Kejiao Li, Zenan Xu, Guanhua Huang, Kun Li, Haoyuan Wu, null Wujiajia, Zihao Zheng, Chenchen Zhang, Kun Shi, Xue Gong, Qi Yi, Ruibin Xiong, Tingqiang Xu, Yuhao Jiang, Jianfeng Yan, Yuyuan Zeng, Guanghui Xu, Jinbao Xue, Zhijiang xu, Zheng Fang, Shuai LI, Qibin Liu, Xiaoxue Li, Zhuoyu Li, Yangyu Tao, Fei Gao, Cheng Jiang, Bochao Wang, Kai Liu, Jianchen Zhu, Wai Lam, Bo Zhou, Di Wang
| Challenge: | Recent progress in large language models is driven by scaling of training compute through pre-training with nexttoken prediction (NTP) or post-training (RL) Pre-training using NTP enables models to acquire extensive knowledge and skills from general data, but it suffers from data inefficiency and catastrophic forgetting in continual learning settings. |
| Approach: | They propose to scale training compute through pre-training with next-token prediction (NTP) or post-training by scaling reinforcement learning (RL) to improve learning from general data. |
| Outcome: | Experiments on multiple benchmarks and models show that the proposed approach improves continual pre-training and provides a strong foundation for post-training on Qwen3-8B-Base. |
No More Stale Feedback: Co-Evolving Critics for Open-World Agent Learning (2026.acl-long)
Copied to clipboard
Zhicong Li, Lingjie Jiang, Yulan Hu, Xingchen Zeng, Yixia Li, Xiangwen Zhang, Guanhua Chen, Zheng Pan, Xin Li, Yong Liu
| Challenge: | Current methods for training Large Language Model agents rely on static or offline critic models, which fail to adapt as the policy evolves. |
| Approach: | They propose a framework that integrates a critique and a policy to optimize the policy and critic through a synchronized co-evolutionary loop. |
| Outcome: | The proposed framework yields more stable training and higher long-horizon task success across open-world environments. |
CARD: Cross-modal Agent Framework for Generative and Editable Residential Design (2025.emnlp-main)
Copied to clipboard
| Challenge: | Architectural design automation has made significant progress, but the complexity of open-world environments makes residential design a challenging task. |
| Approach: | They propose a framework that leverages a system of specialized cross-modal agents to adapt to open-world residential design. |
| Outcome: | The proposed framework enables users to generate and edit residential design without requiring specialized expertise. |
Optimal Transport-Based Token Weighting scheme for Enhanced Preference Optimization (2025.acl-long)
Copied to clipboard
| Challenge: | Existing methods for direct preference optimization assign equal importance to all tokens while humans focus on more meaningful parts. |
| Approach: | They propose to use a transport-based token weighting scheme to enhance direct preference optimization by emphasizing meaningful token pairs and de-emphasizing less relevant ones to yield a more contrastive reward difference estimate. |
| Outcome: | Extensive experiments have validated the proposed method in improving instruction-following ability across various settings. |
History-Aware Hierarchical Transformer for Multi-session Open-domain Dialogue System (2022.findings-emnlp)
Copied to clipboard
| Challenge: | Existing open-domain dialogue systems conduct one-session conversations, but multi-session MSCs are under-investigated. |
| Approach: | They propose a History-Aware Hierarchical Transformer for multi-session open-domain dialogue . they propose to encode history conversations into a history memory and leverage historical information to generate well-informed responses. |
| Outcome: | The proposed model outperforms baseline models on a large-scale MSC dataset. |
PaTaRM: Bridging Pairwise and Pointwise Signals via Preference-Aware Task-Adaptive Reward Modeling (2026.acl-long)
Copied to clipboard
| Challenge: | Existing reward models lack generative and reasoning capabilities, resulting in poor performance. |
| Approach: | They propose a reward-aware task-adaptive reward model that enables pointwise training using readily available pairwise data via a novel Preference-Aware Reward mechanism. |
| Outcome: | The proposed reward model achieves an average relative improvement of 8.7% over the base models on RewardBench and RMBench. |
ChatGLM-Math: Improving Math Problem-Solving in Large Language Models with a Self-Critique Pipeline (2024.findings-emnlp)
Copied to clipboard
Yifan Xu, Xiao Liu, Xinghan Liu, Zhenyu Hou, Yueyan Li, Xiaohan Zhang, Zihan Wang, Aohan Zeng, Zhengxiao Du, Zhao Wenyi, Jie Tang, Yuxiao Dong
| Challenge: | Large language models (LLMs) have shown excellent mastering of human language but struggle in real-world applications that require mathematical problem-solving. |
| Approach: | They propose a pipeline to train a general Math-Critique model from the LLM itself to provide feedback signals and employ rejective fine-tuning and direct preference optimization over the Llm's own generations for data collection. |
| Outcome: | The proposed pipeline outperforms existing LLMs that could be two times larger. |
Stepwise Reasoning Checkpoint Analysis: A Test Time Scaling Method to Enhance LLMs’ Reasoning (2025.emnlp-main)
Copied to clipboard
Zezhong Wang, Xingshan Zeng, Weiwen Liu, Yufei Wang, Liangyou Li, Yasheng Wang, Lifeng Shang, Xin Jiang, Qun Liu, Kam-Fai Wong
| Challenge: | Existing methods that use Chain-of-Thought suffer from path homogenization and inefficient use of intermediate results. |
| Approach: | They propose a framework that introduces checkpoints between reasoning steps to reduce path homogenization and create fault-tolerant mechanisms. |
| Outcome: | The proposed framework reduces path homogenization and creates fault-tolerant mechanism by utilizing high-quality intermediate results. |
Reasoning in Flux: Enhancing Large Language Models Reasoning through Uncertainty-aware Adaptive Guidance (2024.acl-long)
Copied to clipboard
Zhangyue Yin, Qiushi Sun, Qipeng Guo, Zhiyuan Zeng, Xiaonan Li, Junqi Dai, Qinyuan Cheng, Xuanjing Huang, Xipeng Qiu
| Challenge: | Extensive experiments across various reasoning tasks demonstrate that UAG not only enhances the reasoning abilities of LLMs but consistently outperforms several strong baselines with minimal computational overhead. |
| Approach: | They propose an approach to guide LLMs onto an accurate and reliable trajectory by identifying and adjusting uncertainty signals within each step of the reasoning chain. |
| Outcome: | The proposed approach outperforms strong baselines and outperformed strong models with minimal computational overhead. |
KnowCoder-X: Boosting Multilingual Information Extraction via Code (2025.findings-acl)
Copied to clipboard
Yuxin Zuo, Wenxuan Jiang, Wenxuan Liu, Zixuan Li, Long Bai, Hanbin Wang, Yutao Zeng, Xiaolong Jin, Jiafeng Guo, Xueqi Cheng
| Challenge: | Empirical evidence indicates that Large Language Models exhibit spontaneous cross-lingual alignment in Information Extraction (IE) however, a significant imbalance across languages persists, highlighting an underlying deficiency. |
| Approach: | They propose a code LLM with advanced cross-lingual and multilingual capabilities for universal IE that standardizes the representation of multilingual schemas using Python classes and conducts IE alignment instruction tuning on translated instance prediction task. |
| Outcome: | The proposed model surpasses ChatGPT and SoTA by 30.17% without training in 29 unseen languages and significantly improves cross-lingual IE transferability. |
ODE Transformer: An Ordinary Differential Equation-Inspired Model for Sequence Generation (2022.acl-long)
Copied to clipboard
Bei Li, Quan Du, Tao Zhou, Yi Jing, Shuhan Zhou, Xin Zeng, Tong Xiao, JingBo Zhu, Xuebo Liu, Min Zhang
| Challenge: | Residual networks are an Euler discretization of solutions to Ordinary Differential Equations (ODE). |
| Approach: | They propose a residual block of layers in Transformer that can be described as a higher-order solution to ODE. |
| Outcome: | The proposed architecture can gain large improvements over strong baselines at a slight cost in inference efficiency. |
A General Framework to Enhance Fine-tuning-based LLM Unlearning (2025.findings-acl)
Copied to clipboard
Jie Ren, Zhenwei Dai, Xianfeng Tang, Hui Liu, Jingying Zeng, Zhen Li, Rahul Goutam, Suhang Wang, Yue Xing, Qi He, Hui Liu
| Challenge: | Existing approaches to remove copyrighted and privacy-sensitive data from Large Language Models (LLMs) have been proposed to remove specific data from LLMs without requiring full retraining. |
| Approach: | They propose a general framework that enhances the utility of fine-tuning-based methods by distinguishing target data and suppressing related generations. |
| Outcome: | The proposed framework improves the unlearning and utility of fine-tuning-based methods by distinguishing the target data and suppressing related generations. |
Continuity of Topic, Interaction, and Query: Learning to Quote in Online Conversations (2020.emnlp-main)
Copied to clipboard
| Challenge: | Quotations are crucial for successful explanations and persuasions in interpersonal communications. |
| Approach: | They propose to use an encoder-decoder neural framework to continue the context with a quotation via language generation to capture latent topics, interactions with the dialogue history, and coherence to the existing contents. |
| Outcome: | The proposed model outperforms state-of-the-art models on two large-scale datasets in English and Chinese and shows that topic, interaction, and query consistency are helpful to learn how to quote in online conversations. |
From Long Videos to Engaging Clips: A Human-Inspired Video Editing Framework with Multimodal Narrative Understanding (2025.emnlp-industry)
Copied to clipboard
Xiangfeng Wang, Xiao Li, Yadong Wei, null Songxueyu, Yang Song, null Xiaxiaoqiang, Fangrui Zeng, Zaiyi Chen, null Liuliu, Gu Xu, Tong Xu
| Challenge: | Existing methods for video editing rely on textual cues from ASR transcripts and segment selection, often neglecting rich visual context. |
| Approach: | They propose a human-inspired automatic video editing framework that leverages multimodal narrative understanding to address these limitations. |
| Outcome: | The proposed framework outperforms existing baselines across general and advertisement-oriented editing tasks. |
Few-shot Natural Language Generation for Task-Oriented Dialog (2020.findings-emnlp)
Copied to clipboard
| Challenge: | Existing methods for NLG depend on heavily annotated data, which is infeasible for new domains. |
| Approach: | They propose a system that converts a dialog act into a response in natural language . they propose 'nuclear language generation' to simulate a few-shot learning setting . |
| Outcome: | The proposed model outperforms existing methods on a large set of annotated datasets. |
PiKGL: Leveraging Pruned Knowledge Graphs for Explainable Stance Detection (2026.tacl-1)
Copied to clipboard
Bingbing Wang, Jingjie Lin, Zhixin Bai, Xintong Song, Qianlong Wang, Min Yang, Xi Zeng, Jing Li, Ruifeng Xu
| Challenge: | Experimental results demonstrate that a Pruned interpretable knowledge Graph Learning framework for explainable stance detection is state-of-the-art for social media stance prediction. |
| Approach: | They propose a Pruned interpretable knowledge Graph Learning framework for explainable stance detection that incorporates commonsense knowledge and prunes redundant information to ensure precision and minimize noise. |
| Outcome: | The proposed framework achieves state-of-the-art on three public datasets. |
MultiConIR: Towards Multi-Condition Information Retrieval (2025.findings-emnlp)
Copied to clipboard
Xuan Lu, Sifan Liu, Bochao Yin, Yongqi Li, Xinghao Chen, Hui Su, Yaohui Jin, Wenjun Zeng, Xiaoyu Shen
| Challenge: | MultiConIR is a benchmark designed to evaluate retrieval and reranking models under nuanced multi-condition query scenarios. |
| Approach: | They propose a benchmark to evaluate retrieval and reranking models under nuanced multi-condition query scenarios. |
| Outcome: | The proposed benchmark evaluates retrieval and reranking models under nuanced multi-condition query scenarios across five domains. |
MorphoBench: A Benchmark with Difficulty Adaptive to Model Reasoning (2026.findings-acl)
Copied to clipboard
Xukai Wang, Xuanbo Liu, Mingrui Chen, Haitian Zhong, Xuanlin Yang, Bohan Zeng, Jinbo Hu, Hao Liang, Junbo Niu, Xuchen Li, Ruitao Wu, Ruichuan An, Yang Shi, Liu Liu, Qiang Liu, Zhouchen Lin, Xu-Yao Zhang, Wentao Zhang, Bin Dong
| Challenge: | Existing benchmarks designed to evaluate the reasoning capabilities of large models are limited in scope and lack flexibility to adapt difficulty according to evolving reasoning capacities of models. |
| Approach: | They propose a benchmark that incorporates multidisciplinary questions to evaluate the reasoning capabilities of large models and can adjust and update question difficulty based on the reasoning abilities of advanced models. |
| Outcome: | The proposed benchmark incorporates multidisciplinary questions to evaluate the reasoning capabilities of large models and can adjust and update question difficulty based on the reasoning abilities of advanced models. |
Memorize Step by Step: Efficient Long-Context Prefilling with Incremental Memory and Decremental Chunk (2024.emnlp-main)
Copied to clipboard
Zhiyuan Zeng, Qipeng Guo, Xiaoran Liu, Zhangyue Yin, Wentao Shu, Mianqiu Huang, Bo Wang, Yunhua Zhou, Linlin Li, Qun Liu, Xipeng Qiu
| Challenge: | Existing methods to optimize LLM for long sequences for long documents are slow and consume memory. |
| Approach: | They propose a method that starts with a small memory size and gradually increases it . they propose Decremental Chunk based on Incremental Memory (IMDC) which reduces chunk size while increasing memory size . |
| Outcome: | The proposed method is faster (1.45x) and reduces GPU memory consumption by 23.3% compared to fixed-size memory. |
AdaTranS: Adapting with Boundary-based Shrinking for End-to-End Speech Translation (2023.findings-emnlp)
Copied to clipboard
| Challenge: | End-to-end speech translation (ST) models need large amount of training data to perform well. |
| Approach: | They propose a shrinking mechanism to mitigate the length mismatch between speech and text features by predicting word boundaries. |
| Outcome: | The proposed method achieves better performance on the MUST-C dataset, with higher inference speed and lower memory usage. |
GOBench: Stage-Wise Diagnostics and the Visual Paradox in Multimodal Graph Optimization (2026.findings-acl)
Copied to clipboard
| Challenge: | Existing benchmarks fail to represent multimodal problem specifications, score outcomes only and cannot localize where failures occur along the modeling pipeline. |
| Approach: | They propose a Graph Optimization benchmark that aligns multiple modalities with solver-derived oracles and a diagnostic protocol that evaluates intermediate artifacts as well as end results. |
| Outcome: | Graph Optimization benchmark (GOBench) evaluates intermediate artifacts as well as end results . vision reliably increases inference cost, while reliability impact is regime-dependent . current benchmarks fail to represent multimodal problem specifications, fail to localize failures . |
TPTU-v2: Boosting Task Planning and Tool Usage of Large Language Model-based Agents in Real-world Industry Systems (2024.emnlp-industry)
Copied to clipboard
Yilun Kong, Jingqing Ruan, YiHong Chen, Bin Zhang, Tianpeng Bao, Shi Shiwei, Du Qing, Xiaoru Hu, Hangyu Mao, Ziyue Li, Xingyu Zeng, Rui Zhao, Xueqian Wang
| Challenge: | Large language models have demonstrated proficiency in addressing tasks that necessitate a combination of task planning and the usage of external tools. |
| Approach: | They propose a framework to enhance the task planning and tool usage abilities of LLMs in industrial systems. |
| Outcome: | The proposed framework enhances the task planning and tool usage abilities of LLM-based agents in industrial systems. |
RealTranS: End-to-End Simultaneous Speech Translation with Convolutional Weighted-Shrinking Transformer (2021.findings-acl)
Copied to clipboard
| Challenge: | End-to-end simultaneous speech translation (SST) is useful in many scenarios but has not been fully investigated. |
| Approach: | They propose an end-to-end simultaneous speech translation model called RealTranS . they use interleaved convolution and unidirectional Transformer layers to downsample input speech . |
| Outcome: | The proposed model outperforms existing models on public and widely-used datasets in multiple latency settings. |
Unveiling Factual Recall Behaviors of Large Language Models through Knowledge Neurons (2024.emnlp-main)
Copied to clipboard
| Challenge: | Recent advances in Large Language Models have underscored their exceptional reasoning prowess with natural language understanding across a broad spectrum of tasks. |
| Approach: | They examine whether Large Language Models actively recall or retrieve their internal repositories of factual knowledge when faced with reasoning tasks. |
| Outcome: | The proposed model improves reasoning performance while suppressing it leads to notable degradation. |
RU22Fact: Optimizing Evidence for Multilingual Explainable Fact-Checking on Russia-Ukraine Conflict (2024.lrec-main)
Copied to clipboard
| 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. |
Connecting the Dots: Event Graph Schema Induction with Path Language Modeling (2020.emnlp-main)
Copied to clipboard
| Challenge: | Existing methods to automate event extraction focus on uncertainty, re-occurring events and multiple hypotheses. |
| Approach: | They propose a new Event Graph Schema where two event types are connected through multiple paths involving entities that fill important roles in a coherent story. |
| Outcome: | The proposed model is highly effective at inducing salient and coherent schemas. |
Learning Confidence for Transformer-based Neural Machine Translation (2022.acl-long)
Copied to clipboard
| Challenge: | A well-calibrated confidence estimate is not sufficient for neural machine translation (NMT) where probabilities from softmax distribution fail to describe when the model is probably mistaken. |
| Approach: | They propose an unsupervised confidence estimate learning jointly with the training of a neural machine translation model to quantify confidence. |
| Outcome: | The proposed model outperforms standard label smoothing and can predict failures in two real-world scenarios. |
FaStFact: Faster, Stronger Long-Form Factuality Evaluations in LLMs (2025.findings-emnlp)
Copied to clipboard
Yingjia Wan, Haochen Tan, Xiao Zhu, Xinyu Zhou, Zhiwei Li, Qingsong Lv, Changxuan Sun, Jiaqi Zeng, Yi Xu, Jianqiao Lu, Yinhong Liu, Zhijiang Guo
| Challenge: | Prior evaluation pipelines fail to evaluate factuality of long-form LLMs due to inefficiency and costly human assessment. |
| Approach: | They propose a fast and strong evaluation pipeline that can evaluate factuality of long-form LLMs . they propose 'faStFact' to reduce cost of web searching and inference calling . |
| Outcome: | The proposed evaluation pipeline achieves highest alignment with human evaluation and efficiency among existing baselines. |
GLA: Grounding Large Language Models in Molecular Hierarchy for Chemical Understanding (2026.findings-acl)
Copied to clipboard
| Challenge: | Existing molecule-language models obscure the hierarchical organization of chemical semantics . Existing models rely on linear or uniform encodings, causing structural distortion . |
| Approach: | They propose a framework that integrates intrinsic molecular topology into large language models. |
| Outcome: | The proposed framework improves on cross-modal retrieval, captioning, and property prediction benchmarks. |
RewardDS: Privacy-Preserving Fine-Tuning for Large Language Models via Reward Driven Data Synthesis (2025.emnlp-main)
Copied to clipboard
Jianwei Wang, Chengming Shi, Junyao Yang, Haoran Li, Qianli Ma, Huiping Zhuang, Cen Chen, Ziqian Zeng
| Challenge: | Existing solutions to fine-tune large language models for domain-specific tasks are ineffective in addressing privacy concerns. |
| Approach: | They propose a privacy-preserving framework that fine-tunes a reward proxy model and uses reward signals to guide the synthetic data generation. |
| Outcome: | The proposed framework fine-tunes a reward proxy model and uses reward signals to guide the synthetic data generation. |
Agentic Conversational Search with Contextualized Reasoning via Reinforcement Learning (2026.findings-acl)
Copied to clipboard
Fengran Mo, Yifan Gao, Sha Li, Hansi Zeng, Xin Liu, Zhaoxuan Tan, Xian Li, Jianshu Chen, Dakuo Wang, Meng Jiang
| Challenge: | Existing studies focus on single-turn scenarios, which might lack the ability to handle multi-turn interactions. |
| Approach: | They propose a conversational agent that interleaves search and reasoning across turns and provides tailored rewards towards evolving user goals. |
| Outcome: | The proposed agent interleaves search and reasoning across turns, enabling exploratory and adaptive behaviors learned through reinforcement learning (RL) training with tailored rewards towards evolving user goals. |
MM-CRITIC: A Holistic Evaluation of Large Multimodal Models as Multimodal Critique (2025.findings-emnlp)
Copied to clipboard
| Challenge: | e MM-CRITIC is a holistic benchmark for evaluating the critique ability of Large Multimodal Models (LMMs) covering 8 main task types and over 500 tasks, covering 4471 samples. |
| Approach: | They introduce a holistic benchmark for evaluating the critique ability of Large Multimodal Models across multiple dimensions: basic, correction, and comparison. |
| Outcome: | The proposed benchmark covers 8 main task types and over 500 tasks and is composed of 4471 samples. |
ChartEdit: How Far Are MLLMs From Automating Chart Analysis? Evaluating MLLMs’ Capability via Chart Editing (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing evaluations of multimodal large language models rely on limited case studies . however, they lack the ability to generate accurate edits according to the instructions . |
| Approach: | They propose a benchmark for chart editing that includes 1,405 edit instructions applied to 233 real-world charts. |
| Outcome: | The proposed benchmark includes 1,405 diverse editing instructions applied to 233 real-world charts. |
Learning to Edit: Aligning LLMs with Knowledge Editing (2024.acl-long)
Copied to clipboard
Yuxin Jiang, Yufei Wang, Chuhan Wu, Wanjun Zhong, Xingshan Zeng, Jiahui Gao, Liangyou Li, Xin Jiang, Lifeng Shang, Ruiming Tang, Qun Liu, Wei Wang
| Challenge: | Existing knowledge editing techniques rely on memorizing updated knowledge, impeding LLMs from effectively combining the new knowledge with their inherent knowledge when answering questions. |
| Approach: | They propose a Learning to Edit framework that equips LLMs with the ability to apply updated knowledge to input questions through a two-phase process . |
| Outcome: | The proposed framework outperforms existing methods in knowledge editing tasks and compares it with four benchmarks and two LLM architectures. |
Plug-and-Play Knowledge Injection for Pre-trained Language Models (2023.acl-long)
Copied to clipboard
Zhengyan Zhang, Zhiyuan Zeng, Yankai Lin, Huadong Wang, Deming Ye, Chaojun Xiao, Xu Han, Zhiyuan Liu, Peng Li, Maosong Sun, Jie Zhou
| Challenge: | Existing knowledge injection methods are not suitable for enhancing pre-trained language models with external knowledge bases. |
| Approach: | They propose a plug-and-play knowledge injection method where knowledge bases are injected into frozen existing downstream models by a knowledge plugin. |
| Outcome: | The proposed method improves the performance of knowledge injection on knowledge-driven tasks while keeping model parameters frozen. |
Grounded Multimodal Procedural Entity Recognition for Procedural Documents: A New Dataset and Baseline (2024.lrec-main)
Copied to clipboard
| Challenge: | Existing methods to extract procedural knowledge from documents focus on text-only settings, which is insufficient for entity disambiguation. |
| Approach: | They propose a model to detect the entity and the corresponding bounding box groundings in images. |
| Outcome: | The proposed model detects the entity and the corresponding bounding box groundings in image (i.e., visual entities) it is based on a dataset of a WikiHow 1 and EHow 2 document and the results are compared with existing models. |
Jailbreaking? One Step Is Enough! (2025.acl-long)
Copied to clipboard
Weixiong Zheng, Peijian Zeng, YiWei Li, Hongyan Wu, Nankai Lin, Junhao Chen, Aimin Yang, Yongmei Zhou
| Challenge: | Large language models (LLMs) excel in various tasks but remain vulnerable to jailbreak attacks, where adversaries manipulate prompts to generate harmful outputs. |
| Approach: | They propose a Reverse Embedded Defense Attack mechanism that disguises the attack intention as the "defense" intention against harmful content. |
| Outcome: | The proposed method outperforms existing methods on open-source and closed-source models and enables successful jailbreak in one iteration. |
The Devil is in the Details: On the Pitfalls of Event Extraction Evaluation (2023.findings-acl)
Copied to clipboard
| Challenge: | Event extraction (EE) is a fundamental information extraction task aimed at extracting events from plain texts. |
| Approach: | They propose to specify data preprocessing, standardize outputs, and provide pipeline evaluation results to avoid these pitfalls. |
| Outcome: | The results show that the evaluations are reliable and lack pipeline evaluations. |
S2-MAD: Breaking the Token Barrier to Enhance Multi-Agent Debate Efficiency (2025.naacl-long)
Copied to clipboard
Yuting Zeng, Weizhe Huang, Lei Jiang, Tongxuan Liu, XiTai Jin, Chen Tianying Tiana, Jing Li, Xiaohua Xu
| Challenge: | Large language models exhibit limitations when handling complex mathematical reasoning and logical inference tasks. |
| Approach: | They propose a sparsification strategy to reduce token costs within Multi-agent Debate (MAD) this strategy minimizes ineffective exchanges of information and unproductive discussions among agents . |
| Outcome: | The proposed approach reduces token costs by up to 94.5% while maintaining performance degradation below 2.0%. |
Mechanistic Insights into Deferred Semantic Drift in LLMs (2026.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) face a fundamental challenge with delayed disambiguation: how can a model update the meaning of an early, ambiguous token when clarifying context only appears later in the sequence? |
| Approach: | They propose a method to defer semantic re-evaluation to subsequent tokens in a process they call "Deferred Semantic Drift" they demonstrate this mechanism in metaphor comprehension and provide causal validation by steering model outputs towards literal or metaphorical meanings via targeted activation interventions. |
| Outcome: | The proposed model can update the meaning of an ambiguous word when clarifying context arrives only after it has been processed. |
LVLMs and Humans Ground Differently in Referential Communication (2026.acl-long)
Copied to clipboard
Peter Zeng, Weiling Li, Amie J. Paige, Zhengxiang Wang, Panagiotis Kaliosis, Dimitris Samaras, Gregory J. Zelinsky, Susan Brennan, Owen Rambow
| Challenge: | generative AI agents cannot model common ground in a way that enables smooth communication . a recent study examined whether large language models and large vision language models engage in grounding as human discourse partners do . |
| Approach: | They propose to use referential communication to model common ground between a pair of directors and a picture matching system. |
| Outcome: | The proposed experiment shows that generative AI agents cannot model common ground . human conversation relies on common ground accrued and updated by interacting partners . |
For-Value: Efficient Forward-Only Data Valuation for finetuning LLMs and VLMs (2026.acl-long)
Copied to clipboard
Wenlong Deng, Qi Zeng, Jiaming Zhang, Minghui Chen, Zixin Ding, Christos Thrampoulidis, Boying Gong, Xiaoxiao Li
| Challenge: | Existing methods for data valuation rely on gradient computations, making them prohibitive for billion-parameter models. |
| Approach: | They propose a forward-only data valuation framework that enables efficient batch-scalable value estimation while maintaining effectiveness. |
| Outcome: | The proposed framework matches or outperforms gradient-based baselines in detecting influential data and mislabeled data while achieving significant efficiency improvements. |
ACEBench: A Comprehensive Evaluation of LLM Tool Usage (2025.findings-emnlp)
Copied to clipboard
Chen Chen, Xinlong Hao, Weiwen Liu, Xu Huang, Xingshan Zeng, Shuai Yu, Dexun Li, Yuefeng Huang, Xiangcheng Liu, Wang Xinzhi, Wu Liu
| Challenge: | Existing benchmarks for evaluating LLMs’ tool usage face several limitations: limited evaluation scenarios, lacking assessments in real multi-turn dialogue contexts; narrow evaluation dimensions, with insufficient detailed assessments of how LLM use tools; and reliance on LLM or real API executions for evaluation, which introduces significant overhead. |
| Approach: | ACEBench is a benchmark for evaluating tool usage in Large Language Models . it categorizes data into three primary types based on evaluation methodology: Normal, Special, and Agent. |
| Outcome: | ACEBench categorizes data into three primary types based on evaluation methodology: Normal, Special, and Agent. |
Enhancing Multimodal Retrieval via Complementary Information Extraction and Alignment (2025.acl-long)
Copied to clipboard
| Challenge: | Existing studies focus on capturing information in multimodal data that is similar to their paired texts, but often ignores the complementary information contained in multimodule data. |
| Approach: | They propose a multimodal retrieval approach that employs Complementary Information Extraction and Alignment to capture complementary information in multimodal data. |
| Outcome: | The proposed approach achieves significant improvements over divide-and-conquer models and universal dense retrieval models. |
Large Language Models Are Poor Clinical Decision-Makers: A Comprehensive Benchmark (2024.emnlp-main)
Copied to clipboard
Fenglin Liu, Zheng Li, Hongjian Zhou, Qingyu Yin, Jingfeng Yang, Xianfeng Tang, Chen Luo, Ming Zeng, Haoming Jiang, Yifan Gao, Priyanka Nigam, Sreyashi Nag, Bing Yin, Yining Hua, Xuan Zhou, Omid Rohanian, Anshul Thakur, Lei Clifton, David Clifton
| Challenge: | Existing studies focus on evaluating large language models in close-ended QA tasks, but many clinical decisions involve answering open-ended questions without pre-set options. |
| Approach: | They construct a benchmark to better understand large language models in the clinic . they use existing datasets to evaluate LLMs in clinical situations . |
| Outcome: | The proposed model outperforms human experts in multiple medical tasks. |
EfficientLLM: Unified Pruning-Aware Pretraining for Auto-Designed Compact Language Models (2026.acl-long)
Copied to clipboard
Xingrun Xing, Zheng Liu, Shitao Xiao, Boyan Gao, Yiming Liang, Haokun Lin, Xianlin Zeng, Guoqi Li, Jiajun Zhang
| Challenge: | Large language models (LLMs) driven by scaling laws can be developed in large model sizes. |
| Approach: | They propose a pruning-aware pretraining approach that decouples LLM pruning from direct pretraining. |
| Outcome: | The proposed model outperforms pretraining models with 100M 1B parameters in commen sense benchmarks. |
Microblog Conversation Recommendation via Joint Modeling of Topics and Discourse (N18-1)
Copied to clipboard
| Challenge: | Existing methods for recommendation focus on content of individual posts, but we exploit both context and user content and behavior preferences. |
| Approach: | They propose a method that captures conversational context and user content and behavior preferences. |
| Outcome: | The proposed method outperforms methods that only model content without considering discourse on two Twitter datasets. |
When to Continue Thinking: Adaptive Thinking Mode Switching for Efficient Reasoning (2025.findings-emnlp)
Copied to clipboard
Xiaoyun Zhang, Jingqing Ruan, Xing Ma, Yawen Zhu, Haodong Zhao, Hao Li, Jiansong Chen, Ke Zeng, Xunliang Cai
| Challenge: | Large reasoning models (LRMs) incur excessive computational overhead due to redundant reasoning, especially on simple tasks. |
| Approach: | They propose an Adaptive Self-Recovery Reasoning framework that suppresses unnecessary reasoning and enables implicit recovery. |
| Outcome: | The proposed framework suppresses unnecessary reasoning and enables implicit recovery. |
Reinforcement Learning–Guided Adaptive Tuning for Out-of-Distribution Harmful Text Detection (2026.acl-long)
Copied to clipboard
| Challenge: | Existing methods for testing harmful information on social media rely on fixed parameters that fail to handle substantial semantic discrepancies . RLAT can be used to adapt to semantic variations while preventing overfitting from continuous tuning. |
| Approach: | They propose a reinforcement learning-guided adaptive tuning method for harmful text detection that optimizes consistency loss and applies word-level attention constraints to reduce over-reliance on local words. |
| Outcome: | The proposed method outperforms state-of-the-art models in cross-platform and cross-temporal scenarios across multiple public datasets. |
Bridging the Editing Gap in LLMs: FineEdit for Precise and Targeted Text Modifications (2025.findings-emnlp)
Copied to clipboard
| Challenge: | a recent study shows that large language models can perform precise text editing tasks. |
| Approach: | InstrEditBench is a benchmark dataset that compares 30,000 structured editing tasks . experimental evaluations show FineEdit outperforms state-of-the-art models . |
| Outcome: | The proposed model outperforms state-of-the-art models on single-turn edits and mistral-7B-OpenOrca on direct edits. |
Automatic Prompt Optimization with “Gradient Descent” and Beam Search (2023.emnlp-main)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have shown impressive performance as general purpose agents, but their abilities remain highly dependent on prompts which are hand written with onerous trial-and-error effort. |
| Approach: | They propose an algorithm that uses numerical gradient descent to automatically improve prompts by rewriting vague task descriptions into more precise annotation instructions. |
| Outcome: | The proposed algorithm outperforms previous methods and improves performance on three benchmark NLP tasks and the novel problem of LLM jailbreak detection. |
Are All Prompt Components Value-Neutral? Understanding the Heterogeneous Adversarial Robustness of Dissected Prompt in LLMs (2026.eacl-long)
Copied to clipboard
Yujia Zheng, Tianhao Li, Haotian Huang, Tianyu Zeng, Jingyu Lu, Chuangxin Chu, Yuekai Huang, Ziyou Jiang, Qian Xiong, Yuyao Ge, Mingyang Li
| Challenge: | Existing studies treat prompts as flat text, overlooking their internal structure, and different components within a prompt contribute unequally to robustness. |
| Approach: | They propose a framework that decomposes prompts into functional components and a method that selectively modifies components to expose component-wise vulnerabilities. |
| Outcome: | The proposed framework exposes component-wise vulnerabilities while ensuring linguistic plausibility through perplexity-based filtering. |
Neural Conversation Recommendation with Online Interaction Modeling (D19-1)
Copied to clipboard
| Challenge: | Existing models that only use lexical features and ignore past user interactions in online conversations are inadequate to identify and engage in online discussions. |
| Approach: | They propose a framework that automatically recommends conversations based on user's prior conversation behaviors by exploring deep semantic features that measure how a user’s preferences match an ongoing conversation’s context. |
| Outcome: | The proposed model outperforms state-of-the-art models on two large-scale datasets from Twitter and Reddit showing that it incorporates deep semantic features that measure how a user’s preferences match an ongoing conversation’s context. |
RankCoT: Refining Knowledge for Retrieval-Augmented Generation through Ranking Chain-of-Thoughts (2025.acl-long)
Copied to clipboard
| Challenge: | Retrieval-Augmented Generation (RAG) models enable Large Language Models to access external knowledge. |
| Approach: | They propose a knowledge refinement method that incorporates reranking signals to generate CoT-based summarization based on query and retrieval documents. |
| Outcome: | RankCoT generates CoT-based summarization based on query and all retrieval documents . Rank CoT incorporates a self-reflection mechanism that refines the outputs . |
From Personal to Collective: On the Role of Local and Global Knowledge in LLM Personalization (2026.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) are the backbone of modern AI systems supporting a wide range of applications. |
| Approach: | They propose a Local–Global knowledge framework that augments user-specific signals with a global knowledge encoding collective behavioral trends. |
| Outcome: | The proposed framework outperforms existing methods on five personalization benchmarks. |
MMLU-ProX: A Multilingual Benchmark for Advanced Large Language Model Evaluation (2025.emnlp-main)
Copied to clipboard
Weihao Xuan, Rui Yang, Heli Qi, Qingcheng Zeng, Yunze Xiao, Aosong Feng, Dairui Liu, Yun Xing, Junjue Wang, Fan Gao, Jinghui Lu, Yuang Jiang, Huitao Li, Xin Li, Kunyu Yu, Ruihai Dong, Shangding Gu, Yuekang Li, Xiaofei Xie, Felix Juefei-Xu, Foutse Khomh, Osamu Yoshie, Qingyu Chen, Douglas Teodoro, Nan Liu, Randy Goebel, Lei Ma, Edison Marrese-Taylor, Shijian Lu, Yusuke Iwasawa, Yutaka Matsuo, Irene Li
| Challenge: | Existing large language model evaluation benchmarks focus on English, while current multilingual tasks lack parallel questions that specifically assess cross-lingual reasoning abilities. |
| Approach: | They propose a comprehensive benchmark covering 29 languages, built on an English benchmark. |
| Outcome: | The MMLU-ProX is a comprehensive benchmark covering 29 languages, built on an English benchmark. |
Fin-STAR: Structure-as-Semantics to Resolve Implicitness in Financial Retrieval (2026.findings-acl)
Copied to clipboard
Yu Zou, Yan Chen, Lida He, Qi Zhou, Xiaorui Zhou, Aixi Zhong, Yi Wang, Wei Li, Qingyu Wang, Jiatao Li, Wei Gong, Jialei Zeng, Jingmei Zhao, Ke Jiang, Qing Li
| Challenge: | Existing Retrieval-Augmented Generation systems treat structure as a physical navigational skeleton rather than intrinsic semantic knowledge. |
| Approach: | They propose a framework that redefining hierarchy as intrinsic semantics and uses snippets to enrich hierarchical lineage. |
| Outcome: | The proposed framework outperforms state-of-the-art hierarchical and graph-based benchmarks on FinTierQA Gold. |
RAGLAB: A Modular and Research-Oriented Unified Framework for Retrieval-Augmented Generation (2024.emnlp-demo)
Copied to clipboard
Xuanwang Zhang, Yun-Ze Song, Yidong Wang, Shuyun Tang, Xinfeng Li, Zhengran Zeng, Zhen Wu, Wei Ye, Wenyuan Xu, Yue Zhang, Xinyu Dai, Shikun Zhang, Qingsong Wen
| Challenge: | Existing research on Retrieval Augmented Generation (RAG) does not address the problem of hallucinations and real-time updating of knowledge. |
| Approach: | They propose a modular open-source library to equip LLMs with external knowledge. |
| Outcome: | The proposed approach reduces the need for expensive open-source tools and lacks fair comparisons between novel RAG algorithms. |
QuantAgents: Towards Multi-agent Financial System via Simulated Trading (2025.findings-emnlp)
Copied to clipboard
| Challenge: | Existing LLM-based agent models exhibit significant deviations from real-world fund companies. |
| Approach: | They propose a multi-agent financial system that incorporates simulated trading . they propose simulated trades are evaluated without assuming actual risks . |
| Outcome: | The proposed system evaluates various investment strategies without assuming actual risks without involving real-world investors. |
ToolFlow: Boosting LLM Tool-Calling Through Natural and Coherent Dialogue Synthesis (2025.naacl-long)
Copied to clipboard
Zezhong Wang, Xingshan Zeng, Weiwen Liu, Liangyou Li, Yasheng Wang, Lifeng Shang, Xin Jiang, Qun Liu, Kam-Fai Wong
| Challenge: | Large Language Models (LLMs) can be enhanced by using supervised fine-tuning . however, access to fine-timing data can be limited. |
| Approach: | They propose a Graph-based Sampling strategy and a Planned-generation strategy to enhance the coherence between dialogues by using 8,000 synthetic dialogues. |
| Outcome: | The proposed model achieves tool-calling performance comparable to or surpassing GPT-4 while maintaining strong general capabilities. |
FollowBench: A Multi-level Fine-grained Constraints Following Benchmark for Large Language Models (2024.acl-long)
Copied to clipboard
Yuxin Jiang, Yufei Wang, Xingshan Zeng, Wanjun Zhong, Liangyou Li, Fei Mi, Lifeng Shang, Xin Jiang, Qun Liu, Wei Wang
| Challenge: | Existing benchmarks focus on evaluating pure response quality, rather than assessing whether the response follows constraints stated in the instruction. |
| Approach: | They propose a Multi-level Fine-grained Constraints Following Benchmark for Large Language Models that adds a single constraint to the initial instruction at each increased level. |
| Outcome: | The proposed model can follow instructions with more constraints, and is deemed to have better instruction-following ability. |
Topic Memory Networks for Short Text Classification (D18-1)
Copied to clipboard
| Challenge: | Existing classification models for short texts are weak due to data sparsity . |
| Approach: | They propose topic memory networks for short text classification with a novel topic memory mechanism to encode latent topic representations indicative of class labels. |
| Outcome: | The proposed model outperforms state-of-the-art models on short text classification, while generating coherent topics. |
VisFinEval: A Scenario-Driven Chinese Multimodal Benchmark for Holistic Financial Understanding (2025.emnlp-main)
Copied to clipboard
Zhaowei Liu, Xin Guo, Haotian Xia, Lingfeng Zeng, Fangqi Lou, Jinyi Niu, Mengping Li, Qi Qi, Jiahuan Li, Wei Zhang, Yinglong Wang, Weige Cai, Weining Shen, Liwen Zhang
| Challenge: | Existing benchmarks focus on text comprehension, but MLLMs lack the ability to integrate visual data over financial visuals. |
| Approach: | They evaluate 21 state-of-the-art multimodal large language models in a zero-shot setting . they use an annotated question–answer pair from eight common financial image modalities . |
| Outcome: | The new benchmark outperforms existing models but trailed financial experts by 14 percentage points. |
Task-guided Disentangled Tuning for Pretrained Language Models (2022.findings-acl)
Copied to clipboard
| Challenge: | Pretrained language models are fine-tuned on task-specific datasets, but fail to capture task- specific patterns. |
| Approach: | They propose a method which disentangles task-relevant signals from entangled representations. |
| Outcome: | The proposed method improves generalization of representations by disentangling task-relevant signals from the entangled representations. |
On the Generation of Medical Dialogs for COVID-19 (2021.acl-short)
Copied to clipboard
Meng Zhou, Zechen Li, Bowen Tan, Guangtao Zeng, Wenmian Yang, Xuehai He, Zeqian Ju, Subrato Chakravorty, Shu Chen, Xingyi Yang, Yichen Zhang, Qingyang Wu, Zhou Yu, Kun Xu, Eric Xing, Pengtao Xie
| Challenge: | under the pandemic of COVID-19, people experiencing COVI D19-related symptoms have a pressing need to consult doctors. |
| Approach: | They develop a medical dialog system that can provide COVID19-related consultations . they use two dialog datasets containing conversations between doctors and patients . |
| Outcome: | The proposed system can provide COVID19-related consultations, but is too small compared with general-domain dialog datasets. |
Chain-of-Probe: Examining the Necessity and Accuracy of CoT Step-by-Step (2025.findings-naacl)
Copied to clipboard
Zezhong Wang, Xingshan Zeng, Weiwen Liu, Yufei Wang, Liangyou Li, Yasheng Wang, Lifeng Shang, Xin Jiang, Qun Liu, Kam-Fai Wong
| Challenge: | Current research found the issue of Early Answering in large language models where the models already have an answer before generating the Chain-of-Thought (CoT). |
| Approach: | They propose a method to probe changes in confidence during the model’s reasoning and prioritize answers with correct reasoning among multiple candidates. |
| Outcome: | The proposed method reveals that in a significant number of question-answer cases, CoT appears to be unnecessary and this necessity correlates with the simplicity of the task, defined by the reasoning steps required. |
FARSS: Fisher-Optimized Adaptive Low-Rank and Singular-Vector Selection for Knowledge-Preserving Fine-Tuning (2026.findings-acl)
Copied to clipboard
Renxing Chen, Ziwei Xiang, Peisong Wang, Hongjian Fang, Meng Li, Fanhu Zeng, Yanan Zhu, Peipei Yang, Xu-Yao Zhang, Jian Cheng
| Challenge: | Low-rank adaptation methods for large language models have limitations in preserving world knowledge and limiting updates to preserve world knowledge. |
| Approach: | They propose a Fisher-optimized adaptive low Rank and Singular-VectorSelection framework for knowledge-preserving fine-tuning that allows efficient and task-sensitive updates. |
| Outcome: | The proposed framework outperforms existing methods for knowledge-preserving fine-tuning. |
Knowledge-Enhanced Natural Language Inference Based on Knowledge Graphs (2020.coling-main)
Copied to clipboard
| Challenge: | Existing approaches to natural language inference rely on semantic knowledge, but background knowledge is limited to a few specific types. |
| Approach: | They propose a Knowledge Graph-enhanced NLI model that leverages background knowledge stored in knowledge graphs to facilitate inference. |
| Outcome: | The proposed model can leverage background knowledge stored in knowledge graphs to perform the task. |
DAST: Context-Aware Compression in LLMs via Dynamic Allocation of Soft Tokens (2025.findings-acl)
Copied to clipboard
Shaoshen Chen, Yangning Li, Zishan Xu, Yongqin Zeng, Shunlong Wu, Xinshuo Hu, Zifei Shan, Xin Su, Jiwei Tang, Yinghui Li, Hai-Tao Zheng
| Challenge: | Existing semantic vector-based compression methods do not account for the intrinsic information density variations between context chunks, instead allocating soft tokens uniformly across context chunk. |
| Approach: | They propose a method that leverages the LLM's intrinsic understanding of contextual relevance to guide compression. |
| Outcome: | The proposed method surpasses state-of-the-art methods on long context tasks. |
LCQMC:A Large-scale Chinese Question Matching Corpus (C18-1)
Copied to clipboard
| Challenge: | Existing methods for question answering system lack large-scale question matching corpora . lack of large-sized question matching results in problem solving . |
| Approach: | They propose a large-scale Chinese question matching corpus which is released to the public . they use a search engine to collect large-sized question pairs related to high-frequency words . |
| Outcome: | The proposed corpus is more general than paraphrase corpus as it focuses on intent matching rather than paraphrasing. |
LexMatcher: Dictionary-centric Data Curation for LLM-based Machine Translation (2024.findings-emnlp)
Copied to clipboard
| Challenge: | emergence of large language models (LLMs) has brought about new opportunities for machine translation. |
| Approach: | They propose a method for data curation that supplements the infrequent senses of polysemous words. |
| Outcome: | The proposed method outperforms established baselines on the WMT2022 test sets and is applicable to other pre-trained models. |
DrFrattn: Directly Learn Adaptive Policy from Attention for Simultaneous Machine Translation (2025.emnlp-main)
Copied to clipboard
| Challenge: | Existing approaches to learn read/write policies from attention mechanism may compromise effectiveness of attention mechanism . |
| Approach: | They propose a method that directly learns adaptive policies from the attention mechanism . experimental results demonstrate that the method achieves an improved balance between translation accuracy and latency. |
| Outcome: | The proposed method achieves improved balance between translation accuracy and latency. |
Self-Improvement Programming for Temporal Knowledge Graph Question Answering (2024.lrec-main)
Copied to clipboard
| Challenge: | Existing methods implicitly model time constraints by learning time-aware embeddings of questions and candidate answers, which is far from understanding the question comprehensively. |
| Approach: | They propose a temporal-based temporal programming method that leverages the in-context learning ability of Large Language Models to understand combinatory time constraints in questions. |
| Outcome: | The proposed method outperforms existing methods on multiTQ and CronQuestions datasets and is highly efficient on multi-level questions. |
AdaPrompt: Adaptive Model Training for Prompt-based NLP (2022.findings-emnlp)
Copied to clipboard
| Challenge: | Prompt-based learning can tackle zero-shot and few-shot NLP tasks . authors propose a method that makes use of pre-trained language models . |
| Approach: | They propose to map NLP tasks into natural language prompts, which are then filled by pre-trained language models. |
| Outcome: | The proposed method outperforms standard prompt-based methods in few-shot settings. |
LongBench: A Bilingual, Multitask Benchmark for Long Context Understanding (2024.acl-long)
Copied to clipboard
Yushi Bai, Xin Lv, Jiajie Zhang, Hongchang Lyu, Jiankai Tang, Zhidian Huang, Zhengxiao Du, Xiao Liu, Aohan Zeng, Lei Hou, Yuxiao Dong, Jie Tang, Juanzi Li
| Challenge: | Large language models (LLMs) can only handle texts a few thousand tokens long, limiting their applications on longer sequence inputs, such as books, reports, and codebases. |
| Approach: | They propose a bilingual, multi-task benchmark for long context understanding that extends context windows and more sophisticated memory mechanisms to improve models' long context capabilities. |
| Outcome: | The proposed model outperforms open-source models but struggles on longer contexts. |
MirageBackdoor: A Stealthy Attack that Induces Think-Well-Answer-Wrong Reasoning (2026.acl-long)
Copied to clipboard
| Challenge: | Existing CoT backdoor attacks manipulate intermediate reasoning steps to steer the model toward incorrect answers, but these corrupted reasoning traces are readily detected by prevalent process-monitoring defenses. |
| Approach: | They propose a backdoor attack that exploits the model's post-output space to preserve clean CoTs while selectively steering the final answer toward a specific target. |
| Outcome: | Experiments show that MirageBD achieves over 90% success rate across four datasets and five models with a poison ratio of only 5%. |
Consistency Regularization Training for Compositional Generalization (2023.acl-long)
Copied to clipboard
| Challenge: | Existing neural models have difficulty generalizing to unseen combinations of seen components. |
| Approach: | They propose to improve the capability of Transformer on compositional generalization by consistency regularization training without modifying model architectures. |
| Outcome: | The proposed model performs well on semantic parsing and machine translation benchmarks. |
Unpaired Sentiment-to-Sentiment Translation: A Cycled Reinforcement Learning Approach (P18-1)
Copied to clipboard
| Challenge: | Existing studies for sentiment-to-sentiment "translation" only change the underlying sentiment and fail to keep the semantic content. |
| Approach: | They propose a cycled reinforcement learning method that combines neutralization module and emotionalization module. |
| Outcome: | The proposed method outperforms state-of-the-art systems on Yelp and Amazon review datasets. |
Uncertainty Unveiled: Can Exposure to More In-context Examples Mitigate Uncertainty for Large Language Models? (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent advances in handling long sequences have unlocked new possibilities for long-context in-contact learning (ICL). |
| Approach: | They investigate how increased examples influence predictive uncertainty . they quantify uncertainty across different “shot” configurations and focus on EU . |
| Outcome: | The proposed model reduces uncertainty in simple and complex tasks by injecting task-specific knowledge. |
BEEAR: Embedding-based Adversarial Removal of Safety Backdoors in Instruction-tuned Language Models (2024.emnlp-main)
Copied to clipboard
| Challenge: | Safety backdoors in large language models can be triggered while evading detection during normal interactions. |
| Approach: | They propose a bi-level optimization method that uses a key insight: backdoor triggers induce a uniform drift in the model’s embedding space . inner level identifies universal perturbations to the decoder’s embedded spaces that steer the model towards defender-defined unwanted behaviors; outer level fine-tunes the model to reinforce safe behaviors against these perturbations. |
| Outcome: | The proposed mitigation method reduces the success rate of safety backdoor attacks from over 95% to 1% for general harmful behaviors and from 47% to 0% for Sleeper Agents, without compromising the model’s usefulness. |
Evaluating Large Language Models on Wikipedia-Style Survey Generation (2024.findings-acl)
Copied to clipboard
Fan Gao, Hang Jiang, Rui Yang, Qingcheng Zeng, Jinghui Lu, Moritz Blum, Tianwei She, Yuang Jiang, Irene Li
| Challenge: | Recent studies have shown that large language models can perform well in general tasks, but their effectiveness and limitations in domainspecific tasks remain unclear. |
| Approach: | They examine the proficiency of Large Language Models (LLMs) in generating succinct survey articles specific to the niche field of NLP in computer science. |
| Outcome: | The LLMs perform better in generating succinct survey articles specific to the niche field of NLP in computer science, compared to human-authored surveys, but they exhibit bias in evaluation. |
Conditional Memory via Scalable Lookup: A New Axis of Sparsity for Large Language Models (2026.acl-long)
Copied to clipboard
Xin Cheng, Wangding Zeng, Damai Dai, Qinyu Chen, Bingxuan Wang, Zhenda Xie, Kezhao Huang, Xingkai Yu, Zhewen Hao, Han Zhang, Yu-Kun Li, Huishuai Zhang, Dongyan Zhao, Wenfeng Liang
| Challenge: | Mixture-of-Experts (MoE) scales capacity via conditional computation, but lacks knowledge lookup primitive. |
| Approach: | They propose a conditional memory instantiated via Deep Sparse Embedding (DSE) they propose 'u-shaped scaling law' that identifies optimal balance between MoE experts and DSE memory . |
| Outcome: | The proposed model outperforms an iso-parameter and isoFLOPs MoE baseline across knowledge and reasoning benchmarks and is infrastructure-efficient. |
Gradient-Based Adversarial Factual Consistency Evaluation for Abstractive Summarization (2021.emnlp-main)
Copied to clipboard
| Challenge: | Abstractive summarization models often produce inconsistent statements or false facts. |
| Approach: | They propose an efficient weak-supervised adversarial data augmentation approach to generate factual consistency datasets by backpropagating gradients on token embeddings. |
| Outcome: | The proposed model can make interpretable factual errors tracing on public datasets and is cost-effective. |
DiffER: Diffusion Entity-Relation Modeling for Reversal Curse in Diffusion Large Language Models (2026.findings-acl)
Copied to clipboard
| Challenge: | Existing large language models exhibit unidirectional behavior when processing bidirectional relationships . authors propose a solution to alleviate the reversal curse in Diffusion LLMs . |
| Approach: | They propose a model that addresses the "reversal curse" of bidirectional behavior in large language models . they propose 'entity-aware training' and balanced data construction to alleviate asymmetry and missing relations . |
| Outcome: | The proposed model alleviates the "reversal curse" in Diffusion LLMs . the proposed model employs whole-entity masking to mitigate entity fragmentation . |
OS Agents: A Survey on MLLM-based Agents for Computer, Phone and Browser Use (2025.acl-long)
Copied to clipboard
Xueyu Hu, Tao Xiong, Biao Yi, Zishu Wei, Ruixuan Xiao, Yurun Chen, Jiasheng Ye, Meiling Tao, Xiangxin Zhou, Ziyu Zhao, Yuhuai Li, Shengze Xu, Shenzhi Wang, Xinchen Xu, Shuofei Qiao, Zhaokai Wang, Kun Kuang, Tieyong Zeng, Liang Wang, Jiwei Li, Yuchen Eleanor Jiang, Wangchunshu Zhou, Guoyin Wang, Keting Yin, Zhou Zhao, Hongxia Yang, Fan Wu, Shengyu Zhang, Fei Wu
| Challenge: | a new generation of (M)LLMs is enabling the creation of superintelligent AI assistants . OS Agents can complete tasks autonomously and have the potential to significantly enhance the lives of billions of users worldwide. |
| Approach: | They propose to build OS Agents that operate within operating systems' GUIs and GUIs . they examine evaluation metrics and benchmarks to identify promising directions . |
| Outcome: | The proposed agents are based on operating systems (OS) and operating systems frameworks. |
EFSA: Towards Event-Level Financial Sentiment Analysis (2024.acl-long)
Copied to clipboard
| Challenge: | a large-scale Chinese dataset contains 12,160 news articles and 13,725 quintuples . a four-hop Chain-of-Thought LLM-based approach is devised for this task . |
| Approach: | They propose to extend financial sentiment analysis to event-level since events usually serve as the subject of the sentiment in financial text. |
| Outcome: | The proposed method can reach the current state-of-the-art on a large-scale Chinese dataset. |
DynamicKV: Task-Aware Adaptive KV Cache Compression for Long Context LLMs (2025.findings-emnlp)
Copied to clipboard
| Challenge: | Existing KV cache compression methods enforce a fixed pattern, neglecting task-specific characteristics, which hampers the effective retention of essential information while discarding less important tokens. |
| Approach: | They propose a Task-Aware KV cache mechanism that dynamically adjusts the KV caching size across different layers based on the characteristics of the tasks. |
| Outcome: | The proposed method surpasses state-of-the-art methods by 11% on the LongBench dataset even under extreme compression (0.9%) |
OssCSE: Overcoming Surface Structure Bias in Contrastive Learning for Unsupervised Sentence Embedding (2023.emnlp-main)
Copied to clipboard
Zhan Shi, Guoyin Wang, Ke Bai, Jiwei Li, Xiang Li, Qingjun Cui, Belinda Zeng, Trishul Chilimbi, Xiaodan Zhu
| Challenge: | Recent studies show that contrastive learning is effective in sentence representation learning . but, the surface structure bias is a problem in the current model . |
| Approach: | They propose to combine a sentence with a sub-semantic sentence to investigate the surface structure bias. |
| Outcome: | The proposed model achieves state-of-the-art on standard semantic textual similarity tasks using different pre-trained backbones. |
LOTUS: Evolving Multimodal Unlearning via Hyperbolic Entailment and Lorentz Transport (2026.acl-long)
Copied to clipboard
| Challenge: | Existing unlearning methods suffer from a geometric mismatch, causing catastrophic forgetting or unsafe substitution. |
| Approach: | They propose a framework for surgical semantic pruning within the Lorentz manifold. |
| Outcome: | Experiments on MLLMU-Bench show that LOTUS significantly outperforms baselines while maintaining general utility. |
Alleviating Performance Degradation Caused by Out-of-Distribution Issues in Embedding-Based Retrieval (2025.findings-emnlp)
Copied to clipboard
Haotong Bao, Jianjin Zhang, Qi Chen, Weihao Han, Zhengxin Zeng, Ruiheng Chang, Mingzheng Li, Hao Sun, Weiwei Deng, Feng Sun, Qi Zhang
| Challenge: | Recent studies reveal query out-of-distribution issues degrading ANN performance . a distribution regularizer is introduced into the encoder training objective to encourage alignment between query and base embeddings. |
| Approach: | They introduce a distribution regularizer into the encoder training objective to encourage alignment between query and base embeddings. |
| Outcome: | The proposed method consistently improves retrieval performance across multiple datasets. |
AIRepr: An Analyst-Inspector Framework for Evaluating Reproducibility of LLMs in Data Science (2025.findings-emnlp)
Copied to clipboard
| Challenge: | Large language models are increasingly used to automate data analysis, but data science tasks often admit multiple statistically valid solutions. |
| Approach: | They propose a framework to evaluate LLM-generated code and assess its reproducibility . they introduce two reproducibility-enhancing prompting strategies and benchmark them against standard prompting . |
| Outcome: | The proposed framework improves reproducibility of large language models . it provides a foundation for transparent, reliable, and efficient human–AI collaboration in data science. |
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
| 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. |
Backdoor Attacks on Pre-trained Models by Layerwise Weight Poisoning (2021.emnlp-main)
Copied to clipboard
| Challenge: | Pre-trained models can be maliciously poisoned with certain triggers, causing a security threat. |
| Approach: | They propose a stronger weight poisoning attack method that introduces a layerwise weight poison strategy to plant deeper backdoors. |
| Outcome: | The proposed method can be widely applied and provide hints for future models robustness studies. |
Evaluating Generalization Capability of Language Models across Abductive, Deductive and Inductive Logical Reasoning (2025.coling-main)
Copied to clipboard
| Challenge: | Recent research in language models (LMs) have demonstrated remarkable performance on many natural language tasks, yet to what extent LMs possess the capability of generalizing to unseen logical rules remains unclear. |
| Approach: | They propose to use a dataset to assess the generalization capabilities of LMs on ADI reasoning to assess their generalization abilities. |
| Outcome: | The proposed dataset shows that LMs perform poorly on ADI reasoning tasks and lacks generalization capabilities. |
Label-Enhanced Hierarchical Contextualized Representation for Sequential Metaphor Identification (2021.emnlp-main)
Copied to clipboard
| Challenge: | Recent approaches to identify metaphors ignore extra information from data, such as contextual information and broader discourse information. |
| Approach: | They propose a model augmented with hierarchical contextualized representation to extract more information from both sentence-level and discourse-level. |
| Outcome: | The proposed model outperforms state-of-the-art methods on two tasks using a VUA dataset. |
Asymmetric Relational-Geometry Driven Universal Adversarial Perturbations for Vision-Language Models (2026.findings-acl)
Copied to clipboard
| Challenge: | Existing universal adversarial perturbation (UAP) methods suffer from limited cross-model transferability in black-box scenarios. |
| Approach: | They propose an optimization-based framework that learns universal perturbations under an asymmetric relational-geometry driven objective. |
| Outcome: | The proposed framework outperforms state-of-the-art models in black-box transfer settings. |
Activation-Guided Local Editing for Jailbreaking Attacks (2026.acl-long)
Copied to clipboard
| Challenge: | Existing methods for jailbreaking Large Language Models (LLMs) are limited and produce incoherent or unreadable inputs. |
| Approach: | They propose a two-stage framework that performs a one-shot, scenario-based generation of context and rephrases the original malicious query to obscure its harmful intent. |
| Outcome: | The proposed framework achieves state-of-the-art Attack Success Rate, with gains of up to 37.74% over the strongest baseline, and excellent transferability to black-box and large-scale models. |
Latent-Condensed Transformer for Efficient Long Context Modeling (2026.acl-long)
Copied to clipboard
| Challenge: | Existing approaches address these bottlenecks separately: Multi-head Latent Attention (MLA) reduces the KV cache by projecting tokens into a low-dimensional latent space, while sparse attention reduces computation. |
| Approach: | They propose a Latent-Condensed Attention mechanism that performs structured context condensation directly within MLA's latent space. |
| Outcome: | The proposed approach reduces KV cache size and attention cost without adding parameters. |
Attention Calibration for Transformer in Neural Machine Translation (2021.acl-long)
Copied to clipboard
| Challenge: | Attention mechanisms have been ubiquitous in neural machine translation (NMT) however, many studies doubt whether highlyattended inputs have a large impact on the model outputs. |
| Approach: | They propose to introduce a mask perturbation model that automatically evaluates each input’s contribution to the model outputs. |
| Outcome: | The proposed model is more uniform at lower layers while more concentrated on the specific inputs at higher layers. |
Aggregation of Reasoning: A Hierarchical Framework for Enhancing Answer Selection in Large Language Models (2024.lrec-main)
Copied to clipboard
Zhangyue Yin, Qiushi Sun, Qipeng Guo, Zhiyuan Zeng, Xiaonan Li, Tianxiang Sun, Cheng Chang, Qinyuan Cheng, Ding Wang, Xiaofeng Mou, Xipeng Qiu, Xuanjing Huang
| Challenge: | Recent advances in Chain-of-Thought prompting have facilitated significant breakthroughs for Large Language Models (LLMs) in complex reasoning tasks. |
| Approach: | They propose a hierarchical reasoning aggregation framework to address this problem . they propose dynamic sampling to adjust the number of reasoning chains . |
| Outcome: | The proposed framework outperforms existing ensemble methods on complex reasoning tasks. |
TiMem: Temporal-Hierarchical Memory Consolidation for Long-Horizon Conversational Agents (2026.findings-acl)
Copied to clipboard
Kai Li, Xuanqing Yu, Ziyi Ni, Yi Zeng, Yao Xu, Zheqing Zhang, Xin Li, Jitao Sang, Xiaogang Duan, Xuelei Wang, Chengbao Liu, Jie Tan
| Challenge: | Existing memory frameworks provide limited support for temporally structured information across hierarchical levels, leading to fragmented memories and unstable long-horizon personalization. |
| Approach: | They propose a temporal–hierarchical memory framework that organizes conversations through a Temporal Memory Tree. |
| Outcome: | The proposed framework outperforms baselines while reducing the recalled memory length by 52.20%. |
An Efficient Coarse-to-Fine Facet-Aware Unsupervised Summarization Framework Based on Semantic Blocks (2022.coling-1)
Copied to clipboard
| Challenge: | Existing unsupervised summarization methods fail to consider efficiency and effectiveness when the input document is extremely long. |
| Approach: | They propose an efficient Coarse-to-Fine Facet-Aware Ranking framework for unsupervised long document summarization based on the semantic block. |
| Outcome: | The proposed framework can achieve new state-of-the-art unsupervised summarization results on Gov-Report, billSum, arXiv, and PubMed. |
Joint Effects of Context and User History for Predicting Online Conversation Re-entries (P19-1)
Copied to clipboard
| Challenge: | Existing methods for predicting online conversation re-entry focus on modeling engagement patterns in ongoing conversations or ignoring the rich information in users' previous chatting history. |
| Approach: | They propose a neural framework with three main layers to model the conversation context and user history and their interactions with Twitter and Reddit to predict whether a user will return to a conversation they once participated in. |
| Outcome: | The proposed framework outperforms the state-of-the-art methods on two large-scale Twitter and Reddit conversations, and achieves an F1 score of 61.1 on Twitter conversations. |
Towards a Mechanistic Interpretation of Multi-Step Reasoning Capabilities of Language Models (2023.emnlp-main)
Copied to clipboard
Yifan Hou, Jiaoda Li, Yu Fei, Alessandro Stolfo, Wangchunshu Zhou, Guangtao Zeng, Antoine Bosselut, Mrinmaya Sachan
| Challenge: | Recent work has shown that language models (LMs) have strong multi-step (i.e., procedural) reasoning capabilities. |
| Approach: | They propose a mechanistic interpretation of language models for multi-step reasoning tasks by introducing a new probing approach that recovers the reasoning tree from the model’s attention patterns. |
| Outcome: | The proposed model implicitly embeds a reasoning tree resembling the correct reasoning process within it, and detects the information from the model’s attention patterns for most examples. |
MT-Eval: A Multi-Turn Capabilities Evaluation Benchmark for Large Language Models (2024.emnlp-main)
Copied to clipboard
Wai-Chung Kwan, Xingshan Zeng, Yuxin Jiang, Yufei Wang, Liangyou Li, Lifeng Shang, Xin Jiang, Qun Liu, Kam-Fai Wong
| Challenge: | Existing evaluation frameworks focus on single-turn evaluations, overlooking the models’ capabilities in multi-turn interactions. |
| Approach: | They propose a benchmark to evaluate the multi-turn conversational abilities of large language models (LLMs) by analyzing human-LLM conversations and constructing multi-turned queries for each category using GPT-4. |
| Outcome: | The proposed model outperforms open-source models in multi-turn tasks while retaining and recalling historical information. |
Multimodal Misinformation Detection by Learning from Synthetic Data with Multimodal LLMs (2024.findings-emnlp)
Copied to clipboard
| Challenge: | Obtaining large-scale, high-quality real-world fact-checking datasets is costly . generalizability of detectors trained on synthetic data to real-life scenarios remains unclear . |
| Approach: | They propose to use synthetic data to learn from real-world data to detect multimodal misinformation . they propose to combine model-agnostic data selection methods with real-life data distributions . |
| Outcome: | The proposed method improves the performance of a small MLLM on real-world fact-checking datasets, surpassing GPT-4V. |
LC4EE: LLMs as Good Corrector for Event Extraction (2024.findings-acl)
Copied to clipboard
| Challenge: | Event extraction (EE) is a critical task in natural language processing, yet deploying a practical EE system remains challenging. |
| Approach: | They propose to leverage the superior extraction capability of LLMs and instruction-following ability of LRMs to construct a robust and highly available EE system. |
| Outcome: | The proposed method can identify and correct errors in SLMs predictions based on automatically generated feedback information and improve performance. |
Logic-of-Thought: Injecting Logic into Contexts for Full Reasoning in Large Language Models (2025.naacl-long)
Copied to clipboard
Tongxuan Liu, Wenjiang Xu, Weizhe Huang, Yuting Zeng, Jiaxing Wang, Xingyu Wang, Hailong Yang, Jing Li
| Challenge: | Large Language Models (LLMs) have demonstrated remarkable capabilities across various tasks but their performance in complex logical reasoning tasks remains unsatisfactory. |
| Approach: | They propose a propositional logic prompting method which generates expanded logical information descriptions and utilizes them as an additional augmentation to original contexts. |
| Outcome: | Extensive experiments show that Logic-of-Thought boosts the performance of various prompting methods with a striking margin across five logical reasoning tasks. |
KuiLeiXi: a Chinese Open-Ended Text Adventure Game (2021.acl-demo)
Copied to clipboard
Yadong Xi, Xiaoxi Mao, Le Li, Lei Lin, Yanjiang Chen, Shuhan Yang, Xuhan Chen, Kailun Tao, Zhi Li, Gongzheng Li, Lin Jiang, Siyan Liu, Zeng Zhao, Minlie Huang, Changjie Fan, Zhipeng Hu
| Challenge: | Recent advances in pre-trained language models have made it possible to generate human-like text. |
| Approach: | They propose to integrate an open-ended text adventure game in Chinese, named KuiLeiXi, where players interact with the AI until the plot goals are reached. |
| Outcome: | The proposed game lacks incentives and relies on players to explore on their own. |
Mitigating Biases of Large Language Models in Stance Detection with Counterfactual Augmented Calibration (2025.naacl-long)
Copied to clipboard
Ang Li, Jingqian Zhao, Bin Liang, Lin Gui, Hui Wang, Xi Zeng, Xingwei Liang, Kam-Fai Wong, Ruifeng Xu
| Challenge: | Large language models generate biased stances due to spurious correlations and preference towards certain individuals and topics. |
| Approach: | They propose a counterfactual Augmented Calibration Network to calibrate potential bias in stance detection of large language models. |
| Outcome: | The proposed calibration network can mitigate biases of large language models, achieving state-of-the-art results. |