title stringlengths 16 168 | paper_url stringlengths 43 46 | authors listlengths 1 32 | abstract large_stringlengths 346 1.96k | anthology_id stringlengths 17 20 | doi stringlengths 29 32 | award stringclasses 7
values | arxiv_id stringlengths 10 10 ⌀ | arxiv_id_source stringclasses 2
values | embedding listlengths 768 768 |
|---|---|---|---|---|---|---|---|---|---|
Towards Automated Error Discovery: A Study in Conversational AI | https://aclanthology.org/2025.emnlp-main.1/ | [
"Dominic Petrak",
"Thy Thy Tran",
"Iryna Gurevych"
] | Although LLM-based conversational agents demonstrate strong fluency and coherence, they still produce undesirable behaviors (errors) that are challenging to prevent from reaching users during deployment. Recent research leverages large language models (LLMs) to detect errors and guide response-generation models toward ... | 2025.emnlp-main.1 | 10.18653/v1/2025.emnlp-main.1 | null | 2509.10833 | title_snapshot | [
0.01677408441901207,
-0.006571752484887838,
-0.011267205700278282,
0.03930709511041641,
0.053435757756233215,
0.02759828232228756,
0.037908248603343964,
0.009079988114535809,
-0.0015171505510807037,
0.005762510932981968,
-0.03068770095705986,
0.071064792573452,
-0.05661594867706299,
-0.014... |
Break the Checkbox: Challenging Closed-Style Evaluations of Cultural Alignment in LLMs | https://aclanthology.org/2025.emnlp-main.2/ | [
"Mohsinul Kabir",
"Ajwad Abrar",
"Sophia Ananiadou"
] | A large number of studies rely on closed-style multiple-choice surveys to evaluate cultural alignment in Large Language Models (LLMs). In this work, we challenge this constrained evaluation paradigm and explore more realistic, unconstrained approaches. Using the World Values Survey (WVS) and Hofstede Cultural Dimension... | 2025.emnlp-main.2 | 10.18653/v1/2025.emnlp-main.2 | null | 2502.08045 | title_snapshot | [
0.006895057391375303,
-0.00879618339240551,
-0.04348107427358627,
0.026769641786813736,
0.035498231649398804,
0.020766032859683037,
0.038009606301784515,
0.041499219834804535,
-0.02767367660999298,
-0.007768501061946154,
-0.0186980739235878,
0.02236972749233246,
-0.06968329101800919,
-0.02... |
Biased Tales: Cultural and Topic Bias in Generating Children’s Stories | https://aclanthology.org/2025.emnlp-main.3/ | [
"Donya Rooein",
"Vilém Zouhar",
"Debora Nozza",
"Dirk Hovy"
] | Stories play a pivotal role in human communication, shaping beliefs and morals, particularly in children. As parents increasingly rely on large language models (LLMs) to craft bedtime stories, the presence of cultural and gender stereotypes in these narratives raises significant concerns. To address this issue, we pres... | 2025.emnlp-main.3 | 10.18653/v1/2025.emnlp-main.3 | null | 2509.07908 | title_snapshot | [
-0.02190151996910572,
-0.008574865758419037,
-0.031289294362068176,
0.028900375589728355,
0.03652859106659889,
-0.020317258313298225,
0.037286464124917984,
0.034441445022821426,
-0.015700899064540863,
-0.005803718231618404,
-0.05666012316942215,
0.03847110643982887,
-0.06368476152420044,
0... |
Large Language Models as Realistic Microservice Trace Generators | https://aclanthology.org/2025.emnlp-main.4/ | [
"Donghyun Kim",
"Sriram Ravula",
"Taemin Ha",
"Alex Dimakis",
"Daehyeok Kim",
"Aditya Akella"
] | Workload traces are essential to understand complex computer systems’ behavior and manage processing and memory resources. Since real-world traces are hard to obtain, synthetic trace generation is a promising alternative. This paper proposes a first-of-a-kind approach that relies on training a large language model (LLM... | 2025.emnlp-main.4 | 10.18653/v1/2025.emnlp-main.4 | null | 2502.17439 | title_snapshot | [
-0.028535768389701843,
-0.02827896736562252,
-0.02354288287460804,
0.024413948878645897,
0.058196503669023514,
0.031881023198366165,
0.034105084836483,
0.02709246426820755,
-0.0321786068379879,
-0.00436309352517128,
0.0007474947487935424,
0.0003920183808077127,
-0.06036832183599472,
-0.013... |
JUDGEBERT: Assessing Legal Meaning Preservation Between Sentences | https://aclanthology.org/2025.emnlp-main.5/ | [
"David Beauchemin",
"Michelle Albert-Rochette",
"Richard Khoury",
"Pierre-Luc Déziel"
] | Simplifying text while preserving its meaning is a complex yet essential task, especially in sensitive domain applications like legal texts. When applied to a specialized field, like the legal domain, preservation differs significantly from its role in regular texts. This paper introduces FrJUDGE, a new dataset to asse... | 2025.emnlp-main.5 | 10.18653/v1/2025.emnlp-main.5 | null | 2508.16870 | title_snapshot | [
-0.02422008104622364,
-0.06431114673614502,
-0.015076806768774986,
0.035101812332868576,
0.035156119614839554,
-0.004684684332460165,
0.031936436891555786,
0.005657314322888851,
-0.015370882116258144,
-0.017949838191270828,
-0.03224113583564758,
0.0521978884935379,
-0.05160152539610863,
-0... |
QFrCoLA: a Quebec-French Corpus of Linguistic Acceptability Judgments | https://aclanthology.org/2025.emnlp-main.6/ | [
"David Beauchemin",
"Richard Khoury"
] | Large and Transformer-based language models perform outstandingly in various downstream tasks. However, there is limited understanding regarding how these models internalize linguistic knowledge, so various linguistic benchmarks have recently been proposed to facilitate syntactic evaluation of language models across la... | 2025.emnlp-main.6 | 10.18653/v1/2025.emnlp-main.6 | null | 2508.16867 | title_snapshot | [
-0.022529300302267075,
-0.02278587780892849,
-0.004519200883805752,
0.02250555343925953,
0.030868234112858772,
0.036385245621204376,
0.015756627544760704,
0.04443303123116493,
-0.01600053533911705,
-0.011535628698766232,
-0.035936228930950165,
0.06323778629302979,
-0.04605066403746605,
-0.... |
Revisiting LLM Value Probing Strategies: Are They Robust and Expressive? | https://aclanthology.org/2025.emnlp-main.7/ | [
"Siqi Shen",
"Mehar Singh",
"Lajanugen Logeswaran",
"Moontae Lee",
"Honglak Lee",
"Rada Mihalcea"
] | The value orientation of Large Language Models (LLMs) has been extensively studied, as it can shape user experiences across demographic groups.However, two key challenges remain: (1) the lack of systematic comparison across value probing strategies, despite the Multiple Choice Question (MCQ) setting being vulnerable to... | 2025.emnlp-main.7 | 10.18653/v1/2025.emnlp-main.7 | null | 2507.13490 | title_snapshot | [
-0.025236889719963074,
-0.03608259558677673,
-0.009495321661233902,
0.04093966260552406,
0.035322822630405426,
0.022078584879636765,
0.009372507221996784,
0.03870623558759689,
-0.00855945609509945,
-0.0042783115059137344,
-0.01661158725619316,
0.052309781312942505,
-0.04961811751127243,
-0... |
A Systematic Analysis of Base Model Choice for Reward Modeling | https://aclanthology.org/2025.emnlp-main.8/ | [
"Kian Ahrabian",
"Pegah Jandaghi",
"Negar Mokhberian",
"Sai Praneeth Karimireddy",
"Jay Pujara"
] | Reinforcement learning from human feedback (RLHF) and, at its core, reward modeling have become a crucial part of training powerful large language models (LLMs). One commonly overlooked factor in training high-quality reward models (RMs) is the effect of the base model, which is becoming more challenging to choose give... | 2025.emnlp-main.8 | 10.18653/v1/2025.emnlp-main.8 | null | 2505.10775 | title_snapshot | [
-0.03186706826090813,
-0.0018338762456551194,
0.004367852117866278,
0.046308957040309906,
0.05035069212317467,
0.008935345336794853,
0.005901757627725601,
0.025813572108745575,
-0.031672313809394836,
-0.014346005395054817,
-0.009148935787379742,
0.05510515347123146,
-0.061244573444128036,
... |
Comparing Specialised Small and General Large Language Models on Text Classification: 100 Labelled Samples to Achieve Break-Even Performance | https://aclanthology.org/2025.emnlp-main.9/ | [
"Branislav Pecher",
"Ivan Srba",
"Maria Bielikova"
] | When solving NLP tasks with limited labelled data, researchers typically either use a general large language model without further update, or use a small number of labelled samples to tune a specialised smaller model. In this work, we answer an important question – how many labelled samples are required for the special... | 2025.emnlp-main.9 | 10.18653/v1/2025.emnlp-main.9 | null | 2402.12819 | title_snapshot | [
-0.024328693747520447,
-0.04671136289834976,
-0.01230812631547451,
0.036079417914152145,
0.02704056166112423,
0.010770604945719242,
0.03587435558438301,
0.017564859241247177,
-0.03245067596435547,
0.00683560548350215,
-0.021485041826963425,
0.04480932280421257,
-0.06351985782384872,
-0.007... |
Is the Top Still Spinning? Evaluating Subjectivity in Narrative Understanding | https://aclanthology.org/2025.emnlp-main.10/ | [
"Melanie Subbiah",
"Akankshya Mishra",
"Grace Kim",
"Liyan Tang",
"Greg Durrett",
"Kathleen McKeown"
] | Determining faithfulness of a claim to a source document is an important problem across many domains. This task is generally treated as a binary judgment of whether the claim is supported or unsupported in relation to the source. In many cases, though, whether a claim is supported can be ambiguous. For instance, it may... | 2025.emnlp-main.10 | 10.18653/v1/2025.emnlp-main.10 | null | 2504.01132 | title_snapshot | [
-0.009753277525305748,
-0.005642018746584654,
-0.009497827850282192,
0.003556879935786128,
0.04743628576397896,
-0.026221150532364845,
0.027733759954571724,
0.004223274067044258,
-0.021015867590904236,
-0.0028200873639434576,
-0.06477029621601105,
0.03601768612861633,
-0.02960185520350933,
... |
MathTutorBench: A Benchmark for Measuring Open-ended Pedagogical Capabilities of LLM Tutors | https://aclanthology.org/2025.emnlp-main.11/ | [
"Jakub Macina",
"Nico Daheim",
"Ido Hakimi",
"Manu Kapur",
"Iryna Gurevych",
"Mrinmaya Sachan"
] | Evaluating the pedagogical capabilities of AI-based tutoring models is critical for making guided progress in the field. Yet, we lack a reliable, easy-to-use, and simple-to-run evaluation that reflects the pedagogical abilities of models. To fill this gap, we present MathTutorBench, an open-source benchmark for holisti... | 2025.emnlp-main.11 | 10.18653/v1/2025.emnlp-main.11 | null | 2502.18940 | title_snapshot | [
-0.007291925139725208,
-0.045136336237192154,
-0.02095877192914486,
0.02909492701292038,
0.031475577503442764,
0.0024474479723721743,
0.018918627873063087,
0.02213590033352375,
-0.01164040993899107,
-0.0031378583516925573,
-0.02085164189338684,
0.029141975566744804,
-0.03857224062085152,
-... |
Preemptive Detection and Correction of Misaligned Actions in LLM Agents | https://aclanthology.org/2025.emnlp-main.12/ | [
"Haishuo Fang",
"Xiaodan Zhu",
"Iryna Gurevych"
] | Deploying LLM-based agents in real-life applications often faces a critical challenge: the misalignment between agents’ behavior and user intent. Such misalignment may lead agents to unintentionally execute some critical actions that carry negative outcomes (e.g., accidentally triggering a \textit{buy-now} in web shopp... | 2025.emnlp-main.12 | 10.18653/v1/2025.emnlp-main.12 | null | 2407.11843 | title_snapshot | [
0.009785664267838001,
-0.027761494740843773,
-0.00568651407957077,
0.016487495973706245,
0.05551931634545326,
-0.013974870555102825,
0.0355675034224987,
0.03361770510673523,
-0.020772602409124374,
-0.021654793992638588,
-0.03623611107468605,
0.045998889952898026,
-0.06168268993496895,
-0.0... |
Fingerprinting LLMs through Survey Item Factor Correlation: A Case Study on Humor Style Questionnaire | https://aclanthology.org/2025.emnlp-main.13/ | [
"Simon Münker"
] | LLMs increasingly engage with psychological instruments, yet how they represent constructs internally remains poorly understood. We introduce a novel approach to “fingerprinting” LLMs through their factor correlation patterns on standardized psychological assessments to deepen the understanding of LLMs constructs repre... | 2025.emnlp-main.13 | 10.18653/v1/2025.emnlp-main.13 | null | null | null | [
0.01206789631396532,
0.011023597791790962,
0.013234232552349567,
-0.0016891004052013159,
0.07709737122058868,
0.03386140242218971,
0.030506698414683342,
0.023127363994717598,
-0.02166634611785412,
-0.006727918982505798,
-0.01574966125190258,
0.009422545321285725,
-0.02863292209804058,
-0.0... |
Gradient-Attention Guided Dual-Masking Synergetic Framework for Robust Text-based Person Retrieval | https://aclanthology.org/2025.emnlp-main.14/ | [
"Tianlu Zheng",
"Yifan Zhang",
"Xiang An",
"Ziyong Feng",
"Kaicheng Yang",
"Qichuan Ding"
] | Although Contrastive Language-Image Pre-training (CLIP) exhibits strong performance across diverse vision tasks, its application to person representation learning faces two critical challenges: (i) the scarcity of large-scale annotated vision-language data focused on person-centric images, and (ii) the inherent limitat... | 2025.emnlp-main.14 | 10.18653/v1/2025.emnlp-main.14 | null | 2509.09118 | title_snapshot | [
-0.007661794312298298,
-0.011464731767773628,
0.004360248800367117,
0.05344897136092186,
0.027711696922779083,
0.008914383128285408,
0.02822914533317089,
0.0053754327818751335,
-0.03697507828474045,
-0.0395168773829937,
-0.04832875356078148,
0.002846969524398446,
-0.08907653391361237,
-0.0... |
From Problem-Solving to Teaching Problem-Solving: Aligning LLMs with Pedagogy using Reinforcement Learning | https://aclanthology.org/2025.emnlp-main.15/ | [
"David Dinucu-Jianu",
"Jakub Macina",
"Nico Daheim",
"Ido Hakimi",
"Iryna Gurevych",
"Mrinmaya Sachan"
] | Large language models (LLMs) can transform education, but their optimization for direct question-answering often undermines effective pedagogy which requires strategically withholding answers. To mitigate this, we propose an online reinforcement learning (RL)-based alignment framework that can quickly adapt LLMs into e... | 2025.emnlp-main.15 | 10.18653/v1/2025.emnlp-main.15 | null | 2505.15607 | title_snapshot | [
-0.01822853460907936,
-0.021036559715867043,
-0.030822910368442535,
0.040367353707551956,
0.06338433921337128,
0.01485679391771555,
0.014317444525659084,
0.019408894702792168,
-0.02983405441045761,
0.01679909974336624,
-0.04729674011468887,
0.04177040979266167,
-0.06831838190555573,
-0.022... |
CompKBQA: Component-wise Task Decomposition for Knowledge Base Question Answering | https://aclanthology.org/2025.emnlp-main.16/ | [
"Yuhang Tian",
"Dandan Song",
"Zhijing Wu",
"Pan Yang",
"Changzhi Zhou",
"Jun Yang",
"Hao Wang",
"Huipeng Ma",
"Chenhao Li",
"Luan Zhang"
] | Knowledge Base Question Answering (KBQA) aims to extract accurate answers from the Knowledge Base (KB). Traditional Semantic Parsing (SP)-based methods are widely used but struggle with complex queries. Recently, large language models (LLMs) have shown promise in improving KBQA performance. However, the challenge of ge... | 2025.emnlp-main.16 | 10.18653/v1/2025.emnlp-main.16 | null | null | null | [
-0.00577358715236187,
-0.020848732441663742,
0.005412606988102198,
0.03516782820224762,
0.06062418222427368,
0.002277107210829854,
0.020660052075982094,
-0.026998132467269897,
-0.018300645053386688,
0.022432170808315277,
-0.058967430144548416,
0.05943929776549339,
-0.05999705195426941,
-0.... |
Permutative Preference Alignment from Listwise Ranking of Human Judgments | https://aclanthology.org/2025.emnlp-main.17/ | [
"Yang Zhao",
"Yixin Wang",
"Mingzhang Yin"
] | Aligning Large Language Models (LLMs) with human preferences is crucial in ensuring desirable and controllable model behaviors. Current methods, such as Reinforcement Learning from Human Feedback (RLHF) and Direct Preference Optimization (DPO), rely on the Bradley-Terry (B-T) model to maximize the likelihood of pairwis... | 2025.emnlp-main.17 | 10.18653/v1/2025.emnlp-main.17 | null | 2410.04346 | title_snapshot | [
-0.003403435694053769,
0.007591971196234226,
-0.01667514629662037,
0.02783346362411976,
0.02075805515050888,
0.02496364153921604,
0.009694353677332401,
0.01740109734237194,
-0.014499462209641933,
-0.04278244078159332,
-0.008060449734330177,
0.04552318900823593,
-0.08257170021533966,
-0.034... |
ToneCraft: Cantonese Lyrics Generation with Harmony of Tones and Pitches | https://aclanthology.org/2025.emnlp-main.18/ | [
"Junyu Cheng",
"Chang Pan",
"Shuangyin Li"
] | Lyrics generation has garnered increasing attention within the artificial intelligence community. Our task focuses on generating harmonious Cantonese lyrics. Unlike other languages, Cantonese has a unique system of nine contours and six tones, making it essential to satisfy the harmony rules that ensure the alignment b... | 2025.emnlp-main.18 | 10.18653/v1/2025.emnlp-main.18 | null | null | null | [
0.011265200562775135,
0.005177828948944807,
-0.01038727443665266,
0.034805234521627426,
0.0490860790014267,
0.0454958975315094,
0.052005741745233536,
0.03314138948917389,
-0.034435126930475235,
-0.031813185662031174,
-0.014019593596458435,
0.009557840414345264,
-0.0736398994922638,
0.00697... |
SensorLLM: Aligning Large Language Models with Motion Sensors for Human Activity Recognition | https://aclanthology.org/2025.emnlp-main.19/ | [
"Zechen Li",
"Shohreh Deldari",
"Linyao Chen",
"Hao Xue",
"Flora D. Salim"
] | We introduce SensorLLM, a two-stage framework that enables Large Language Models (LLMs) to perform human activity recognition (HAR) from sensor time-series data. Despite their strong reasoning and generalization capabilities, LLMs remain underutilized for motion sensor data due to the lack of semantic context in time-s... | 2025.emnlp-main.19 | 10.18653/v1/2025.emnlp-main.19 | null | 2410.10624 | title_snapshot | [
0.039567090570926666,
-0.016515633091330528,
-0.006957633886486292,
0.0027202137280255556,
0.059237029403448105,
0.016718925908207893,
0.048577699810266495,
0.017855439335107803,
-0.035095132887363434,
-0.01331708487123251,
-0.030863747000694275,
0.016288502141833305,
-0.06460640579462051,
... |
MixLoRA-DSI: Dynamically Expandable Mixture-of-LoRA Experts for Rehearsal-Free Generative Retrieval over Dynamic Corpora | https://aclanthology.org/2025.emnlp-main.20/ | [
"Tuan-Luc Huynh",
"Thuy-Trang Vu",
"Weiqing Wang",
"Trung Le",
"Dragan Gasevic",
"Yuan-Fang Li",
"Thanh-Toan Do"
] | Continually updating model-based indexes in generative retrieval with new documents remains challenging, as full retraining is computationally expensive and impractical under resource constraints. We propose MixLoRA-DSI, a novel framework that combines an expandable mixture of Low-Rank Adaptation experts with a layer-w... | 2025.emnlp-main.20 | 10.18653/v1/2025.emnlp-main.20 | null | 2507.09924 | title_snapshot | [
0.0013181809335947037,
-0.04664233326911926,
-0.01378446351736784,
0.04337843134999275,
0.024637671187520027,
0.01222268957644701,
-0.004286702256649733,
-0.009187460877001286,
-0.03573057800531387,
-0.0065968651324510574,
-0.03377002477645874,
0.024636046960949898,
-0.0519721619784832,
0.... |
ViClaim: A Multilingual Multilabel Dataset for Automatic Claim Detection in Videos | https://aclanthology.org/2025.emnlp-main.21/ | [
"Patrick Giedemann",
"Pius von Däniken",
"Jan Milan Deriu",
"Alvaro Rodrigo",
"Anselmo Peñas",
"Mark Cieliebak"
] | The growing influence of video content as a medium for communication and misinformation underscores the urgent need for effective tools to analyze claims in multilingual and multi-topic settings. Existing efforts in misinformation detection largely focus on written text, leaving a significant gap in addressing the comp... | 2025.emnlp-main.21 | 10.18653/v1/2025.emnlp-main.21 | null | 2504.12882 | title_snapshot | [
0.03820370137691498,
-0.015397031791508198,
0.01717408560216427,
0.056181639432907104,
0.024818439036607742,
-0.0026707726065069437,
0.03407555818557739,
0.01505753118544817,
-0.022018861025571823,
-0.00912137608975172,
-0.03841032460331917,
0.047011084854602814,
-0.0516611747443676,
0.010... |
DeepResearcher: Scaling Deep Research via Reinforcement Learning in Real-world Environments | https://aclanthology.org/2025.emnlp-main.22/ | [
"Yuxiang Zheng",
"Dayuan Fu",
"Xiangkun Hu",
"Xiaojie Cai",
"Lyumanshan Ye",
"Pengrui Lu",
"Pengfei Liu"
] | Large Language Models (LLMs) with web search capabilities show significant potential for deep research, yet current methods—brittle prompt engineering or RAG-based reinforcement learning in controlled environments—fail to capture real-world complexities. In this paper, we introduce DeepResearcher, the first comprehensi... | 2025.emnlp-main.22 | 10.18653/v1/2025.emnlp-main.22 | null | 2504.03160 | title_snapshot | [
-0.034041665494441986,
-0.06159686669707298,
-0.010898950509727001,
0.052669018507003784,
0.03707267343997955,
0.01431444101035595,
0.029445676133036613,
0.01213767472654581,
-0.004824507515877485,
-0.03696173429489136,
-0.03214734047651291,
0.024660782888531685,
-0.04017532989382744,
-0.0... |
Mixture of Length and Pruning Experts for Knowledge Graphs Reasoning | https://aclanthology.org/2025.emnlp-main.23/ | [
"Enjun Du",
"Siyi Liu",
"Yongqi Zhang"
] | Knowledge Graph (KG) reasoning, which aims to infer new facts from structured knowledge repositories, plays a vital role in Natural Language Processing (NLP) systems. Its effectiveness critically depends on constructing informative and contextually relevant reasoning paths. However, existing graph neural networks (GNNs... | 2025.emnlp-main.23 | 10.18653/v1/2025.emnlp-main.23 | null | 2507.20498 | title_snapshot | [
-0.01630253717303276,
-0.022135980427265167,
0.0024357344955205917,
0.05281759053468704,
0.027109656482934952,
0.026735445484519005,
0.03866814449429512,
-0.011194335296750069,
-0.04143736511468887,
-0.01861921325325966,
-0.04249799624085426,
0.06499497592449188,
-0.035878393799066544,
-0.... |
MPRF: Interpretable Stance Detection through Multi-Path Reasoning Framework | https://aclanthology.org/2025.emnlp-main.24/ | [
"ZhaoDan Zhang",
"Jin Zhang",
"Hui Xu",
"Jiafeng Guo",
"Xueqi Cheng"
] | Stance detection, a critical task in Natural Language Processing (NLP), aims to identify the attitude expressed in text toward specific targets. Despite advancements in Large Language Models (LLMs), challenges such as limited interpretability and handling nuanced content persist. To address these issues, we propose the... | 2025.emnlp-main.24 | 10.18653/v1/2025.emnlp-main.24 | null | null | null | [
0.0071084038354456425,
-0.042916439473629,
0.01063546072691679,
0.048967860639095306,
0.024021293967962265,
0.0402248278260231,
0.013332300819456577,
0.00598140899091959,
-0.02101070247590542,
-0.013814802281558514,
-0.01540831383317709,
0.04411589354276657,
-0.06345812976360321,
-0.014301... |
Analyzing the Effects of Supervised Fine-Tuning on Model Knowledge from Token and Parameter Levels | https://aclanthology.org/2025.emnlp-main.25/ | [
"Junjie Ye",
"Yuming Yang",
"Yang Nan",
"Shuo Li",
"Qi Zhang",
"Tao Gui",
"Xuanjing Huang",
"Peng Wang",
"Zhongchao Shi",
"Jianping Fan"
] | Large language models (LLMs) acquire substantial world knowledge during pre-training, which is further shaped by post-training techniques such as supervised fine-tuning (SFT). However, the impact of SFT on a model’s knowledge remains underexplored, limiting our ability to control knowledge behavior in fine-tuned models... | 2025.emnlp-main.25 | 10.18653/v1/2025.emnlp-main.25 | null | 2509.16596 | title_snapshot | [
-0.014765756204724312,
-0.02342117577791214,
-0.0009062594617716968,
0.027805794030427933,
0.06797647476196289,
-0.01208381075412035,
0.03566721826791763,
0.00927671231329441,
-0.015555247664451599,
0.01667845994234085,
-0.018376801162958145,
0.058199964463710785,
-0.03392210602760315,
-0.... |
JI^2S: Joint Influence‐Aware Instruction Data Selection for Efficient Fine‐Tuning | https://aclanthology.org/2025.emnlp-main.26/ | [
"Jingyu Wei",
"Bo Liu",
"Tianjiao Wan",
"Baoyun Peng",
"Xingkong Ma",
"Mengmeng Guo"
] | Instruction tuning (IT) improves large language models (LLMs) by aligning their outputs with human instructions, but its success depends critically on training data quality, and datasets such as Alpaca often contain noisy or suboptimal examples that undermine fine‐tuning. Prior selection strategies score samples using ... | 2025.emnlp-main.26 | 10.18653/v1/2025.emnlp-main.26 | null | null | null | [
-0.007297221105545759,
-0.018249990418553352,
-0.028160803020000458,
0.03533423691987991,
0.017028439790010452,
0.013417859561741352,
0.03872105851769447,
0.017562657594680786,
0.013467954471707344,
-0.01872861199080944,
0.004470454063266516,
0.045199595391750336,
-0.04444195330142975,
-0.... |
SoundMind: RL-Incentivized Logic Reasoning for Audio-Language Models | https://aclanthology.org/2025.emnlp-main.27/ | [
"Xingjian Diao",
"Chunhui Zhang",
"Keyi Kong",
"Weiyi Wu",
"Chiyu Ma",
"Zhongyu Ouyang",
"Peijun Qing",
"Soroush Vosoughi",
"Jiang Gui"
] | While large language models have demonstrated impressive reasoning abilities, their extension to the audio modality, particularly within large audio-language models (LALMs), remains underexplored. Addressing this gap requires a systematic approach that involves a capable base model, high-quality reasoning-oriented audi... | 2025.emnlp-main.27 | 10.18653/v1/2025.emnlp-main.27 | null | 2506.12935 | title_snapshot | [
-0.02569638006389141,
-0.020629938691854477,
-0.012238454073667526,
0.05491746962070465,
0.05215988680720329,
0.020120589062571526,
0.030588245019316673,
-0.006007074378430843,
-0.04150119051337242,
-0.015174904838204384,
-0.02125970460474491,
0.04737819731235504,
-0.06625110656023026,
-0.... |
Seeing More, Saying More: Lightweight Language Experts are Dynamic Video Token Compressors | https://aclanthology.org/2025.emnlp-main.28/ | [
"Xiangchen Wang",
"Jinrui Zhang",
"Teng Wang",
"Haigang Zhang",
"Feng Zheng"
] | Recent advancements in large video-language models have revolutionized video understanding tasks. However, their efficiency is significantly constrained by processing high volumes of visual tokens. Existing token compression strategies apply a fixed compression ratio, ignoring the variability in semantic density among ... | 2025.emnlp-main.28 | 10.18653/v1/2025.emnlp-main.28 | null | 2509.00969 | title_snapshot | [
0.008321802131831646,
-0.01692044921219349,
-0.00355990557000041,
0.05455000326037407,
0.024384653195738792,
0.0356583334505558,
-0.005311987828463316,
0.025861795991659164,
-0.04931202903389931,
-0.031132860109210014,
-0.048542555421590805,
0.009077313356101513,
-0.03308505937457085,
0.00... |
RoT: Enhancing Table Reasoning with Iterative Row-Wise Traversals | https://aclanthology.org/2025.emnlp-main.29/ | [
"Xuanliang Zhang",
"Dingzirui Wang",
"Keyan Xu",
"Qingfu Zhu",
"Wanxiang Che"
] | The table reasoning task, crucial for efficient data acquisition, aims to answer questions based on the given table. Recently, reasoning large language models (RLLMs) with Long Chain-of-Thought (Long CoT) significantly enhance reasoning capabilities, leading to brilliant performance on table reasoning. However, Long Co... | 2025.emnlp-main.29 | 10.18653/v1/2025.emnlp-main.29 | null | 2505.15110 | title_snapshot | [
-0.03164839372038841,
-0.005238095764070749,
0.013277669437229633,
0.0361250601708889,
0.03875107690691948,
0.010229459032416344,
0.016360078006982803,
-0.003446421353146434,
-0.04922555759549141,
-0.0004059688944835216,
-0.018124757334589958,
0.03200170025229454,
-0.0495353639125824,
0.01... |
T-MAD: Target-driven Multimodal Alignment for Stance Detection | https://aclanthology.org/2025.emnlp-main.30/ | [
"ZhaoDan Zhang",
"Jin Zhang",
"Xueqi Cheng",
"Hui Xu"
] | Multimodal Stance Detection (MSD) aims to determine a user’s stance - support, oppose, or neutral - toward a target by analyzing multimodal content such as texts and images from social media. Existing MSD methods struggle with generalizing to unseen targets and handling modality inconsistencies. To address these challe... | 2025.emnlp-main.30 | 10.18653/v1/2025.emnlp-main.30 | null | null | null | [
0.04782884940505028,
-0.012294656597077847,
0.0046528601087629795,
0.02726777456700802,
-0.0001625890436116606,
0.04462761804461479,
0.04510428011417389,
0.01643977127969265,
-0.04477223381400108,
-0.0031679989770054817,
-0.020939651876688004,
0.038734156638383865,
-0.06892269849777222,
-0... |
Emotion Transfer with Enhanced Prototype for Unseen Emotion Recognition in Conversation | https://aclanthology.org/2025.emnlp-main.31/ | [
"Kun Peng",
"Cong Cao",
"Hao Peng",
"Guanlin Wu",
"Zhifeng Hao",
"Lei Jiang",
"Yanbing Liu",
"Philip S. Yu"
] | Current Emotion Recognition in Conversation (ERC) research follows a closed-domain assumption. However, there is no clear consensus on emotion classification in psychology, which presents a challenge for models when it comes to recognizing previously unseen emotions in real-world applications. To bridge this gap, we in... | 2025.emnlp-main.31 | 10.18653/v1/2025.emnlp-main.31 | null | 2508.19533 | title_snapshot | [
-0.011128983460366726,
-0.03179095312952995,
-0.002637302502989769,
0.0512784868478775,
0.028423583135008812,
0.04304180666804314,
0.04854881018400192,
0.003343034302815795,
0.01956564374268055,
-0.038427192717790604,
-0.013432852923870087,
0.027701836079359055,
-0.04516686871647835,
-0.01... |
PBI-Attack: Prior-Guided Bimodal Interactive Black-Box Jailbreak Attack for Toxicity Maximization | https://aclanthology.org/2025.emnlp-main.32/ | [
"Ruoxi Cheng",
"Yizhong Ding",
"Shuirong Cao",
"Ranjie Duan",
"Xiaoshuang Jia",
"Shaowei Yuan",
"Simeng Qin",
"Zhiqiang Wang",
"Xiaojun Jia"
] | Understanding the vulnerabilities of Large Vision Language Models (LVLMs) to jailbreak attacks is essential for their responsible real-world deployment. Most previous work requires access to model gradients, or is based on human knowledge (prompt engineering) to complete jailbreak, and they hardly consider the interact... | 2025.emnlp-main.32 | 10.18653/v1/2025.emnlp-main.32 | null | 2412.05892 | title_snapshot | [
-0.004445957951247692,
-0.026813441887497902,
-0.0446120984852314,
0.021063342690467834,
0.041052497923374176,
0.02372170425951481,
0.0374995693564415,
-0.019327767193317413,
-0.012079235166311264,
-0.014690748415887356,
-0.009289865382015705,
-0.01253801304847002,
-0.0899861603975296,
-0.... |
Training a Utility-based Retriever Through Shared Context Attribution for Retrieval-Augmented Language Models | https://aclanthology.org/2025.emnlp-main.33/ | [
"Yilong Xu",
"Jinhua Gao",
"Xiaoming Yu",
"Yuanhai Xue",
"Baolong Bi",
"Huawei Shen",
"Xueqi Cheng"
] | Retrieval-Augmented Language Models boost task performance, owing to the retriever that provides external knowledge. Although crucial, the retriever primarily focuses on semantics relevance, which may not always be effective for generation. Thus, utility-based retrieval has emerged as a promising topic, prioritizing pa... | 2025.emnlp-main.33 | 10.18653/v1/2025.emnlp-main.33 | null | 2504.00573 | title_snapshot | [
-0.011157230474054813,
-0.03456457331776619,
-0.012868929654359818,
0.04850418493151665,
0.04303375259041786,
0.012702759355306625,
0.026187125593423843,
0.03311975672841072,
-0.010029603727161884,
-0.010732164606451988,
-0.06367694586515427,
0.060095347464084625,
-0.04964267835021019,
-0.... |
SportReason: Evaluating Retrieval-Augmented Reasoning across Tables and Text for Sports Question Answering | https://aclanthology.org/2025.emnlp-main.34/ | [
"Kaiyue Feng",
"Siyue Zhang",
"Bingsen Chen",
"Yilun Zhao",
"Chen Zhao"
] | We present SportReason, a benchmark for retrieval-augmented reasoning on numerical sports questions. Unlike existing benchmarks limited to one or two evidence units, SportReason requires combining and reasoning across free-text, structured tables, and semi-structured infoboxes. We provide 3,000 human-verified QA pairs ... | 2025.emnlp-main.34 | 10.18653/v1/2025.emnlp-main.34 | null | null | null | [
-0.017953261733055115,
-0.024389024823904037,
-0.010631733573973179,
0.060240842401981354,
0.03965812176465988,
-0.05021820217370987,
0.032498471438884735,
0.021069107577204704,
-0.020207710564136505,
-0.002045585308223963,
-0.011927979066967964,
0.056021030992269516,
-0.07297159731388092,
... |
MAC-Tuning: LLM Multi-Compositional Problem Reasoning with Enhanced Knowledge Boundary Awareness | https://aclanthology.org/2025.emnlp-main.35/ | [
"Junsheng Huang",
"Zhitao He",
"Yuchen Huang",
"Sandeep Polisetty",
"Qingyun Wang",
"Yi R. Fung"
] | With the widespread application of large language models (LLMs), the issue of generating non-existing facts, known as hallucination, has garnered increasing attention. Previous research in enhancing LLM confidence estimation mainly focuses on the single problem setting. However, LLM awareness of its internal parameteri... | 2025.emnlp-main.35 | 10.18653/v1/2025.emnlp-main.35 | null | 2504.21773 | title_snapshot | [
-0.004879178944975138,
-0.005714649800211191,
0.009518236853182316,
0.025124046951532364,
0.06958093494176865,
-0.0060377176851034164,
0.04895304888486862,
0.009216812439262867,
-0.03272577375173569,
0.010704714804887772,
-0.035902611911296844,
0.04696352034807205,
-0.05351501330733299,
-0... |
CODI: Compressing Chain-of-Thought into Continuous Space via Self-Distillation | https://aclanthology.org/2025.emnlp-main.36/ | [
"Zhenyi Shen",
"Hanqi Yan",
"Linhai Zhang",
"Zhanghao Hu",
"Yali Du",
"Yulan He"
] | Chain-of-Thought (CoT) reasoning enhances Large Language Models (LLMs) by encouraging step-by-step reasoning in natural language. However, leveraging a latent continuous space for reasoning may offer benefits in terms of both efficiency and robustness. Prior implicit CoT methods attempt to bypass language completely by... | 2025.emnlp-main.36 | 10.18653/v1/2025.emnlp-main.36 | null | 2502.21074 | title_snapshot | [
-0.011450586840510368,
-0.010563630610704422,
-0.023611394688487053,
0.05118246003985405,
0.058497343212366104,
0.022751858457922935,
0.03287527337670326,
0.007303073536604643,
0.012728463858366013,
-0.017896616831421852,
-0.03685998171567917,
0.029883727431297302,
-0.08051452785730362,
0.... |
PAFT: Prompt-Agnostic Fine-Tuning | https://aclanthology.org/2025.emnlp-main.37/ | [
"Chenxing Wei",
"Mingwen Ou",
"Ying He",
"Yao Shu",
"Fei Yu"
] | Fine-tuning large language models (LLMs) often causes overfitting to specific prompt wording, where minor phrasing variations drastically reduce performance. To address this, we propose Prompt-Agnostic Fine-Tuning (PAFT), a method that enhances robustness through dynamic prompt variation during training. PAFT first gen... | 2025.emnlp-main.37 | 10.18653/v1/2025.emnlp-main.37 | SAC Highlight Award | 2502.12859 | title_snapshot | [
-0.013411164283752441,
-0.05564074590802193,
0.011979160830378532,
0.049967434257268906,
0.04148038104176521,
0.020115887746214867,
0.014902383089065552,
0.005585161969065666,
-0.025969384238123894,
-0.008668134920299053,
-0.020525436848402023,
0.029537584632635117,
-0.06053527072072029,
-... |
Theorem-Validated Reverse Chain-of-Thought Problem Generation for Geometric Reasoning | https://aclanthology.org/2025.emnlp-main.38/ | [
"Deng Linger",
"Linghao Zhu",
"Yuliang Liu",
"Yu Wang",
"Qunyi Xie",
"Jingjing Wu",
"Gang Zhang",
"Yingying Zhu",
"Xiang Bai"
] | Large Multimodal Models (LMMs) face limitations in geometric reasoning due to insufficient Chain of Thought (CoT) image-text training data. While existing approaches leverage template-based or LLM-assisted methods for geometric CoT data creation, they often face challenges in achieving both diversity and precision. To ... | 2025.emnlp-main.38 | 10.18653/v1/2025.emnlp-main.38 | null | 2410.17885 | title_snapshot | [
0.008889642544090748,
0.0016330506186932325,
0.008751965127885342,
0.03933073952794075,
0.034282904118299484,
0.009845803491771221,
0.013727623037993908,
0.032627176493406296,
-0.0015104239573702216,
-0.02838560938835144,
-0.02177058532834053,
0.022792916744947433,
-0.054222673177719116,
0... |
TACO: Enhancing Multimodal In-context Learning via Task Mapping-Guided Sequence Configuration | https://aclanthology.org/2025.emnlp-main.39/ | [
"Yanshu Li",
"Jianjiang Yang",
"Tian Yun",
"Pinyuan Feng",
"Jinfa Huang",
"Ruixiang Tang"
] | Multimodal in-context learning (ICL) has emerged as a key mechanism for harnessing the capabilities of large vision–language models (LVLMs). However, its effectiveness remains highly sensitive to the quality of input ICL sequences, particularly for tasks involving complex reasoning or open-ended generation. A major lim... | 2025.emnlp-main.39 | 10.18653/v1/2025.emnlp-main.39 | null | 2505.17098 | title_snapshot | [
0.015101538971066475,
0.018142975866794586,
-0.0159357450902462,
0.0402233712375164,
0.02612706646323204,
0.0013455492444336414,
0.03357964754104614,
0.06772679090499878,
-0.0037534066941589117,
0.0011459194356575608,
-0.03356359899044037,
0.02633930929005146,
-0.044211916625499725,
-0.041... |
Towards Controllable Speech Synthesis in the Era of Large Language Models: A Systematic Survey | https://aclanthology.org/2025.emnlp-main.40/ | [
"Tianxin Xie",
"Yan Rong",
"Pengfei Zhang",
"Wenwu Wang",
"Li Liu"
] | Text-to-speech (TTS) has advanced from generating natural-sounding speech to enabling fine-grained control over attributes like emotion, timbre, and style. Driven by rising industrial demand and breakthroughs in deep learning, e.g., diffusion and large language models (LLMs), controllable TTS has become a rapidly growi... | 2025.emnlp-main.40 | 10.18653/v1/2025.emnlp-main.40 | null | 2412.06602 | title_snapshot | [
-0.043263740837574005,
-0.047142669558525085,
-0.042173177003860474,
0.02095358818769455,
0.03303326293826103,
0.026867061853408813,
0.0551341213285923,
0.030542241409420967,
-0.021862614899873734,
-0.04908081516623497,
-0.04678165167570114,
0.01669578067958355,
-0.039361558854579926,
0.00... |
Automating Steering for Safe Multimodal Large Language Models | https://aclanthology.org/2025.emnlp-main.41/ | [
"Lyucheng Wu",
"Mengru Wang",
"Ziwen Xu",
"Tri Cao",
"Nay Oo",
"Bryan Hooi",
"Shumin Deng"
] | Recent progress in Multimodal Large Language Models (MLLMs) has unlocked powerful cross-modal reasoning abilities, but also raised new safety concerns, particularly when faced with adversarial multimodal inputs. To improve the safety of MLLMs during inference, we introduce a modular and adaptive inference-time interven... | 2025.emnlp-main.41 | 10.18653/v1/2025.emnlp-main.41 | null | 2507.13255 | title_snapshot | [
-0.003210550406947732,
-0.01937151327729225,
0.0010172287002205849,
0.02510797046124935,
0.023477640002965927,
0.0035406439565122128,
0.048781536519527435,
0.00013957088231109083,
-0.02003883384168148,
-0.008344975300133228,
-0.03953291475772858,
0.04737066105008125,
-0.05396631360054016,
... |
EMNLP: Educator-role Moral and Normative Large Language Models Profiling | https://aclanthology.org/2025.emnlp-main.42/ | [
"Yilin Jiang",
"Mingzi Zhang",
"Sheng Jin",
"Zengyi Yu",
"Xiangjie Kong",
"Binghao Tu"
] | Simulating Professions (SP) enables Large Language Models (LLMs) to emulate professional roles. However, comprehensive psychological and ethical evaluation in these contexts remains lacking. This paper introduces EMNLP, an Educator-role Moral and Normative LLMs Profiling framework for personality profiling, moral devel... | 2025.emnlp-main.42 | 10.18653/v1/2025.emnlp-main.42 | null | 2508.15250 | title_snapshot | [
-0.021567801013588905,
-0.01619168184697628,
-0.020709335803985596,
0.029272807762026787,
0.039246417582035065,
0.00801310408860445,
0.029334042221307755,
0.0016983459936454892,
-0.047105953097343445,
-0.01640784554183483,
-0.04524948075413704,
0.02710835076868534,
-0.050990164279937744,
-... |
TracSum: A New Benchmark for Aspect-Based Summarization with Sentence-Level Traceability in Medical Domain | https://aclanthology.org/2025.emnlp-main.43/ | [
"Bohao Chu",
"Meijie Li",
"Sameh Frihat",
"Chengyu Gu",
"Georg Lodde",
"Elisabeth Livingstone",
"Norbert Fuhr"
] | While document summarization with LLMs has enhanced access to textual information, concerns about the factual accuracy of these summaries persist (e.g., hallucination), especially in the medical domain. Tracing source evidence from which summaries are derived enables users to assess their accuracy, thereby alleviating ... | 2025.emnlp-main.43 | 10.18653/v1/2025.emnlp-main.43 | null | 2508.13798 | title_snapshot | [
-0.001968247117474675,
-0.025636427104473114,
0.010019209235906601,
-0.004621436819434166,
0.03343484178185463,
-0.03976046293973923,
0.05006541311740875,
0.03846852481365204,
-0.03290661424398422,
-0.031715329736471176,
-0.02728961408138275,
0.017909465357661247,
-0.04003608226776123,
-0.... |
Context Reasoner: Incentivizing Reasoning Capability for Contextualized Privacy and Safety Compliance via Reinforcement Learning | https://aclanthology.org/2025.emnlp-main.44/ | [
"Wenbin Hu",
"Haoran Li",
"Huihao Jing",
"Qi Hu",
"Ziqian Zeng",
"Sirui Han",
"Xu Heli",
"Tianshu Chu",
"Peizhao Hu",
"Yangqiu Song"
] | While Large Language Models (LLMs) exhibit remarkable capabilities, they also introduce significant safety and privacy risks. Current mitigation strategies often fail to preserve contextual reasoning capabilities in risky scenarios. Instead, they rely heavily on sensitive pattern matching to protect LLMs, which limits ... | 2025.emnlp-main.44 | 10.18653/v1/2025.emnlp-main.44 | null | 2505.14585 | title_snapshot | [
-0.004121498204767704,
-0.009597440250217915,
-0.00882488302886486,
0.046401552855968475,
0.04250887781381607,
-0.01417126040905714,
0.03093692846596241,
-0.02459854818880558,
-0.03224349766969681,
0.010075253434479237,
-0.02414822392165661,
0.042631588876247406,
-0.06177166476845741,
-0.0... |
Towards General-Domain Word Sense Disambiguation: Distilling Large Language Model into Compact Disambiguator | https://aclanthology.org/2025.emnlp-main.45/ | [
"Liqiang Ming",
"Sheng-hua Zhong",
"Yuncong Li"
] | Word Sense Disambiguation (WSD) aims to determine the correct meaning of a word in context from a predefined inventory, and remains a fundamental challenge in natural language understanding. Existing methods rely heavily on manually annotated data, which limits coverage and generalization. In this work, we propose a sc... | 2025.emnlp-main.45 | 10.18653/v1/2025.emnlp-main.45 | null | null | null | [
0.0007976629422046244,
-0.01676865853369236,
0.0022710685152560472,
0.06855116784572601,
0.06270629167556763,
-0.0016389259835705161,
0.03729226067662239,
0.007573658134788275,
0.004564688075333834,
-0.005005992017686367,
-0.02500118501484394,
0.017929960042238235,
-0.0636512041091919,
0.0... |
SLoW: Select Low-frequency Words! Automatic Dictionary Selection for Translation on Large Language Models | https://aclanthology.org/2025.emnlp-main.46/ | [
"Hongyuan Lu",
"Zixuan Li",
"Zefan Zhang",
"Wai Lam"
] | There are more than 7,000 languages around the world, and current Large Language Models (LLMs) only support hundreds of languages. Dictionary-based prompting methods can enhance translation on them, but most methods use all the available dictionaries, which could be expensive. Instead, it will be flexible to have a tra... | 2025.emnlp-main.46 | 10.18653/v1/2025.emnlp-main.46 | null | 2507.18902 | title_snapshot | [
-0.03479083627462387,
-0.0399448499083519,
0.012914229184389114,
0.014952372759580612,
0.02167356386780739,
0.0378224179148674,
0.02172108180820942,
0.03162096068263054,
-0.016173597425222397,
-0.026757698506116867,
-0.037292663007974625,
0.026841076090931892,
-0.04680190607905388,
0.00128... |
Parallel Continuous Chain-of-Thought with Jacobi Iteration | https://aclanthology.org/2025.emnlp-main.47/ | [
"Haoyi Wu",
"Zhihao Teng",
"Kewei Tu"
] | Continuous chain-of-thought has been shown to be effective in saving reasoning tokens for large language models. By reasoning with continuous latent thought tokens, continuous CoT is able to perform implicit reasoning in a compact manner. However, the sequential dependencies between latent thought tokens spoil parallel... | 2025.emnlp-main.47 | 10.18653/v1/2025.emnlp-main.47 | null | 2506.18582 | title_snapshot | [
-0.02929929830133915,
-0.0446709468960762,
-0.027478156611323357,
0.0630561038851738,
0.03516422584652901,
0.04184192419052124,
0.04203128442168236,
0.030443653464317322,
-0.012680953368544579,
-0.017606908455491066,
-0.019848475232720375,
0.02066516876220703,
-0.05986742675304413,
0.00161... |
EQA-RM: A Generative Embodied Reward Model with Test-time Scaling | https://aclanthology.org/2025.emnlp-main.48/ | [
"Yuhang Chen",
"Zhen Tan",
"Tianlong Chen"
] | Reward Models (RMs), vital for large model alignment, are underexplored for complex embodied tasks like Embodied Question Answering (EQA) where nuanced evaluation of agents’ spatial, temporal, and logical understanding is critical yet not considerred by generic approaches. We introduce EQA-RM, a novel generative multim... | 2025.emnlp-main.48 | 10.18653/v1/2025.emnlp-main.48 | null | 2506.10389 | title_snapshot | [
0.010947512462735176,
-0.01404043473303318,
0.014425905421376228,
0.06478992104530334,
0.03249974548816681,
0.027129579335451126,
-0.0023612624499946833,
-0.011553486809134483,
-0.03601603955030441,
-0.0029704091139137745,
-0.015916423872113228,
0.047716714441776276,
-0.0566692054271698,
-... |
Refusal-Aware Red Teaming: Exposing Inconsistency in Safety Evaluations | https://aclanthology.org/2025.emnlp-main.49/ | [
"Yongkang Chen",
"Xiaohu Du",
"Xiaotian Zou",
"Chongyang Zhao",
"Huan Deng",
"Hu Li",
"Xiaohui Kuang"
] | The responsible deployment of Large Language Models (LLMs) necessitates rigorous safety evaluations. However, a critical challenge arises from inconsistencies between an LLM’s internal refusal decisions and external safety assessments, hindering effective validation. This paper introduces the concept of the ‘refusal ga... | 2025.emnlp-main.49 | 10.18653/v1/2025.emnlp-main.49 | null | null | null | [
-0.005543231498450041,
-0.026698125526309013,
-0.058988090604543686,
0.07178764045238495,
0.053993407636880875,
0.00021267181728035212,
0.045281797647476196,
-0.016333239153027534,
-0.041033271700143814,
-0.015261801891028881,
-0.034430019557476044,
0.0704273208975792,
-0.06532294303178787,
... |
OmniThink: Expanding Knowledge Boundaries in Machine Writing through Thinking | https://aclanthology.org/2025.emnlp-main.50/ | [
"Zekun Xi",
"Wenbiao Yin",
"Jizhan Fang",
"Jialong Wu",
"Runnan Fang",
"Yong Jiang",
"Pengjun Xie",
"Fei Huang",
"Huajun Chen",
"Ningyu Zhang"
] | Machine writing with large language models often relies on retrieval-augmented generation. However, these approaches remain confined within the boundaries of the model’s predefined scope, limiting the generation of content with rich information. Specifically, vanilla-retrieved information tends to lack depth, novelty, ... | 2025.emnlp-main.50 | 10.18653/v1/2025.emnlp-main.50 | null | 2501.09751 | title_snapshot | [
-0.019687918946146965,
-0.012861311435699463,
-0.012291708961129189,
0.04368419945240021,
0.05488090589642525,
-0.0034229084849357605,
0.01927398517727852,
0.018671143800020218,
-0.04128587245941162,
-0.013889295049011707,
-0.027670325711369514,
0.011116642504930496,
-0.03258463740348816,
... |
LinkAlign: Scalable Schema Linking for Real-World Large-Scale Multi-Database Text-to-SQL | https://aclanthology.org/2025.emnlp-main.51/ | [
"Yihan Wang",
"Peiyu Liu",
"Xin Yang"
] | Schema linking is a critical bottleneck in applying existing Text-to-SQL models to real-world, large-scale, multi-database environments. Through error analysis, we identify two major challenges in schema linking: (1) Database Retrieval: accurately selecting the target database from a large schema pool, while effectivel... | 2025.emnlp-main.51 | 10.18653/v1/2025.emnlp-main.51 | null | 2503.18596 | title_snapshot | [
-0.00040516065200790763,
-0.010561278089880943,
-0.018188657239079475,
0.04240207374095917,
0.05256867781281471,
0.012305091135203838,
0.003147912211716175,
0.0017958828248083591,
-0.009210440330207348,
-0.0609348826110363,
-0.01459023728966713,
0.006000559311360121,
-0.08723276853561401,
... |
On Relation-Specific Neurons in Large Language Models | https://aclanthology.org/2025.emnlp-main.52/ | [
"Yihong Liu",
"Runsheng Chen",
"Lea Hirlimann",
"Ahmad Dawar Hakimi",
"Mingyang Wang",
"Amir Hossein Kargaran",
"Sascha Rothe",
"François Yvon",
"Hinrich Schuetze"
] | In large language models (LLMs), certain neurons can store distinct pieces of knowledge learned during pretraining. While factual knowledge typically appears as a combination of relations and entities, it remains unclear whether some neurons focus on a relation itself – independent of any entity. We hypothesize such ne... | 2025.emnlp-main.52 | 10.18653/v1/2025.emnlp-main.52 | null | 2502.17355 | title_snapshot | [
-0.04953441768884659,
0.008949634619057178,
0.02344922348856926,
0.027500564232468605,
0.033682700246572495,
-0.0015408247709274292,
0.03727075830101967,
0.03421929478645325,
-0.04255365952849388,
0.02449895441532135,
-0.038367290049791336,
0.05179429426789284,
-0.06527729332447052,
0.0124... |
IPIGuard: A Novel Tool Dependency Graph-Based Defense Against Indirect Prompt Injection in LLM Agents | https://aclanthology.org/2025.emnlp-main.53/ | [
"Hengyu An",
"Jinghuai Zhang",
"Tianyu Du",
"Chunyi Zhou",
"Qingming Li",
"Tao Lin",
"Shouling Ji"
] | Large language model (LLM) agents are widely deployed in real-world applications, where they leverage tools to retrieve and manipulate external data for complex tasks. However, when interacting with untrusted data sources (e.g., fetching information from public websites), tool responses may contain injected instruction... | 2025.emnlp-main.53 | 10.18653/v1/2025.emnlp-main.53 | null | 2508.15310 | title_snapshot | [
-0.027317317202687263,
-0.03491963818669319,
-0.02032054401934147,
0.04032251983880997,
0.0258981641381979,
-0.013215209357440472,
0.04536598175764084,
-0.007185166701674461,
-0.020462699234485626,
-0.01364457793533802,
-0.007012833375483751,
0.03195157274603844,
-0.06483875960111618,
-0.0... |
ProtoVQA: An Adaptable Prototypical Framework for Explainable Fine-Grained Visual Question Answering | https://aclanthology.org/2025.emnlp-main.54/ | [
"Xingjian Diao",
"Weiyi Wu",
"Keyi Kong",
"Peijun Qing",
"Xinwen Xu",
"Ming Cheng",
"Soroush Vosoughi",
"Jiang Gui"
] | Visual Question Answering (VQA) is increasingly used in diverse applications ranging from general visual reasoning to safety-critical domains such as medical imaging and autonomous systems, where models must provide not only accurate answers but also explanations that humans can easily understand and verify. Prototype-... | 2025.emnlp-main.54 | 10.18653/v1/2025.emnlp-main.54 | null | 2509.16680 | title_snapshot | [
0.014140673913061619,
-0.01576988399028778,
-0.01259029284119606,
0.06754729151725769,
0.025041596964001656,
0.03919012099504471,
0.02023189328610897,
-0.019341783598065376,
-0.010769438929855824,
-0.023865636438131332,
-0.031175557523965836,
0.037310194224119186,
-0.05444520339369774,
-0.... |
SEA: Supervised Embedding Alignment for Token-Level Visual-Textual Integration in MLLMs | https://aclanthology.org/2025.emnlp-main.55/ | [
"Yuanyang Yin",
"Yaqi Zhao",
"Yajie Zhang",
"Yuanxing Zhang",
"Ke Lin",
"Jiahao Wang",
"Xin Tao",
"Pengfei Wan",
"Wentao Zhang",
"Feng Zhao"
] | Multimodal Large Language Models (MLLMs) have demonstrated remarkable capabilities by integrating visual and textual inputs, yet modality alignment remains one of the most challenging aspects. Current MLLMs typically rely on simple adapter architectures and pretraining approaches to bridge vision encoders with large la... | 2025.emnlp-main.55 | 10.18653/v1/2025.emnlp-main.55 | null | 2408.11813 | title_snapshot | [
0.009454872459173203,
-0.025577526539564133,
-0.006608910858631134,
0.01054114568978548,
0.022213242948055267,
0.01238294318318367,
0.02664436399936676,
0.03141245245933533,
-0.04127880185842514,
0.007841458544135094,
-0.0194763895124197,
0.01510689128190279,
-0.07212652266025543,
-0.01024... |
Molecular String Representation Preferences in Pretrained LLMs: A Comparative Study in Zero- & Few-Shot Molecular Property Prediction | https://aclanthology.org/2025.emnlp-main.56/ | [
"George Arthur Baker",
"Mario Sanz-Guerrero",
"Katharina von der Wense"
] | Large Language Models (LLMs) have demonstrated capabilities for natural language formulations of molecular property prediction tasks, but little is known about how performance depends on the representation of input molecules to the model; the status quo approach is to use SMILES strings, although alternative chemical n... | 2025.emnlp-main.56 | 10.18653/v1/2025.emnlp-main.56 | null | null | null | [
-0.008889253251254559,
0.010492485947906971,
0.0025463979691267014,
0.05497891083359718,
0.059161752462387085,
0.0038826081436127424,
0.01352181751281023,
0.027294037863612175,
0.011626158840954304,
-0.008875127881765366,
-0.02062898688018322,
0.05367688834667206,
-0.0721476748585701,
0.02... |
Weight-Aware Activation Sparsity with Constrained Bayesian Optimization Scheduling for Large Language Models | https://aclanthology.org/2025.emnlp-main.57/ | [
"Ming Wang",
"Miao Zhang",
"Xuebo Liu",
"Liqiang Nie"
] | Activation sparsity provides a dynamic, input-dependent alternative to weight pruning for accelerating inference in large language models (LLMs), effectively reducing unnecessary computations and memory accesses during the forward pass. Despite its promise, existing activation sparsification methods suffer from two maj... | 2025.emnlp-main.57 | 10.18653/v1/2025.emnlp-main.57 | null | null | null | [
-0.032827164977788925,
0.0011553774820640683,
0.008804419077932835,
0.03150414302945137,
0.009879602119326591,
0.047086942940950394,
0.02764379046857357,
0.006986423395574093,
-0.022873887792229652,
-0.010547797195613384,
0.0022899846080690622,
0.010172623209655285,
-0.03934011980891228,
0... |
DatawiseAgent: A Notebook-Centric LLM Agent Framework for Adaptive and Robust Data Science Automation | https://aclanthology.org/2025.emnlp-main.58/ | [
"Ziming You",
"Yumiao Zhang",
"Dexuan Xu",
"Yiwei Lou",
"Yandong Yan",
"Wei Wang",
"Huamin Zhang",
"Yu Huang"
] | Existing large language model (LLM) agents for automating data science show promise, but they remain constrained by narrow task scopes, limited generalization across tasks and models, and over-reliance on state-of-the-art (SOTA) LLMs. We introduce DatawiseAgent, a notebook-centric LLM agent framework for adaptive and r... | 2025.emnlp-main.58 | 10.18653/v1/2025.emnlp-main.58 | null | 2503.07044 | title_snapshot | [
-0.046754054725170135,
-0.055940162390470505,
-0.020397771149873734,
0.0312926359474659,
0.06639131903648376,
0.014184057712554932,
0.013225012458860874,
0.006565958261489868,
-0.012155664153397083,
-0.011318853124976158,
-0.029069969430565834,
0.016290366649627686,
-0.06276798248291016,
0... |
VC4VG: Optimizing Video Captions for Text-to-Video Generation | https://aclanthology.org/2025.emnlp-main.59/ | [
"Yang Du",
"Zhuoran Lin",
"Kaiqiang Song",
"Biao Wang",
"Zhicheng Zheng",
"Tiezheng Ge",
"Bo Zheng",
"Qin Jin"
] | Recent advances in text-to-video (T2V) generation highlight the critical role of high-quality video-text pairs in training models capable of producing coherent and instruction-aligned videos. However, strategies for optimizing video captions specifically for T2V training remain underexplored. In this paper, we introduc... | 2025.emnlp-main.59 | 10.18653/v1/2025.emnlp-main.59 | null | 2510.24134 | title_snapshot | [
0.02508457936346531,
-0.021841932088136673,
0.028570592403411865,
0.057515762746334076,
0.02214352786540985,
0.005238377954810858,
0.03243131563067436,
0.0031180251389741898,
-0.027304749935865402,
-0.029369769617915154,
-0.04343376308679581,
0.031553350389003754,
-0.0528155080974102,
0.01... |
LaMP-QA: A Benchmark for Personalized Long-form Question Answering | https://aclanthology.org/2025.emnlp-main.60/ | [
"Alireza Salemi",
"Hamed Zamani"
] | Personalization is essential for question answering systems that are user-centric. Despite its importance, personalization in answer generation has been relatively underexplored. This is mainly due to lack of resources for training and evaluating personalized question answering systems. We address this gap by introduci... | 2025.emnlp-main.60 | 10.18653/v1/2025.emnlp-main.60 | null | 2506.00137 | title_snapshot | [
0.031241189688444138,
-0.013487390242516994,
0.00861724466085434,
0.03781644254922867,
0.04707859084010124,
0.013429545797407627,
0.010747641324996948,
0.01247523445636034,
-0.025158310309052467,
0.002747531980276108,
-0.03818117454648018,
0.03709973022341728,
-0.055683430284261703,
-0.022... |
The LLM Already Knows: Estimating LLM-Perceived Question Difficulty via Hidden Representations | https://aclanthology.org/2025.emnlp-main.61/ | [
"Yubo Zhu",
"Dongrui Liu",
"Zecheng Lin",
"Wei Tong",
"Sheng Zhong",
"Jing Shao"
] | Estimating the difficulty of input questions as perceived by large language models (LLMs) is essential for accurate performance evaluation and adaptive inference. Existing methods typically rely on repeated response sampling, auxiliary models, or fine-tuning the target model itself, which may incur substantial computat... | 2025.emnlp-main.61 | 10.18653/v1/2025.emnlp-main.61 | null | 2509.12886 | title_snapshot | [
0.005189497023820877,
-0.028324054554104805,
-0.011967694386839867,
0.03575477749109268,
0.05475318059325218,
0.03355911746621132,
0.023412181064486504,
0.009355703368782997,
-0.02064257487654686,
0.030181651934981346,
-0.04083505645394325,
0.042959064245224,
-0.028917893767356873,
-0.0137... |
MCIP: Protecting MCP Safety via Model Contextual Integrity Protocol | https://aclanthology.org/2025.emnlp-main.62/ | [
"Huihao Jing",
"Haoran Li",
"Wenbin Hu",
"Qi Hu",
"Xu Heli",
"Tianshu Chu",
"Peizhao Hu",
"Yangqiu Song"
] | As Model Context Protocol (MCP) introduces an easy-to-use ecosystem for users and developers, it also brings underexplored safety risks. Its decentralized architecture, which separates clients and servers, poses unique challenges for systematic safety analysis. This paper proposes a novel framework to enhance MCP safet... | 2025.emnlp-main.62 | 10.18653/v1/2025.emnlp-main.62 | null | 2505.14590 | title_snapshot | [
-0.016049401834607124,
-0.016397710889577866,
-0.03741759434342384,
0.03808792307972908,
0.05942569300532341,
-0.020100025460124016,
0.023348843678832054,
-0.02372671477496624,
-0.036828599870204926,
-0.046114444732666016,
-0.024901989847421646,
0.0016670546028763056,
-0.054476119577884674,
... |
SAKI-RAG: Mitigating Context Fragmentation in Long-Document RAG via Sentence-level Attention Knowledge Integration | https://aclanthology.org/2025.emnlp-main.63/ | [
"Wenyu Tao",
"Xiaofen Xing",
"Zeliang Li",
"Xiangmin Xu"
] | Traditional Retrieval-Augmented Generation (RAG) frameworks often segment documents into larger chunks to preserve contextual coherence, inadvertently introducing redundant noise. Recent advanced RAG frameworks have shifted toward finer-grained chunking to improve precision. However, in long-document scenarios, such ch... | 2025.emnlp-main.63 | 10.18653/v1/2025.emnlp-main.63 | null | null | null | [
-0.01293247751891613,
-0.05735958740115166,
-0.03304735943675041,
0.04603421315550804,
-0.008637771010398865,
-0.03664473444223404,
0.011044038459658623,
0.0043559023179113865,
-0.03395099937915802,
-0.004101315978914499,
-0.05513728782534599,
0.014164578169584274,
-0.04810313507914543,
0.... |
Skeletons Matter: Dynamic Data Augmentation for Text-to-Query | https://aclanthology.org/2025.emnlp-main.64/ | [
"Yuchen Ji",
"Bo Xu",
"Jie Shi",
"Jiaqing Liang",
"Deqing Yang",
"Yu Mao",
"Hai Chen",
"Yanghua Xiao"
] | The task of translating natural language questions into query languages has long been a central focus in semantic parsing. Recent advancements in Large Language Models (LLMs) have significantly accelerated progress in this field. However, existing studies typically focus on a single query language, resulting in methods... | 2025.emnlp-main.64 | 10.18653/v1/2025.emnlp-main.64 | null | 2511.18934 | title_snapshot | [
-0.0075504896230995655,
-0.030828796327114105,
-0.0259429682046175,
0.061003658920526505,
0.01854860968887806,
0.008413244970142841,
0.02840721420943737,
0.02360714040696621,
-0.0237995907664299,
-0.03841723874211311,
-0.04126668721437454,
0.03702516853809357,
-0.052507754415273666,
-0.014... |
CondenseLM: LLMs-driven Text Dataset Condensation via Reward Matching | https://aclanthology.org/2025.emnlp-main.65/ | [
"Cheng Shen",
"Yew-Soon Ong",
"Joey Tianyi Zhou"
] | Dataset condensation has emerged as a promising technique to improve data efficiency under limited data budgets. However, when applied to the text level, existing methods struggle to compress more information into samples through optimization. Thus, these methods provide no obvious advantage over simpler coreset select... | 2025.emnlp-main.65 | 10.18653/v1/2025.emnlp-main.65 | null | null | null | [
0.023836052045226097,
-0.038723722100257874,
-0.004763389937579632,
0.05516146123409271,
0.06733953207731247,
0.008526192046701908,
0.003452550619840622,
0.011689702980220318,
-0.017646649852395058,
-0.010192363522946835,
-0.03395146504044533,
0.024606216698884964,
-0.05896087735891342,
-0... |
MovieCORE: COgnitive REasoning in Movies | https://aclanthology.org/2025.emnlp-main.66/ | [
"Gueter Josmy Faure",
"Min-Hung Chen",
"Jia-Fong Yeh",
"Ying Cheng",
"Hung-Ting Su",
"Yung-Hao Tang",
"Shang-Hong Lai",
"Winston H. Hsu"
] | This paper introduces MovieCORE, a novel video question answering (VQA) dataset designed to probe deeper cognitive understanding of movie content. Unlike existing datasets that focus on surface-level comprehension, MovieCORE emphasizes questions that engage System-2 thinking while remaining specific to the video materi... | 2025.emnlp-main.66 | 10.18653/v1/2025.emnlp-main.66 | null | 2508.19026 | title_snapshot | [
0.03134176880121231,
-0.006556149572134018,
0.008607217110693455,
0.09449130296707153,
0.04120096564292908,
0.003899357048794627,
0.016100946813821793,
0.0093832453712821,
-0.003016200615093112,
-0.009126236662268639,
-0.032903872430324554,
0.049734387546777725,
-0.057840317487716675,
-0.0... |
Think Wider, Detect Sharper: Reinforced Reference Coverage for Document-Level Self-Contradiction Detection | https://aclanthology.org/2025.emnlp-main.67/ | [
"Yuhao Chen",
"Yuanjie Lyu",
"Shuochen Liu",
"Chao Zhang",
"Junhui Lv",
"Tong Xu"
] | Detecting self-contradictions within documents is a challenging task for ensuring textual coherence and reliability. While large language models (LLMs) have advanced in many natural language understanding tasks, document-level self-contradiction detection (DSCD) remains insufficiently studied. Recent approaches leverag... | 2025.emnlp-main.67 | 10.18653/v1/2025.emnlp-main.67 | null | null | null | [
0.00011597774573601782,
-0.03734205290675163,
-0.028570447117090225,
0.05896679311990738,
0.042712289839982986,
-0.031192263588309288,
0.042615532875061035,
0.019340012222528458,
-0.021282313391566277,
-0.01862923800945282,
-0.02978035807609558,
0.0789620503783226,
-0.03656342253088951,
-0... |
DRISHTIKON: A Multimodal Multilingual Benchmark for Testing Language Models’ Understanding on Indian Culture | https://aclanthology.org/2025.emnlp-main.68/ | [
"Arijit Maji",
"Raghvendra Kumar",
"Akash Ghosh",
"Anushka",
"Nemil Shah",
"Abhilekh Borah",
"Vanshika Shah",
"Nishant Mishra",
"Sriparna Saha"
] | We introduce DRISHTIKON, a first-of-its-kind multimodal and multilingual benchmark centered exclusively on Indian culture, designed to evaluate the cultural understanding of generative AI systems. Unlike existing benchmarks with a generic or global scope, DRISHTIKON offers deep, fine-grained coverage across India’s div... | 2025.emnlp-main.68 | 10.18653/v1/2025.emnlp-main.68 | null | 2509.19274 | title_snapshot | [
-0.001983510097488761,
-0.018206216394901276,
-0.03287789598107338,
0.06472843885421753,
0.015308811329305172,
0.04146680608391762,
0.048229631036520004,
0.03403112664818764,
-0.013903451152145863,
-0.005701845046132803,
-0.040249455720186234,
0.032735757529735565,
-0.05664324387907982,
0.... |
LingGym: How Far Are LLMs from Thinking Like Field Linguists? | https://aclanthology.org/2025.emnlp-main.69/ | [
"Changbing Yang",
"Franklin Ma",
"Freda Shi",
"Jian Zhu"
] | This paper introduces LingGym, a new benchmark that evaluates LLMs’ capacity for meta-linguistic reasoning using Interlinear Glossed Text (IGT) and grammatical descriptions extracted from 18 typologically diverse reference grammars. Unlike previous work that focuses on specific downstream tasks, we assess whether LLMs ... | 2025.emnlp-main.69 | 10.18653/v1/2025.emnlp-main.69 | Outstanding Paper | 2511.00343 | title_snapshot | [
-0.013566604815423489,
0.024012621492147446,
-0.007663416676223278,
0.03055901825428009,
0.043627724051475525,
0.03765323758125305,
0.03603154420852661,
0.03478844836354256,
-0.024669243022799492,
0.01791761815547943,
-0.010405492037534714,
0.05672512203454971,
-0.05806659534573555,
-0.015... |
Learning from Few Samples: A Novel Approach for High-Quality Malcode Generation | https://aclanthology.org/2025.emnlp-main.70/ | [
"Haijian Ma",
"Daizong Liu",
"Xiaowen Cai",
"Pan Zhou",
"Yulai Xie"
] | Intrusion Detection Systems (IDS) play a crucial role in network security defense. However, a significant challenge for IDS in training detection models is the shortage of adequately labeled malicious samples. To address these issues, this paper introduces a novel semi-supervised framework GANGRL-LLM, which integrates ... | 2025.emnlp-main.70 | 10.18653/v1/2025.emnlp-main.70 | null | 2508.18148 | title_snapshot | [
-0.0032868641428649426,
-0.045749083161354065,
-0.01074132788926363,
0.0587141253054142,
0.0510425865650177,
-0.0025245279539376497,
0.03569880500435829,
0.016529515385627747,
-0.016654251143336296,
-0.021786794066429138,
-0.0031290084589272738,
0.019447797909379005,
-0.07796577364206314,
... |
Personality Matters: User Traits Predict LLM Preferences in Multi-Turn Collaborative Tasks | https://aclanthology.org/2025.emnlp-main.71/ | [
"Sarfaroz Yunusov",
"Kaige Chen",
"Kazi Nishat Anwar",
"Ali Emami"
] | As Large Language Models (LLMs) increasingly integrate into everyday workflows, where users shape outcomes through multi-turn collaboration, a critical question emerges: do users with different personality traits systematically prefer certain LLMs over others? We conduc-ted a study with 32 participants evenly distribut... | 2025.emnlp-main.71 | 10.18653/v1/2025.emnlp-main.71 | null | 2508.21628 | title_snapshot | [
0.0022698293905705214,
-0.017846059054136276,
0.018043914809823036,
0.015773536637425423,
0.03885361924767494,
0.0005954050575383008,
0.04233841225504875,
0.027298420667648315,
0.0031380075961351395,
-0.027856172993779182,
-0.03177099674940109,
0.04526383802294731,
-0.05483657121658325,
-0... |
VisualWebInstruct: Scaling up Multimodal Instruction Data through Web Search | https://aclanthology.org/2025.emnlp-main.72/ | [
"Yiming Jia",
"Jiachen Li",
"Xiang Yue",
"Bo Li",
"Ping Nie",
"Kai Zou",
"Wenhu Chen"
] | Vision-Language Models have made significant progress on many perception-focused tasks. However, their progress on reasoning-focused tasks remains limited due to the lack of high-quality and diverse training data. In this work, we aim to address the scarcity of reasoning-focused multimodal datasets. We propose VisualWe... | 2025.emnlp-main.72 | 10.18653/v1/2025.emnlp-main.72 | null | 2503.10582 | title_snapshot | [
0.022665541619062424,
-0.02622980810701847,
-0.003466809866949916,
0.05416345223784447,
0.030548762530088425,
0.01317417062819004,
0.0282415933907032,
0.03749074786901474,
-0.013184896670281887,
-0.03593135625123978,
-0.03297358751296997,
0.014302043244242668,
-0.08116548508405685,
0.00457... |
Thinking Out Loud: Do Reasoning Models Know When They’re Right? | https://aclanthology.org/2025.emnlp-main.73/ | [
"Qingcheng Zeng",
"Weihao Xuan",
"Leyang Cui",
"Rob Voigt"
] | Large reasoning models (LRMs) have recently demonstrated impressive capabilities in complex reasoning tasks by leveraging increased test-time computation and exhibiting behaviors reminiscent of human-like self-reflection. While LRMs show a clear capacity for valuable self-reflection, how this ability interacts with oth... | 2025.emnlp-main.73 | 10.18653/v1/2025.emnlp-main.73 | null | 2504.06564 | title_snapshot | [
-0.010853307321667671,
-0.014947420917451382,
-0.036982323974370956,
0.05169709771871567,
0.057953644543886185,
-0.017877018079161644,
0.04949657991528511,
0.020511331036686897,
-0.045473285019397736,
-0.005036925431340933,
-0.0131043316796422,
0.0536496527493,
-0.035963721573352814,
-0.00... |
Seeing is Believing, but How Much? A Comprehensive Analysis of Verbalized Calibration in Vision-Language Models | https://aclanthology.org/2025.emnlp-main.74/ | [
"Weihao Xuan",
"Qingcheng Zeng",
"Heli Qi",
"Junjue Wang",
"Naoto Yokoya"
] | Uncertainty quantification is essential for assessing the reliability and trustworthiness of modern AI systems. Among existing approaches, verbalized uncertainty, where models express their confidence through natural language, has emerged as a lightweight and interpretable solution in large language models (LLMs). Howe... | 2025.emnlp-main.74 | 10.18653/v1/2025.emnlp-main.74 | null | 2505.20236 | title_snapshot | [
0.014481225982308388,
0.01314334012567997,
-0.0166900847107172,
0.04787394404411316,
0.026577621698379517,
0.042130596935749054,
0.05249245837330818,
0.04629918187856674,
-0.04952539876103401,
-0.032911140471696854,
-0.040200479328632355,
0.0553353875875473,
-0.06351006776094437,
0.0021110... |
Enhancing Efficiency and Exploration in Reinforcement Learning for LLMs | https://aclanthology.org/2025.emnlp-main.75/ | [
"Mengqi Liao",
"Xiangyu Xi",
"Chen Ruinian",
"Jia Leng",
"Yangen Hu",
"Ke Zeng",
"Shuai Liu",
"Huaiyu Wan"
] | Reasoning large language models (LLMs) excel in complex tasks, which has drawn significant attention to reinforcement learning (RL) for LLMs. However, existing approaches allocate an equal number of rollouts to all questions during the RL process, which is inefficient. This inefficiency stems from the fact that trainin... | 2025.emnlp-main.75 | 10.18653/v1/2025.emnlp-main.75 | null | 2505.18573 | title_snapshot | [
-0.020714187994599342,
-0.0261316429823637,
0.0000863842069520615,
0.05095301568508148,
0.07402677834033966,
0.019922658801078796,
0.0050807129591703415,
-0.0008551908540539443,
-0.054575104266405106,
0.006454891990870237,
-0.04267992824316025,
0.04908011853694916,
-0.04493870958685875,
-0... |
LLM Bias Detection and Mitigation through the Lens of Desired Distributions | https://aclanthology.org/2025.emnlp-main.76/ | [
"Ingroj Shrestha",
"Padmini Srinivasan"
] | Although prior work on bias mitigation has focused on promoting social equality and demographic parity, less attention has been given to aligning LLM’s outputs to desired distributions. For example, we might want to align a model with real-world distributions to support factual grounding. Thus, we define bias as deviat... | 2025.emnlp-main.76 | 10.18653/v1/2025.emnlp-main.76 | null | 2510.06354 | title_snapshot | [
-0.003499225014820695,
-0.01871207170188427,
-0.013274316675961018,
0.007788713090121746,
0.030595894902944565,
0.04924371838569641,
0.03711848333477974,
0.00968860276043415,
-0.04163085296750069,
-0.012129592709243298,
-0.03340549021959305,
0.030044088140130043,
-0.07860530912876129,
-0.0... |
MEBench: Benchmarking Large Language Models for Cross-Document Multi-Entity Question Answering | https://aclanthology.org/2025.emnlp-main.77/ | [
"Teng Lin",
"Yuyu Luo",
"Honglin Zhang",
"Jicheng Zhang",
"Chunlin Liu",
"Kaishun Wu",
"Nan Tang"
] | Cross-Document Multi-entity question answering (MEQA) demands the integration of scattered information across documents to resolve complex queries involving entities, relationships, and contextual dependencies. Although Large Language Models (LLMs) and Retrieval-augmented Generation (RAG) systems show promise, their pe... | 2025.emnlp-main.77 | 10.18653/v1/2025.emnlp-main.77 | null | 2502.18993 | title_snapshot | [
-0.010608856566250324,
-0.014283321797847748,
-0.004041138105094433,
0.056074101477861404,
0.018862280994653702,
0.020009517669677734,
0.0043071573600173,
0.011202090419828892,
-0.01780364289879799,
-0.004690233618021011,
0.005934341810643673,
0.046525612473487854,
-0.054122231900691986,
-... |
POSITION BIAS MITIGATES POSITION BIAS: Mitigate Position Bias Through Inter-Position Knowledge Distillation | https://aclanthology.org/2025.emnlp-main.78/ | [
"Yifei Wang",
"Feng Xiong",
"Yong Wang",
"Linjing Li",
"Xiangxiang Chu",
"Daniel Dajun Zeng"
] | Positional bias (PB), manifesting as non-uniform sensitivity across different contextual locations, significantly impairs long-context comprehension and processing capabilities. Previous studies have addressed PB either by modifying the underlying architectures or by employing extensive contextual awareness training. H... | 2025.emnlp-main.78 | 10.18653/v1/2025.emnlp-main.78 | null | 2508.15709 | title_snapshot | [
-0.04280607029795647,
0.01755702495574951,
-0.020563356578350067,
0.05835678428411484,
0.023964865133166313,
-0.009233975782990456,
0.02612888254225254,
0.0044500865042209625,
-0.042930494993925095,
-0.017626382410526276,
-0.030947314575314522,
-0.005873359274119139,
-0.05148465558886528,
... |
MMLU-ProX: A Multilingual Benchmark for Advanced Large Language Model Evaluation | https://aclanthology.org/2025.emnlp-main.79/ | [
"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"... | Existing large language model (LLM) evaluation benchmarks primarily focus on English, while current multilingual tasks lack parallel questions that specifically assess cross-lingual reasoning abilities. This dual limitation makes it challenging to assess LLMs’ performance in the multilingual setting comprehensively. To... | 2025.emnlp-main.79 | 10.18653/v1/2025.emnlp-main.79 | null | 2503.10497 | title_snapshot | [
-0.028809932991862297,
-0.01798054948449135,
0.009018477983772755,
0.03382067009806633,
0.0338599793612957,
0.019301917403936386,
0.025150705128908157,
0.01832514815032482,
-0.03447176516056061,
0.001649963203817606,
-0.0063453977927565575,
0.041152238845825195,
-0.051754891872406006,
-0.0... |
NL-Debugging: Exploiting Natural Language as an Intermediate Representation for Code Debugging | https://aclanthology.org/2025.emnlp-main.80/ | [
"Weiming Zhang",
"Qingyao Li",
"Xinyi Dai",
"Jizheng Chen",
"Kounianhua Du",
"Weiwen Liu",
"Yasheng Wang",
"Ruiming Tang",
"Yong Yu",
"Weinan Zhang"
] | Debugging is a critical aspect of LLM’s coding ability. Early debugging efforts primarily focused on code-level analysis, which often falls short when addressing complex programming errors that require a deeper understanding of algorithmic logic. Recent advancements in large language models (LLMs) have shifted attentio... | 2025.emnlp-main.80 | 10.18653/v1/2025.emnlp-main.80 | null | 2505.15356 | title_snapshot | [
-0.021966373547911644,
-0.01855853758752346,
-0.02541213110089302,
0.013650402426719666,
0.06830301135778427,
0.0016418123850598931,
0.037557683885097504,
0.011606758460402489,
-0.0367763377726078,
-0.012597446329891682,
-0.04052203148603439,
0.0407874658703804,
-0.049264516681432724,
-0.0... |
Persuasion Dynamics in LLMs: Investigating Robustness and Adaptability in Knowledge and Safety with DuET-PD | https://aclanthology.org/2025.emnlp-main.81/ | [
"Bryan Chen Zhengyu Tan",
"Daniel Wai Kit Chin",
"Zhengyuan Liu",
"Nancy F. Chen",
"Roy Ka-Wei Lee"
] | Large Language Models (LLMs) can struggle to balance gullibility to misinformation and resistance to valid corrections in persuasive dialogues, a critical challenge for reliable deployment. We introduce **DuET-PD** (**Du**al **E**valuation for **T**rust in **P**ersuasive **D**ialogues), a framework evaluating multi-tur... | 2025.emnlp-main.81 | 10.18653/v1/2025.emnlp-main.81 | null | 2508.17450 | title_snapshot | [
-0.02079162932932377,
-0.018286701291799545,
-0.02406785637140274,
0.038005344569683075,
0.022001191973686218,
0.030855879187583923,
0.05346464365720749,
-0.010324429720640182,
-0.025104409083724022,
-0.014546236954629421,
-0.02280719392001629,
0.07017120718955994,
-0.04220535606145859,
-0... |
POINTS-Reader: Distillation-Free Adaptation of Vision-Language Models for Document Conversion | https://aclanthology.org/2025.emnlp-main.82/ | [
"Yuan Liu",
"Zhongyin Zhao",
"Le Tian",
"Haicheng Wang",
"Xubing Ye",
"Yangxiu You",
"Zilin Yu",
"Chuhan Wu",
"Zhou Xiao",
"Yang Yu",
"Jie Zhou"
] | High-quality labeled data is essential for training accurate document conversion models, particularly in domains with complex formats such as tables, formulas, and multi-column text. However, manual annotation is both costly and time-consuming, while automatic labeling using existing models often lacks accuracy in hand... | 2025.emnlp-main.82 | 10.18653/v1/2025.emnlp-main.82 | null | 2509.01215 | title_snapshot | [
-0.008898104541003704,
-0.011883802711963654,
-0.018528100103139877,
0.03697287663817406,
0.035144902765750885,
-0.011124171316623688,
0.006211950443685055,
0.0030592000111937523,
-0.018576765432953835,
-0.015819378197193146,
-0.048507820814847946,
-0.0016197971999645233,
-0.0591694489121437... |
Large Language Models for Automated Literature Review: An Evaluation of Reference Generation, Abstract Writing, and Review Composition | https://aclanthology.org/2025.emnlp-main.83/ | [
"Xuemei Tang",
"Xufeng Duan",
"Zhenguang Cai"
] | Large language models (LLMs) have emerged as a potential solution to automate the complex processes involved in writing literature reviews, such as literature collection, organization, and summarization. However, it is yet unclear how good LLMs are at automating comprehensive and reliable literature reviews. This study... | 2025.emnlp-main.83 | 10.18653/v1/2025.emnlp-main.83 | null | 2412.13612 | title_snapshot | [
-0.005740721710026264,
0.007294579409062862,
-0.011152086779475212,
0.006366036366671324,
0.05777684226632118,
-0.0314946174621582,
-0.0031313803046941757,
0.03617550805211067,
-0.028570586815476418,
0.006824775133281946,
-0.04502619430422783,
0.030099451541900635,
-0.053897835314273834,
0... |
CoBia: Constructed Conversations Can Trigger Otherwise Concealed Societal Biases in LLMs | https://aclanthology.org/2025.emnlp-main.84/ | [
"Nafiseh Nikeghbal",
"Amir Hossein Kargaran",
"Jana Diesner"
] | Improvements in model construction, including fortified safety guardrails, allow Large language models (LLMs) to increasingly pass standard safety checks. However, LLMs sometimes slip into revealing harmful behavior, such as expressing racist viewpoints, during conversations. To analyze this systematically, we introduc... | 2025.emnlp-main.84 | 10.18653/v1/2025.emnlp-main.84 | null | 2510.09871 | title_snapshot | [
0.0007976593333296478,
-0.03165467083454132,
-0.06865137070417404,
0.028549209237098694,
0.030977848917245865,
0.025727219879627228,
0.04272463917732239,
0.010716519318521023,
0.0026279555168002844,
0.0004775497072841972,
-0.026920650154352188,
0.04675458371639252,
-0.10181444138288498,
-0... |
From Schema to State: Zero-Shot Scheme-Only Dialogue State Tracking via Diverse Synthetic Dialogue and Step-by-Step Distillation | https://aclanthology.org/2025.emnlp-main.85/ | [
"Huan Xu",
"Zequn Li",
"Wen Tang",
"Jian Jun Zhang"
] | Dialogue State Tracking (DST) is crucial for linking user intentions to appropriate services in task-oriented dialogue systems. We propose a zero-shot, scheme-only approach that tackles two main challenges: generating synthetic dialogues that balance diversity with schema alignment, and efficiently distilling knowledge... | 2025.emnlp-main.85 | 10.18653/v1/2025.emnlp-main.85 | null | null | null | [
-0.010787292383611202,
-0.028043586760759354,
-0.02811071090400219,
0.05829690396785736,
0.061067529022693634,
0.04023157060146332,
0.046288348734378815,
0.016185322776436806,
0.01201357040554285,
-0.007626356091350317,
-0.03780728951096535,
0.037592556327581406,
-0.05316270515322685,
-0.0... |
Beyond the Surface: Measuring Self-Preference in LLM Judgments | https://aclanthology.org/2025.emnlp-main.86/ | [
"Zhi-Yuan Chen",
"Hao Wang",
"Xinyu Zhang",
"Enrui Hu",
"Yankai Lin"
] | Recent studies show that large language models (LLMs) exhibit self-preference bias when serving as judges, meaning they tend to favor their own responses over those generated by other models. Existing methods typically measure this bias by calculating the difference between the scores a judge model assigns to its own r... | 2025.emnlp-main.86 | 10.18653/v1/2025.emnlp-main.86 | null | 2506.02592 | title_snapshot | [
0.0024857039097696543,
-0.012495182454586029,
-0.006452654022723436,
0.0027440753765404224,
0.04487469047307968,
0.01252788957208395,
0.022043980658054352,
0.02719881758093834,
-0.02136598899960518,
-0.042178574949502945,
-0.007239760830998421,
0.04487941786646843,
-0.06298447400331497,
-0... |
Beyond Input Activations: Identifying Influential Latents by Gradient Sparse Autoencoders | https://aclanthology.org/2025.emnlp-main.87/ | [
"Dong Shu",
"Xuansheng Wu",
"Haiyan Zhao",
"Mengnan Du",
"Ninghao Liu"
] | Sparse Autoencoders (SAEs) have recently emerged as powerful tools for interpreting and steering the internal representations of large language models (LLMs). However, conventional approaches to analyzing SAEs typically rely solely on input-side activations, without considering the influence between each latent feature... | 2025.emnlp-main.87 | 10.18653/v1/2025.emnlp-main.87 | null | 2505.08080 | title_snapshot | [
-0.008382455445826054,
-0.00046836782712489367,
-0.014486395753920078,
0.017844030633568764,
0.041141290217638016,
0.06205686181783676,
0.04722448065876961,
0.02231762371957302,
-0.0435129813849926,
-0.019488122314214706,
-0.00997268594801426,
0.005692342296242714,
-0.03526882827281952,
0.... |
Utility-Focused LLM Annotation for Retrieval and Retrieval-Augmented Generation | https://aclanthology.org/2025.emnlp-main.88/ | [
"Hengran Zhang",
"Minghao Tang",
"Keping Bi",
"Jiafeng Guo",
"Shihao Liu",
"Daiting Shi",
"Dawei Yin",
"Xueqi Cheng"
] | This paper explores the use of large language models (LLMs) for annotating document utility in training retrieval and retrieval-augmented generation (RAG) systems, aiming to reduce dependence on costly human annotations. We address the gap between retrieval relevance and generative utility by employing LLMs to annotate... | 2025.emnlp-main.88 | 10.18653/v1/2025.emnlp-main.88 | null | 2504.05220 | title_snapshot | [
-0.019517861306667328,
-0.03721771016716957,
0.011689567938446999,
0.028882402926683426,
0.060722995549440384,
-0.008419680409133434,
-0.01509167067706585,
0.00815619807690382,
-0.01730901561677456,
-0.0018788868328556418,
-0.07206845283508301,
0.04576534032821655,
-0.051293134689331055,
-... |
CiteBART: Learning to Generate Citations for Local Citation Recommendation | https://aclanthology.org/2025.emnlp-main.89/ | [
"Ege Yiğit Çelik",
"Selma Tekir"
] | Local citation recommendation (LCR) suggests a set of papers for a citation placeholder within a given context. This paper introduces CiteBART, citation-specific pre-training within an encoder-decoder architecture, where author-date citation tokens are masked to learn to reconstruct them to fulfill LCR. The global vers... | 2025.emnlp-main.89 | 10.18653/v1/2025.emnlp-main.89 | null | 2412.17534 | title_snapshot | [
0.020828984677791595,
-0.01863921247422695,
0.010469693690538406,
0.032108161598443985,
0.03469612076878548,
-0.0154863391071558,
-0.001633552135899663,
0.01642761565744877,
-0.007091254461556673,
-0.02311951294541359,
-0.008682286366820335,
0.017031123861670494,
-0.05125141888856888,
-0.0... |
Autoformalization in the Wild: Assessing LLMs on Real-World Mathematical Definitions | https://aclanthology.org/2025.emnlp-main.90/ | [
"Lan Zhang",
"Marco Valentino",
"Andre Freitas"
] | Thanks to their linguistic capabilities, LLMs offer an opportunity to bridge the gap between informal mathematics and formal languages through autoformalization. However, it is still unclear how well LLMs generalize to sophisticated and naturally occurring mathematical statements. To address this gap, we investigate th... | 2025.emnlp-main.90 | 10.18653/v1/2025.emnlp-main.90 | Best Resource Paper | 2502.12065 | title_snapshot | [
-0.00412104744464159,
-0.02688232809305191,
-0.015013855881989002,
0.005638224072754383,
0.06385752558708191,
0.016964536160230637,
0.009328117594122887,
-0.014940911903977394,
0.0017682465258985758,
-0.004681080114096403,
-0.017427979037165642,
0.03368157520890236,
-0.038414862006902695,
... |
Culture Cartography: Mapping the Landscape of Cultural Knowledge | https://aclanthology.org/2025.emnlp-main.91/ | [
"Caleb Ziems",
"William Barr Held",
"Jane Yu",
"Amir Goldberg",
"David Grusky",
"Diyi Yang"
] | To serve global users safely and productively, LLMs need culture-specific knowledge that might not be learned during pre-training. How do we find knowledge that is (1) salient to in-group users, but (2) unknown to LLMs? The most common solutions are single-initiative: either researchers define challenging questions tha... | 2025.emnlp-main.91 | 10.18653/v1/2025.emnlp-main.91 | null | 2510.27672 | title_snapshot | [
0.019753610715270042,
-0.03222471848130226,
-0.031644899398088455,
0.034822314977645874,
0.04111231863498688,
-0.0014889492886140943,
0.04045955464243889,
0.03703804314136505,
-0.028869694098830223,
0.0033518136478960514,
-0.04724453389644623,
0.005309394095093012,
-0.03471099212765694,
-0... |
Interpretability Analysis of Arithmetic In-Context Learning in Large Language Models | https://aclanthology.org/2025.emnlp-main.92/ | [
"Gregory Polyakov",
"Christian Hepting",
"Carsten Eickhoff",
"Seyed Ali Bahrainian"
] | Large language models (LLMs) exhibit sophisticated behavior, notably solving arithmetic with only a few in-context examples (ICEs). Yet the computations that connect those examples to the answer remain opaque. We probe four open-weight LLMs, Pythia-12B, Llama-3.1-8B, MPT-7B, and OPT-6.7B, on basic arithmetic to illustr... | 2025.emnlp-main.92 | 10.18653/v1/2025.emnlp-main.92 | null | null | null | [
-0.02966708317399025,
0.01783186011016369,
-0.024159077554941177,
0.0352676697075367,
0.048332579433918,
0.03381483629345894,
0.019256453961133957,
0.025704387575387955,
-0.03634316101670265,
0.01676090806722641,
-0.008777127601206303,
0.03252233564853668,
-0.05736136808991432,
0.006391133... |
SwarmAgentic: Towards Fully Automated Agentic System Generation via Swarm Intelligence | https://aclanthology.org/2025.emnlp-main.93/ | [
"Yao Zhang",
"Chenyang Lin",
"Shijie Tang",
"Haokun Chen",
"Shijie Zhou",
"Yunpu Ma",
"Volker Tresp"
] | The rapid progress of Large Language Models has advanced agentic systems in decision-making, coordination, and task execution. Yet, existing agentic system generation frameworks lack full autonomy, missing from-scratch agent generation, self-optimizing agent functionality, and collaboration, limiting adaptability and s... | 2025.emnlp-main.93 | 10.18653/v1/2025.emnlp-main.93 | null | 2506.15672 | title_snapshot | [
-0.02142229862511158,
-0.0222554262727499,
0.014639399945735931,
0.044950686395168304,
0.051161374896764755,
0.018286390230059624,
0.014543462544679642,
0.02201012708246708,
-0.036867063492536545,
-0.05256210267543793,
0.0015802555717527866,
-0.002057565376162529,
-0.07821649312973022,
-0.... |
We Politely Insist: Your LLM Must Learn the Persian Art of Taarof | https://aclanthology.org/2025.emnlp-main.94/ | [
"Nikta Gohari Sadr",
"Sahar Heidariasl",
"Karine Megerdoomian",
"Laleh Seyyed-Kalantari",
"Ali Emami"
] | Large language models (LLMs) struggle to navigate culturally specific communication norms, limiting their effectiveness in global contexts. We focus on Persian *taarof*, a social norm in Iranian interactions, which is a sophisticated system of ritual politeness that emphasizes deference, modesty, and indirectness, yet ... | 2025.emnlp-main.94 | 10.18653/v1/2025.emnlp-main.94 | null | 2509.01035 | title_snapshot | [
-0.017561348155140877,
-0.003470970783382654,
-0.028796719387173653,
-0.003529380541294813,
0.019993413239717484,
-0.018048051744699478,
0.04727046936750412,
0.03213317692279816,
-0.0017047199653461576,
0.007178927306085825,
-0.0458795502781868,
0.0401238352060318,
-0.058504994958639145,
-... |
Unstructured Evidence Attribution for Long Context Query Focused Summarization | https://aclanthology.org/2025.emnlp-main.95/ | [
"Dustin Wright",
"Zain Muhammad Mujahid",
"Lu Wang",
"Isabelle Augenstein",
"David Jurgens"
] | Large language models (LLMs) are capable of generating coherent summaries from very long contexts given a user query, and extracting and citing evidence spans helps improve the trustworthiness of these summaries. Whereas previous work has focused on evidence citation with fixed levels of granularity (e.g. sentence, par... | 2025.emnlp-main.95 | 10.18653/v1/2025.emnlp-main.95 | null | 2502.14409 | title_snapshot | [
0.0020052804611623287,
-0.04152714088559151,
-0.03581102192401886,
0.004224252887070179,
0.05711047723889351,
-0.02016638033092022,
0.03495844453573227,
0.039437126368284225,
-0.031146010383963585,
-0.03713322803378105,
-0.03930762782692909,
0.032647810876369476,
-0.05646707862615585,
0.00... |
RAVEN: Query-Guided Representation Alignment for Question Answering over Audio, Video, Embedded Sensors, and Natural Language | https://aclanthology.org/2025.emnlp-main.96/ | [
"Subrata Biswas",
"Mohammad Nur Hossain Khan",
"Bashima Islam"
] | Multimodal question answering (QA) often requires identifying which video, audio, or sensor tokens are relevant to the question. Yet modality disagreements are common: off-camera speech, background noise, or motion outside the field of view often mislead fusion models that weight all streams equally. We present RAVEN, ... | 2025.emnlp-main.96 | 10.18653/v1/2025.emnlp-main.96 | null | 2505.17114 | title_snapshot | [
0.011937564238905907,
-0.020098229870200157,
0.008765961974859238,
0.09194300323724747,
0.009127150289714336,
0.0372307263314724,
0.046851228922605515,
0.027957985177636147,
-0.03630789741873741,
-0.019339565187692642,
-0.018980253487825394,
0.05372755974531174,
-0.0641096755862236,
-0.027... |
Cache-of-Thought: Master-Apprentice Framework for Cost-Effective Vision Language Model Reasoning | https://aclanthology.org/2025.emnlp-main.97/ | [
"Mingyuan Wu",
"Jize Jiang",
"Haozhen Zheng",
"Meitang Li",
"Zhaoheng Li",
"Beitong Tian",
"Bo Chen",
"Yongjoo Park",
"Minjia Zhang",
"ChengXiang Zhai",
"Klara Nahrstedt"
] | Vision Language Models (VLMs) have achieved remarkable success in a wide range of vision applications of increasing complexity and scales, yet choosing the right VLM model size involves a trade-off between response quality and cost. While smaller VLMs are cheaper to run, they typically produce responses only marginally... | 2025.emnlp-main.97 | 10.18653/v1/2025.emnlp-main.97 | null | 2502.20587 | title_snapshot | [
0.009168023243546486,
-0.0019792206585407257,
-0.00039888775791041553,
0.04926292225718498,
0.030458886176347733,
0.026132144033908844,
0.01280254777520895,
0.03550664708018303,
-0.01712469756603241,
-0.040990352630615234,
-0.035945549607276917,
0.032404109835624695,
-0.08651946485042572,
... |
Video Compression Commander: Plug-and-Play Inference Acceleration for Video Large Language Models | https://aclanthology.org/2025.emnlp-main.98/ | [
"Xuyang Liu",
"Yiyu Wang",
"Junpeng Ma",
"Linfeng Zhang"
] | Video large language models (VideoLLM) excel at video understanding, but face efficiency challenges due to the quadratic complexity of abundant visual tokens. Our systematic analysis of token compression methods for VideoLLMs reveals two critical issues: (i) overlooking distinctive visual signals across frames, leading... | 2025.emnlp-main.98 | 10.18653/v1/2025.emnlp-main.98 | null | 2505.14454 | title_snapshot | [
0.00988223310559988,
-0.0049875895492732525,
-0.00003109082535956986,
0.041606273502111435,
0.020288487896323204,
0.03691266104578972,
-0.00990905612707138,
0.01709320768713951,
-0.033359404653310776,
-0.028456036001443863,
-0.012633881531655788,
0.003989655524492264,
-0.039114855229854584,
... |
Router-Tuning: A Simple and Effective Approach for Dynamic Depth | https://aclanthology.org/2025.emnlp-main.99/ | [
"Shwai He",
"Tao Ge",
"Guoheng Sun",
"Bowei Tian",
"Xiaoyang Wang",
"Dong Yu"
] | The Mixture of Depths (MoD) was introduced to improve computational efficiency by dynamically skipping less important layers, reducing redundant computation while maintaining model capacity. Despite its promise, existing MoD approaches remain under-explored and face two main challenges: (1) high training costs due to t... | 2025.emnlp-main.99 | 10.18653/v1/2025.emnlp-main.99 | null | 2410.13184 | title_judge | [
-0.010436122305691242,
-0.03844035789370537,
-0.0017818063497543335,
0.030948607251048088,
0.0495065338909626,
0.0016012227861210704,
0.03263898193836212,
0.014315085485577583,
-0.04476083815097809,
-0.06973513960838318,
0.026544678956270218,
0.006088682916015387,
-0.039194997400045395,
0.... |
Foot-In-The-Door: A Multi-turn Jailbreak for LLMs | https://aclanthology.org/2025.emnlp-main.100/ | [
"Zixuan Weng",
"Xiaolong Jin",
"Jinyuan Jia",
"Xiangyu Zhang"
] | Ensuring AI safety is crucial as large language models become increasingly integrated into real-world applications. A key challenge is jailbreak, where adversarial prompts bypass built-in safeguards to elicit harmful disallowed outputs. Inspired by psychological foot-in-the-door principles, we introduce FITD, a novel m... | 2025.emnlp-main.100 | 10.18653/v1/2025.emnlp-main.100 | null | 2502.19820 | title_snapshot | [
-0.019082287326455116,
-0.04192448779940605,
-0.02924400009214878,
0.0053516775369644165,
0.06451569497585297,
0.02958993799984455,
0.055294498801231384,
-0.01961810328066349,
-0.015476012602448463,
-0.012437534518539906,
-0.005255075637251139,
-0.02992881089448929,
-0.09151865541934967,
-... |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.