Findings of the Association for Computational Linguistics: NAACL 2025
From Lazy to Prolific: Tackling Missing Labels in Open Vocabulary Extreme Classification by Positive-Unlabeled Sequence Learning (2025.findings-naacl)
Copied to clipboard
| Challenge: | Extreme multi-label classification (OXMC) is a challenging and critical task in natural language processing. |
| Approach: | They propose to use PUSL to reframe OXMC as an infinite keyphrase generation task . they propose to adopt evaluation metrics to reliably assess OXML models with incomplete ground truths. |
| Outcome: | The proposed approach improves on a highly imbalanced e-commerce dataset with missing labels . it generates 30% more unique labels and 72% of its predictions align with actual user queries . |
DiffZOO: A Purely Query-Based Black-Box Attack for Red-teaming Text-to-Image Generative Model via Zeroth Order Optimization (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing text-to-image (T2I) synthesis diffusion models raise misuse concerns, particularly in creating prohibited or not-safe-for-work (NSFW) images. |
| Approach: | They propose a method which uses zeroth order optimization to procure gradient approximations and harnesses both C-PRV and D-PRv to enhance attack prompts within a discrete prompt space. |
| Outcome: | The proposed method achieves an 8.5% higher average attack success rate than previous works on multiple state-of-the-art safety mechanisms. |
MedOdyssey: A Medical Domain Benchmark for Long Context Evaluation Up to 200K Tokens (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing benchmarks in the generic domain have evaluated long-context capabilities for LLMs. |
| Approach: | They propose a medical long-context benchmark with seven length levels ranging from 4K to 200K tokens. |
| Outcome: | The proposed benchmarks have seven length levels ranging from 4K to 200K tokens. |
Can LLMs Learn Macroeconomic Narratives from Social Media? (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing evaluation strategies for analyzing economic data with narratives are limited due to the complexity of the interplay of numerous factors and the difficulty in isolating causal relationships. |
| Approach: | They propose to use two Twitter datasets to capture economy-related narratives and use them to construct models using large language models. |
| Outcome: | The proposed models are able to predict macroeconomic fluctuations using the extracted or extracted narratives in two Twitter datasets. |
Code-Optimise: Self-Generated Preference Data for Correctness and Efficiency (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing studies have shown that CLMs can generate accurate solutions with no regard for runtime, but at a substantial cost to correctness (down by up to 30%) |
| Approach: | They propose a framework that incorporates correctness and runtime as learning signals via self-generated preference data. |
| Outcome: | The proposed framework reduces the baseline runtimes by 6% and the average length of the generated solutions is reduced by up to 48% on MBPP and 23% on HumanEval. |
People will agree what I think: Investigating LLM’s False Consensus Effect (2025.findings-naacl)
Copied to clipboard
| Challenge: | Recent studies have focused on the False Consensus Effect (FCE) where individuals overestimate the extent to which others share their beliefs or behaviors. |
| Approach: | They conduct two studies to examine the FCE phenomenon in Large Language Models (LLMs) they find that popular LLMs have FCE and that they have different prompting styles. |
| Outcome: | The proposed model is popular among LLM users and specifies the conditions when FCE becomes more or less prevalent compared to normal usage. |
LawInstruct: A Resource for Studying Language Model Adaptation to the Legal Domain (2025.findings-naacl)
Copied to clipboard
Joel Niklaus, Lucia Zheng, Arya D. McCarthy, Christopher Hahn, Brian M Rosen, Peter Henderson, Daniel E. Ho, Garrett Honke, Percy Liang, Christopher D Manning
| Challenge: | In general, instruction tuning is important for direct user interaction, but the legal domain is underrepresented in typical instruction datasets. |
| Approach: | They aggregate 58 annotated legal datasets and write instructions for each to create LawInstruct. |
| Outcome: | The proposed model improves on LegalBench across all model sizes, but no drop in MMLU. |
Stephanie: Step-by-Step Dialogues for Mimicking Human Interactions in Social Conversations (2025.findings-naacl)
Copied to clipboard
Hao Yang, Hongyuan Lu, Xinhua Zeng, Yang Liu, Xiang Zhang, Haoran Yang, Yumeng Zhang, Shan Huang, Yiran Wei, Wai Lam
| Challenge: | a new paradigm for dialogue systems is being developed to mimic human interactions . the current single-step dialogue paradigm lacks the depth and fluidity of human interactions. |
| Approach: | They propose a step-by-step dialogue paradigm that mimics human interactions . they use a dataset to fine-tune existing language models . |
| Outcome: | The proposed system mimics the dynamic nature of human conversations . it is compared with existing paradigms and will be released later this year . |
ConShift: Sense-based Language Variation Analysis using Flexible Alignment (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing methods for semantic variation analysis are limited due to the limited evaluation datasets available for word-level and sense-level variants. |
| Approach: | They propose a family of alignment-based algorithms that enable semantic variation analysis at the sense-level. |
| Outcome: | The proposed algorithms can detect multiple sense-level language variations while providing explanations through visualization of related concepts. |
Breaking the Stigma! Unobtrusively Probe Symptoms in Depression Disorder Diagnosis Dialogue (2025.findings-naacl)
Copied to clipboard
| Challenge: | Recent efforts to improve depression-diagnosis-oriented dialogue systems ignore the Stigma preventing users from open conversations about their struggles. |
| Approach: | They propose a method to promote a sense of unobtrusiveness within the dialogue system and assessing depression disorder by probing symptoms. |
| Outcome: | The proposed method improves on baselines including unobtrusiveness evaluation of dialogue content and diagnostic accuracy. |
ToVo: Toxicity Taxonomy via Voting (2025.findings-naacl)
Copied to clipboard
Tinh Son Luong, Thanh-Thien Le, Thang Viet Doan, Linh Ngo Van, Thien Huu Nguyen, Nguyen Thi Ngoc Diep
| Challenge: | Existing toxic content detection models face limitations due to the closed-source nature of training data and the paucity of explanations for their evaluation mechanism. |
| Approach: | They propose a mechanism that integrates voting and chain-of-thought processes to produce a high-quality open-source dataset for toxic content detection. |
| Outcome: | The proposed model improves transparency and customizability while facilitating better fine-tuning for specific use cases. |
HALLUCANA: Fixing LLM Hallucination with A Canary Lookahead (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing methods to detect factual hallucinations w.r.t. world knowledge require assistive generations (samples, drafts, etc.) Existing approaches to factuality hallucinism detection, such as SelfCheckGPT, require assistively generation (sequences, Drafts, and etc.). Existing studies on factualism hallucinosation detection require assistives generations (Sample, draft, etc). |
| Approach: | They propose a canary lookahead which detects and corrects factual hallucinations of Large Language Models in long-form generation by exploiting the internal factuality representation in the LLM hidden space. |
| Outcome: | The proposed method improves generation quality by 2.5x while consuming over 6 times less compute. |
Enhancing Adversarial Transferability in Visual-Language Pre-training Models via Local Shuffle and Sample-based Attack (2025.findings-naacl)
Copied to clipboard
| Challenge: | Visual-Language Pre-training (VLP) models are vulnerable to adversarial examples . previous studies have focused on improving adversariality of models . |
| Approach: | They propose a local shuffle and sample-based attack that randomly shufts one of the local image blocks and generates adversarial images and samples around them. |
| Outcome: | The proposed attack outperforms other advanced attacks on Large Vision-Language Models and outperformed previous attacks on Visual-Langue Pre-training models. |
Dis2Dis: Explaining Ambiguity in Fact-Checking (2025.findings-naacl)
Copied to clipboard
| Challenge: | Ambiguity is a linguistic tool for encoding information efficiently, yet it also causes misunderstandings and disagreements. |
| Approach: | They propose a constrained generation task for explaining ambiguous claims in fact-checking by editing them to spell out an interpretation that can be unequivocally supported by the given evidence. |
| Outcome: | The proposed model disambiguates claims 72% of the time compared to a simple copy baseline and a Large Language Model baseline. |
Enhancing Visual-Language Modality Alignment in Large Vision Language Models via Self-Improvement (2025.findings-naacl)
Copied to clipboard
Xiyao Wang, Jiuhai Chen, Zhaoyang Wang, Yuhang Zhou, Yiyang Zhou, Huaxiu Yao, Tianyi Zhou, Tom Goldstein, Parminder Bhatia, Taha Kass-Hout, Furong Huang, Cao Xiao
| Challenge: | Existing methods for visual and language alignment depend on external models or data, leading to uncontrollable and unstable results. |
| Approach: | They propose a framework that enhances visual and language alignment without external dependencies by incorporating an in-context self-critic mechanism that constructs preference pairs for tuning. |
| Outcome: | The proposed framework outperforms existing methods and improves performance on 14 hallucination and comprehensive benchmarks. |
RePD: Defending Jailbreak Attack through a Retrieval-based Prompt Decomposition Process (2025.findings-naacl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) are still susceptible to jailbreak exploits despite rigorous pre-training and fine-tuning focused on ethical alignment . |
| Approach: | They propose a Retrieval-based attack Retriest-based Prompt Decomposition framework that decomposes harmful queries embedded within user prompts into a one-shot learning example to effectively teach the LLM to discern and separate malicious components. |
| Outcome: | The proposed framework is capable of decomposing harmful queries from the original query and enhancing the resilience of large language models against jailbreak attacks without compromising their performance. |
ChatCRS: Incorporating External Knowledge and Goal Guidance for LLM-based Conversational Recommender Systems (2025.findings-naacl)
Copied to clipboard
| Challenge: | Experimental results show that ChatCRS improves language quality and informativeness by 17% and proactivity by 27%. |
| Approach: | They propose a framework to decompose the CRS task into several sub-tasks . they propose 'knowledge retrieval agent' and 'goal-planning agent' |
| Outcome: | The proposed framework improves language quality and informativeness by 17% and proactivity by 27% on two multi-goal CRS datasets. |
Data-Efficiently Learn Large Language Model for Universal 3D Scene Perception (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing methods for 3D scene understanding are limited to specific downstream tasks, hindering their practicality in real-world applications. |
| Approach: | They propose a 3D visual perceptual ability and advanced reasoning capabilities for 3D scenes by aligning 3D representations into the feature space of advanced LLMs. |
| Outcome: | The proposed system achieves a 82.2% relative score compared with state-of-the-art methods with limited data. |
UnifiedMLLM: Enabling Unified Representation for Multi-modal Multi-tasks With Large Language Model (2025.findings-naacl)
Copied to clipboard
Zhaowei Li, Wei Wang, YiQing Cai, Qi Xu, Pengyu Wang, Dong Zhang, Hang Song, Botian Jiang, Zhida Huang, Tao Wang
| Challenge: | Representative models like LLaVA and MiniGPT-4 have great capabilities in various tasks. |
| Approach: | They propose a unified model to represent various multi-modal tasks using a single representation. |
| Outcome: | The proposed model outperforms existing models in a variety of tasks while maintaining generality and scalability. |
PEMV: Improving Spatial Distribution for Emotion Recognition in Conversations Using Proximal Emotion Mean Vectors (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing research focuses on the analysis of contextual structure in dialogue and the interactions between different emotions. |
| Approach: | They propose a method that generates Proximal Emotion Mean Vectors (PEMVs) based on emotion feature queues to optimize the spatial representation of text features. |
| Outcome: | The proposed method achieves state-of-the-art performance on three widely used benchmark datasets. |
DiscoverGPT: Multi-task Fine-tuning Large Language Model for Related Table Discovery (2025.findings-naacl)
Copied to clipboard
Xuming Hu, Xiao Qin, Chuan Lei, Asterios Katsifodimos, Zhengyuan Shen, Balasubramaniam Srinivasan, Huzefa Rangwala
| Challenge: | Existing methods to learn and evaluate the table semantic relatedness of tabular data are based on pretrain-and-finetune paradigms. |
| Approach: | They propose a multi-task fine-tuning framework that holistically discovers and leverages the intricate relationships among the supervisions to optimize the performance on the data discovery task. |
| Outcome: | The proposed framework outperforms the best performing baseline by up to 7% in F1 score. |
Can GPT-4 Sway Experts’ Investment Decisions? (2025.findings-naacl)
Copied to clipboard
| Challenge: | In the post-Turing era, evaluating large language models involves assessing generated text based on readers’ decisions rather than merely its indistinguishability from human-produced content. |
| Approach: | They propose to use GPT-4 to evaluate generated text from the aspects of grammar, convincingness, logical coherence, and usefulness to determine its validity. |
| Outcome: | The proposed model can generate persuasive analyses affecting the decisions of amateurs and experts. |
PolyJoin: Semantic Multi-key Joinable Table Search in Data Lakes (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing joinable table search methods focus on single key (unary) joins, where a single column is the join key, but are ineffective when dealing with join keys composed of multiple columns (n-ary joins) Existing methods are inefficient when dealing . with joins composed of n-aries, which are prevalent on web table corpora. |
| Approach: | They propose a joinable table search method that finds multi-key joinable tables on the web, given a query table. |
| Outcome: | The proposed method outperforms the state-of-the-art methods on two real-world web table benchmarks. |
Marrying LLMs with Dynamic Forecasting: A Graph Mixture-of-expert Perspective (2025.findings-naacl)
Copied to clipboard
| Challenge: | Recent data-driven approaches often use graph neural networks (GNNs) to learn relationships in dynamical systems. |
| Approach: | They propose a framework which leverages large language models to enhance generalization capabilities of dynamical system modeling. |
| Outcome: | The proposed framework improves on existing methods and compares to baselines. |
DialogGen: Multi-modal Interactive Dialogue System with Multi-turn Text-Image Generation (2025.findings-naacl)
Copied to clipboard
Minbin Huang, Yanxin Long, Xinchi Deng, Ruihang Chu, Jiangfeng Xiong, Xiaodan Liang, Hong Cheng, Qinglin Lu, Wei Liu
| Challenge: | Text-to-image (T2I) generation models have advanced in recent years, but effective interaction with these models is challenging for average users due to the need for specialized prompt engineering knowledge and the inability to perform multi-turn image generation. |
| Approach: | They propose to use off-the-shelf MLLMs and T2I models to build a multi-modal interactive dialogue system (MIDS) that can generate correct output modalities and coherence of output images. |
| Outcome: | The proposed pipeline can generate correct output modalities and coherent multi-modal outputs compared with other state-of-the-art models. |
RELexED: Retrieval-Enhanced Legal Summarization with Exemplar Diversity (2025.findings-naacl)
Copied to clipboard
| Challenge: | Current approaches to legal summarization struggle with content theme deviation and inconsistent writing styles due to the content of the source document. |
| Approach: | They propose a retrieval-augmented framework that utilizes exemplar summaries along with the source document to guide the model. |
| Outcome: | The proposed model outperforms models that do not utilize exemplars and those that rely on similarity-based exemplar selection. |
CLaMP 2: Multimodal Music Information Retrieval Across 101 Languages Using Large Language Models (2025.findings-naacl)
Copied to clipboard
Shangda Wu, Yashan Wang, Ruibin Yuan, Guo Zhancheng, Xu Tan, Ge Zhang, Monan Zhou, Jing Chen, Xuefeng Mu, Yuejie Gao, Yuanliang Dong, Jiafeng Liu, Xiaobing Li, Feng Yu, Maosong Sun
| Challenge: | Current music information retrieval systems struggle to meet linguistic diversity challenges . current systems struggle with text queries in non-English languages . |
| Approach: | They propose a music information retrieval system that supports both ABC notation and MIDI . CLaMP 2 includes a multilingual text encoder and a multiple-modal music encoder . |
| Outcome: | The proposed system achieves state-of-the-art results in multilingual semantic search and music classification across modalities. |
LogRules: Enhancing Log Analysis Capability of Large Language Models through Rules (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing large language models (LLMs) exhibit hallucinations when analyzing logs due to the implicit knowledge and rules in logs that LLMs cannot capture. |
| Approach: | They propose a lightweight log analysis framework that generates and utilizes rules through LLMs. |
| Outcome: | The proposed framework outperforms LLM-based methods in log parsing and anomaly detection tasks and achieves better performance compared to case-based approaches. |
Audio Description Generation in the Era of LLMs and VLMs: A Review of Transferable Generative AI Technologies (2025.findings-naacl)
Copied to clipboard
| Challenge: | Audio descriptions (ADs) are acoustic commentaries designed to assist blind and visually impaired individuals in accessing digital media content. |
| Approach: | They examine how state-of-the-art NLP and CV technologies can be applied to generate ADs . they identify essential research directions for the future . |
| Outcome: | The proposed technologies can be applied to generate audio descriptions (ADs) the process is time-consuming and costly, and requires significant human effort . the authors identify key research directions for the future . |
Adaptive Retrieval-Augmented Generation for Conversational Systems (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing studies have shown the effectiveness of retrieving and augmenting external knowledge for informative responses. |
| Approach: | They propose to use a gating model to predict if a conversational system requires retrieval-augmented generation to generate high-quality responses with high confidence. |
| Outcome: | The proposed model can predict if a conversational system requires RAG to generate high-quality responses with high confidence. |
Multimodal Generation with Consistency Transferring (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing methods for multimodal content generation are limited to unimodal content production due to high training complexity, significant costs, and inadequate emphasis on model constraints. |
| Approach: | They propose a method to generate multimodal content with constraints on adjacent steps and a layer-based layer-constrained transfer between adjacent steps to improve denoising capabilities. |
| Outcome: | The proposed method improves the model’s ability to capture actions and depict backgrounds more effectively and improves video generation speed by approximately 40% and quality by about 39.3%. |
On the Impact of Noise in Differentially Private Text Rewriting (2025.findings-naacl)
Copied to clipboard
| Challenge: | a field of text privatization often requires the addition of noise to vector representations of text . however, noise addition leads to considerable utility loss, highlighting one drawback of DP in NLP. |
| Approach: | They propose a sentence-infilling privatization technique that adds noise to vector representations of text to provide privacy guarantees. |
| Outcome: | The proposed method shows that non-DP privatization methods excel in utility preservation and can find an acceptable privacy-utility trade-off, but cannot outperform DP methods in empirical privacy protections. |
Teaching Large Language Models Number-Focused Headline Generation With Key Element Rationales (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing studies focus only on textual quality and numerical accuracy for headline generation. |
| Approach: | They propose a framework for using rationales of key elements of Topic, Entities, and Numerical reasoning in news articles to enhance LLMs' ability to generate topic-aligned texts with precise numerical accuracy. |
| Outcome: | The proposed framework improves the ability of large language models to generate high-quality texts with precise numerical accuracy. |
Zero-Shot Strategies for Length-Controllable Summarization (2025.findings-naacl)
Copied to clipboard
| Challenge: | Large language models struggle with precise length control, particularly in zero-shot settings. |
| Approach: | They propose to use length approximation, target adjustment, sample filtering and automated revisions to improve LLMs' length control capabilities. |
| Outcome: | The proposed methods improve length control in large language models while maintaining or enhancing summary quality without the need for model fine-tuning or architectural changes. |
SIMPLOT: Enhancing Chart Question Answering by Distilling Essentials (2025.findings-naacl)
Copied to clipboard
| Challenge: | Recent advances in vision-language models have accelerated research into models capable of advanced reasoning based on images. |
| Approach: | They propose a method that leverages vision-language models to convert charts into table format . they use Large Language Model (LLM) for reasoning to extract only the essential information . |
| Outcome: | The proposed method extracts only the elements necessary for chart reasoning without the need for additional annotations or datasets. |
InstructAny2Pix: Image Editing with Multi-Modal Prompts (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing image editing methods struggle with complex instructions involving multiple objects or reference images. |
| Approach: | They propose a novel image editing model that leverages a multi-modal LLM to execute complex edit instructions. |
| Outcome: | The proposed model outperforms existing models and benchmarks in two multi-modal datasets. |
Lost in Overlap: Exploring Logit-based Watermark Collision in LLMs (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing watermarking methods embed imperceptible identifiers into text to address copyright concerns. |
| Approach: | They propose a new philosophy for watermark attacks that addresses watermark collision . they demonstrate that collision poses a threat to all logit-based watermark algorithms . |
| Outcome: | The proposed method improves watermark collision performance on top of other methods. |
Prompt-Guided Selective Masking Loss for Context-Aware Emotive Text-to-Speech (2025.findings-naacl)
Copied to clipboard
| Challenge: | Emotional dialogue speech synthesis (EDSS) aims to generate expressive speech by leveraging the dialogue context between interlocutors. |
| Approach: | They propose a large language model to generate holistic emotion tags based on prior dialogue context and pinpoint key words in the target utterance that align with the predicted emotional state. |
| Outcome: | The proposed method improves emotional expressiveness and facilitates automatic emotion speech generation during inference. |
Identifying and Mitigating Social Bias Knowledge in Language Models (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing methods for debiasing may generate incorrect or nonsensical predictions but leave aside individual commonsense facts, resulting in modified knowledge that elicits unreasonable or undesired predictions. |
| Approach: | They propose a framework that identifies encoding locations of biases within language models and then applies the Fairness-Stamp (FAST) they also propose 'BiaScope' to evaluate the retention of commonsense knowledge and generalization across paraphrased social biase. |
| Outcome: | The proposed framework surpasses state-of-the-art baselines with superior debiasing performance while not compromising the overall model capability for knowledge retention and prediction. |
DiaSynth: Synthetic Dialogue Generation Framework for Low Resource Dialogue Applications (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing research is limited by general or niche datasets that lack sufficient scale for training dialogue systems. |
| Approach: | They propose a synthetic dialogue generation framework that uses Large Language Models and Chain of Thought reasoning to generate dynamic, domain-specific dialogues with simulated personas and diverse conversational features. |
| Outcome: | The proposed framework outperforms existing frameworks on dialogue summarization and quality increases as the size of the LLM increases from 3B to 8B. |
Do Not Design, Learn: A Trainable Scoring Function for Uncertainty Estimation in Generative LLMs (2025.findings-naacl)
Copied to clipboard
Duygu Nur Yaldiz, Yavuz Faruk Bakman, Baturalp Buyukates, Chenyang Tao, Anil Ramakrishna, Dimitrios Dimitriadis, Jieyu Zhao, Salman Avestimehr
| Challenge: | Existing methods for probability-based UE are limited by their inability to handle biased probabilities and complex semantic dependencies between tokens. |
| Approach: | They propose a learning-based scoring function that captures complex dependencies between tokens and probabilities and produces more reliable responses. |
| Outcome: | The proposed function outperforms existing scoring functions in question-answering and arithmetical reasoning tasks with different datasets. |
Joint Learning Event-Specific Probe and Argument Library with Differential Optimization for Document-Level Multi-Event Extraction (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing methods for document-level multi-event extraction neglect the fine-grained difference between events in multi-documents, which leads to event confusion and missing. |
| Approach: | They propose an event-specific probe-based method to sniff multiple events by querying each corresponding argument library. |
| Outcome: | The proposed method outperforms the state-of-the-art method in the recall of multi-events. |
Synonym-unaware Fast Adversarial Training against Textual Adversarial Attacks (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing adversarial defense methods rely on predetermined linguistic knowledge and assume that attackers’ synonym candidates are known, which is often unrealistic. |
| Approach: | They propose a Fast Adversarial Training method that leverages single-step perturbation generation and effective perturbation initialization to improve model robustness without requiring synonym awareness. |
| Outcome: | Experiments show that the proposed method outperforms existing models under character-level and word-level attacks while still maintaining the correct syntax. |
Tethering Broken Themes: Aligning Neural Topic Models with Labels and Authors (2025.findings-naacl)
Copied to clipboard
| Challenge: | Recent studies suggest that topic models do not align well with human intentions. |
| Approach: | They propose a method to align neural topic models with both labels and authorship information. |
| Outcome: | The proposed method improves existing models in terms of topic quality and alignment. |
Towards Zero-Shot Multimodal Machine Translation (2025.findings-naacl)
Copied to clipboard
| Challenge: | Current multimodal machine translation systems rely on fully supervised data, which is costly to collect and prevents extension of MMT to language pairs with no such data. |
| Approach: | They propose a method to bypass the need for fully supervised data to train MMT systems . they adapt a strong text-only machine translation model to a visually conditioned language model and a divergence test set to evaluate how well models use images to disambiguate translations. |
| Outcome: | The proposed method can generalize to languages with no fully supervised training data. |
Large-Scale Corpus Construction and Retrieval-Augmented Generation for Ancient Chinese Poetry: New Method and Data Insights (2025.findings-naacl)
Copied to clipboard
| Challenge: | Ancient Chinese poetry presents unique challenges for Large Language Models due to data scarcity and limited ability of general LLMs when dealing with ACP. |
| Approach: | They propose a specialized Retrieval-Augmented Generation framework to improve LLMs' performance . they use 1.1 million ancient poems and 990K related texts to address hallucination issues . |
| Outcome: | The proposed framework improves performance of LLMs in ancient Chinese poetry domain from 49.2% to 89.0%. |
OpenBioNER: Lightweight Open-Domain Biomedical Named Entity Recognition Through Entity Type Description (2025.findings-naacl)
Copied to clipboard
Alessio Cocchieri, Giacomo Frisoni, Marcos Martínez Galindo, Gianluca Moro, Giuseppe Tagliavini, Francesco Candoli
| Challenge: | Biomedical Named Entity Recognition (BioNER) is a computationally expensive and limited tool . specialized 7B NER LLMs and GPT-4o can't match textual spans with entity types . |
| Approach: | They propose a lightweight BERT-based cross-encoder architecture that can identify any biomedical entity using only its description. |
| Outcome: | The proposed system outperforms existing models that match textual spans with entity types rather than descriptions on biomedical benchmarks. |
Dialetto, ma Quanto Dialetto? Transcribing and Evaluating Dialects on a Continuum (2025.findings-naacl)
Copied to clipboard
| Challenge: | Recent work on dialect variation in NLP treats dialects as discrete categories . dialect variation is a focus of increasing interest in the field . |
| Approach: | They examine performance differences between Italian dialects by incorporating performance data from different regions of the world. |
| Outcome: | The results show that performance disparities are due to dialects that are more similar to the standard variety. |
Linguistically Grounded Analysis of Language Models using Shapley Head Values (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing methods for probing language models for morphosyntactic constructions are not well understood . language models gain knowledge of grammatical phenomena during pretraining, but exactly how this knowledge is encoded is not well established. |
| Approach: | They propose a method for probing language models via Shapley Head Values . they use a BLiMP dataset to test their method on linguistic constructions based on a Shaply Head Value method . |
| Outcome: | The proposed method can be used to investigate linguistic knowledge in language models . it shows that attention heads responsible for processing related linguistic phenomena cluster together . |
How Do Large Language Models Perform in Dynamical System Modeling (2025.findings-naacl)
Copied to clipboard
| Challenge: | Recent data-driven methods often use graph neural networks (GNNs) to learn interactions between objects. |
| Approach: | They propose prompting techniques for dynamical system modeling and evaluate their performance . they find that large language models demonstrate competitive performance without training . |
| Outcome: | The proposed methods show competitive performance without training compared to state-of-the-art methods in dynamical system modeling. |
LMMs-Eval: Reality Check on the Evaluation of Large Multimodal Models (2025.findings-naacl)
Copied to clipboard
Kaichen Zhang, Bo Li, Peiyuan Zhang, Fanyi Pu, Joshua Adrian Cahyono, Kairui Hu, Shuai Liu, Yuanhan Zhang, Jingkang Yang, Chunyuan Li, Ziwei Liu
| Challenge: | Current large foundational models have demonstrated transformative capabilities, approaching or surpassing human-level performances in many tasks. |
| Approach: | They propose a unified and standardized multimodal benchmark framework with over 50 tasks and more than 10 models to promote transparent and reproducible evaluations. |
| Outcome: | The proposed framework has 50 tasks and more than 10 models to promote transparent and reproducible evaluations. |
Pairwise Prompt-Based Tuning with Parameter Efficient Fast Adaptation for Generalized Zero-Shot Intent Detection (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing methods to generalize from seen intents to unseen intents are not effective . Xian et al., 2019: a novel approach to generalized zero-shot intent detection is needed . |
| Approach: | They propose a pairwise prompt-based tuning model with parameter efficient fast adaptation . they leverage hybrid contrastive learning in discriminant space and masked language modeling . |
| Outcome: | The proposed model can generalize to unseen intents with the help of seen intents . the proposed model is based on a pairwise prompt-based tuning model with fast adaptation . |
FaithfulPersona: Balancing Faithfulness and Personalization in Code Explanations through Self-Critique (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing methods for generating faithful code explanations face challenges balancing faithfulness to the original code and personalization for diverse user needs. |
| Approach: | They propose a benchmark and method for generating faithful personalized code explanations using code samples and user profiles. |
| Outcome: | The proposed method achieves 3.7% improvement in Pass@5 compared to the strong baseline method, Self-Consistency, while maintaining high personalization with a 61.08% win rate in the LLM-as-a-Judge evaluation. |
Efficient Multi-Agent Collaboration with Tool Use for Online Planning in Complex Table Question Answering (2025.findings-naacl)
Copied to clipboard
| Challenge: | Recent studies on complex table question answering focus more and more on complex instances, as they are ubiquitous in table data analysis. |
| Approach: | They propose a framework that requires neither fine-tuning nor closed-source models to solve complex table question answering (TQA) their framework outperforms previous SoTA systems on three out of four benchmarks and performs comparably to the larger and more expensive closed-sourced model GPT-4 on two benchmarks. |
| Outcome: | The proposed framework outperforms closed-source models and closed-sourced models on three out of four benchmarks and performs comparable to the larger and more expensive closed-Source model GPT-4 on two benchmarks. |
Ground Every Sentence: Improving Retrieval-Augmented LLMs with Interleaved Reference-Claim Generation (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing methods to enhance credibility and verifiability of large language models (LLMs) mainly focus on passage-level or paragraph-level references or citations, which fall short in verifikatability. |
| Approach: | They propose a method that provides sentence-level citations in LLM-generated responses. |
| Outcome: | The proposed method achieves 90% accuracy in long-form question-answering tasks. |
Understanding the Role of Mental Models in User Interaction with an Adaptive Dialog Agent (2025.findings-naacl)
Copied to clipboard
| Challenge: | Adaptive dialog systems can help users align their behavior with user expectations, but there is little research into what mental models users form when interacting with a task-oriented dialog system. |
| Approach: | They propose to use a publicly available dataset to explore user mental models of dialog systems to better align with users' mental models. |
| Outcome: | The proposed model can improve dialog efficiency, success, and user perception of the interaction, even when done implicitly. |
CoPERLex: Content Planning with Event-based Representations for Legal Case Summarization (2025.findings-naacl)
Copied to clipboard
| Challenge: | Recent efforts to produce concise legal summarization have shifted towards abstractive approaches . |
| Approach: | They propose a framework that integrates content selection and planning components to generate coherent summaries based on both the content and the structured plan. |
| Outcome: | The proposed framework shows that it integrates content selection and planning components over entity-centric approaches in the context of legal judgements. |
DisComp: A Two-Stage Prompt Optimization Framework Combining Task-Agnostic and Task-Aware Compression (2025.findings-naacl)
Copied to clipboard
| Challenge: | Extended prompts can lead to substantial computational overhead and increased hardware demands, limiting the scalability and efficiency of large language models. |
| Approach: | They propose a two-stage prompt compression framework that combines task-agnostic and task-based strategies to efficiently compress prompt length without compromising performance. |
| Outcome: | The proposed framework outperforms task-agnostic and task-specific compression methods on three benchmark datasets and is up to 6.56 faster at inference compared to the best token-level compression method. |
A Large-Scale Benchmark for Vietnamese Sentence Paraphrases (2025.findings-naacl)
Copied to clipboard
| Challenge: | 1.2M original–paraphrase pairs were generated using a hybrid approach to generate high-quality paraphrases. |
| Approach: | They present a high-quality Vietnamese dataset for sentence paraphrasing . they used automatic paraphrase generation and manual evaluation to ensure high quality . |
| Outcome: | The proposed dataset is the first large-scale study on Vietnamese paraphrasing . it combines automatic paraphrase generation with manual evaluation to ensure high quality . |
RAMQA: A Unified Framework for Retrieval-Augmented Multi-Modal Question Answering (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing ranking methods rely on small encoder-based ranking models, which are incompatible with modern decoder--based generative large language models (LLMs) Existing methods based on small LLaVA rankers are incompatible with advanced LLMs. |
| Approach: | They propose a framework that combines learning-to-rank methods with generative permutation-enhanced ranking techniques. |
| Outcome: | The proposed framework improves on two benchmarks, WebQA and MultiModalQA, showing significant improvements over baselines. |
MultiCAT: Multimodal Communication Annotations for Teams (2025.findings-naacl)
Copied to clipboard
Adarsh Pyarelal, John M Culnan, Ayesha Qamar, Meghavarshini Krishnaswamy, Yuwei Wang, Cheonkam Jeong, Chen Chen, Md Messal Monem Miah, Shahriar Hormozi, Jonathan Tong, Ruihong Huang
| Challenge: | Recent flagship models from OpenAI and Google are only capable of 1-on-1 interactions with humans, limiting the potential for integration into human-machine teams of the future. |
| Approach: | They propose a dataset that allows team members to make multiple types of predictions on the same dataset. |
| Outcome: | The proposed dataset builds upon data from teams working collaboratively to save victims in a simulated search and rescue mission. |
Prototype Tuning: A Meta-Learning Approach for Few-Shot Document-Level Relation Extraction with Large Language Models (2025.findings-naacl)
Copied to clipboard
| Challenge: | Few-Shot Document-Level Relation Extraction (FSDLRE) aims to develop models capable of generalizing to new categories with minimal support examples. |
| Approach: | They propose a meta-training approach to train Large Language Models to improve their ICL capabilities . they construct simulated episodes using relation types that do not overlap with test corpus . |
| Outcome: | Experimental results show that the proposed approach outperforms baseline models on few-shot tasks. |
LegalSeg: Unlocking the Structure of Indian Legal Judgments Through Rhetorical Role Classification (2025.findings-naacl)
Copied to clipboard
Shubham Kumar Nigam, Tanmay Dubey, Govind Sharma, Noel Shallum, Kripabandhu Ghosh, Arnab Bhattacharya
| Challenge: | a lack of large-scale annotated datasets hinders effective training of ML models . despite advances in semantic segmentation, challenges persist in distinguishing between closely related roles . |
| Approach: | They propose a large annotated dataset for semantic segmentation of legal documents . they use a rhetorical role classification model to compare performance against other models . |
| Outcome: | The largest annotated dataset for this task outperforms models relying on sentence-level features. |
Claim-Guided Textual Backdoor Attack for Practical Applications (2025.findings-naacl)
Copied to clipboard
| Challenge: | a novel backdoor attack is based on textual claims to trick models into misbehaving on targeted claims. |
| Approach: | a new backdoor attack is designed to trick models into misbehaving on targeted claims . the code and data will be available at https://github.com/minkyoo9/CGBA . |
| Outcome: | a new backdoor attack exploits the power of textual claims to trick models into misbehaving on claims without affecting their performance on clean data. |
ToolSandbox: A Stateful, Conversational, Interactive Evaluation Benchmark for LLM Tool Use Capabilities (2025.findings-naacl)
Copied to clipboard
Jiarui Lu, Thomas Holleis, Yizhe Zhang, Bernhard Aumayer, Feng Nan, Haoping Bai, Shuang Ma, Shen Ma, Mengyu Li, Guoli Yin, Zirui Wang, Ruoming Pang
| Challenge: | Recent advances in large language models have led to a growing interest in tool assisted LLMs . toolSandbox includes stateful tool execution, implicit state dependencies between tools . |
| Approach: | a new tool-based evaluation tool is released to help LLMs evaluate their tool-use capabilities. a tool-driven evaluation tool includes stateful tool execution, implicit state dependencies between tools and a built-in user simulator. |
| Outcome: | the toolSandbox evaluation benchmark shows that open source and proprietary models have a performance gap . the benchmarks show that even the most capable LLMs are challenged by state dependent tasks . |
SusGen-GPT: A Data-Centric LLM for Financial NLP and Sustainability Report Generation (2025.findings-naacl)
Copied to clipboard
Qilong Wu, Xiaoneng Xiang, Huang Hejia, Xuan Wang, Yeo Wei Jie, Ranjan Satapathy, Ricardo Shirota Filho, Bharadwaj Veeravalli
| Challenge: | Existing tools for financial reporting and ESG analysis are lacking . large language models are not proficient across general finance and ESE domains . |
| Approach: | They propose a dataset that includes seven financial NLP tasks and a benchmark to improve sustainability report generation. |
| Outcome: | SusGen-30k, a high-quality dataset, shows state-of-the-art performance . it surpasses all other models except GPT-4 in six adapted tasks and two off-the shelf tasks . |
GrEmLIn: A Repository of Green Baseline Embeddings for 87 Low-Resource Languages Injected with Multilingual Graph Knowledge (2025.findings-naacl)
Copied to clipboard
| Challenge: | Contextualized word embeddings are available for many languages, but their coverage is limited for low resourced languages. |
| Approach: | They propose a method that integrates multilingual graph knowledge into the embeddings to make them green. |
| Outcome: | The proposed method outperforms state-of-the-art embeddings on lexical similarity task while being parameter-free at inference time. |
In-Context Example Selection via Similarity Search Improves Low-Resource Machine Translation (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing studies have shown that in-context examples for machine translation are beneficial for high-resource languages. |
| Approach: | They propose to use in-context examples for machine translation (MT) they argue that similarity-based selection can improve MT . |
| Outcome: | The proposed approach improves machine translation (MT) and low-resource languages. |
Self-Training Large Language Models for Tool-Use Without Demonstrations (2025.findings-naacl)
Copied to clipboard
| Challenge: | Recent work augmented LLMs with tools to mitigate factual inaccuracies and computational errors. |
| Approach: | They propose a method to synthesise tool-use traces using the LLM itself. |
| Outcome: | The proposed method improves performance on a long-tail knowledge task, but not on other datasets. |
Can Large Language Models Generate High-quality Patent Claims? (2025.findings-naacl)
Copied to clipboard
| Challenge: | Large language models (LLMs) have shown exceptional performance across various text generation tasks, but remain under-explored in the patent domain, which offers highly structured and precise language. |
| Approach: | They construct a dataset to investigate the performance of current LLMs in patent claim generation. |
| Outcome: | The proposed model outperforms state-of-the-art general LLMs in patent claim generation. |
Obliviate: Neutralizing Task-agnostic Backdoors within the Parameter-efficient Fine-tuning Paradigm (2025.findings-naacl)
Copied to clipboard
| Challenge: | Parameter-efficient fine-tuning (PEFT) is a key training strategy for large language models, but its reliance on fewer trainable parameters poses security risks, such as task-agnostic backdoors. |
| Approach: | They propose a method that amplifies benign neurons within PEFT layers and penalizes the influence of trigger tokens. |
| Outcome: | The proposed method reduces the success rate of state-of-the-art task-agnostic backdoors by 83.6%. |
CORAL: Benchmarking Multi-turn Conversational Retrieval-Augmented Generation (2025.findings-naacl)
Copied to clipboard
Yiruo Cheng, Kelong Mao, Ziliang Zhao, Guanting Dong, Hongjin Qian, Yongkang Wu, Tetsuya Sakai, Ji-Rong Wen, Zhicheng Dou
| Challenge: | Existing research focuses on single-turn RAG, leaving a gap in addressing multi-turn conversations . a new benchmark is designed to assess RAG systems in realistic multi-turned conversations based on Wikipedia . |
| Approach: | They propose a large-scale benchmark to assess RAG systems in multi-turn contexts . CORAL includes diverse information-seeking conversations automatically derived from Wikipedia . authors propose unified framework to standardize various conversational RAG methods . |
| Outcome: | The proposed framework supports three core tasks of conversational RAG: passage retrieval, response generation, and citation labeling. |
Beyond English: The Impact of Prompt Translation Strategies across Languages and Tasks in Multilingual LLMs (2025.findings-naacl)
Copied to clipboard
| Challenge: | Current LLMs are primarily trained on English data but also include data from other languages. |
| Approach: | They propose to use a pre-translation strategy to translate a task prompt into English before inference . they use 'a modular entity' that could be translated into four different languages . |
| Outcome: | The proposed strategies are based on a set of pre-trained data across 35 languages covering both low and high-resource languages. |
QuaLLM: An LLM-based Framework to Extract Quantitative Insights from Online Forums (2025.findings-naacl)
Copied to clipboard
| Challenge: | Qualitative and quantitative methods to analyze text data on online forums are infeasible to scale or require significant human effort to translate outputs to human readable forms. |
| Approach: | They propose a novel LLM-based framework to analyze and extract quantitative insights from text data on online forums. |
| Outcome: | The proposed framework analyzes over one million comments from two of Reddit’s rideshare worker communities, marking the largest study of its type. |
The Promises and Pitfalls of LLM Annotations in Dataset Labeling: a Case Study on Media Bias Detection (2025.findings-naacl)
Copied to clipboard
Tomáš Horych, Christoph Mandl, Terry Ruas, Andre Greiner-Petter, Bela Gipp, Akiko Aizawa, Timo Spinde
| Challenge: | Recent research suggests using Large Language Models (LLMs) to automate the annotation process, reducing these costs while maintaining data quality. |
| Approach: | They propose to use Large Language Models to automate annotation process and train classifiers on large datasets. |
| Outcome: | The proposed model outperforms all of the annotator LLMs on two media bias benchmark datasets (BABE and BASIL) while maintaining data quality. |
Mechanistic Unveiling of Transformer Circuits: Self-Influence as a Key to Model Reasoning (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing studies have shown that large language models implicitly embed reasoning trees, but their internal mechanisms remain largely opaque due to the complexity of non-linear interactions and high-dimensional operations. |
| Approach: | They propose to use circuit analysis and self-influence functions to map the reasoning process of large models. |
| Outcome: | The proposed model is able to map human-interpretable reasoning paths and a model's underlying circuits reveal human-mediated reasoning processes. |
Intrinsic Model Weaknesses: How Priming Attacks Unveil Vulnerabilities in Large Language Models (2025.findings-naacl)
Copied to clipboard
| Challenge: | Large language models (LLMs) have significant impact on various industries and societal functions due to advanced instruction-following capabilities. |
| Approach: | They developed and tested novel attack strategies on popular LLMs to expose their vulnerabilities in generating harmful content. |
| Outcome: | The proposed attacks achieved an ASR of 100% on open-source models, including Meta’s Llama-3.2, Google’s Gemma-2, Mistral’s Mistral-NeMo, Falcon’s Falcon-mamba, Apple’s DCLM, Microsoft’s Phi3, and Qwen’s Qwend2.5, among others. |
AdParaphrase: Paraphrase Dataset for Analyzing Linguistic Features toward Generating Attractive Ad Texts (2025.findings-naacl)
Copied to clipboard
| Challenge: | Effective linguistic choices that attract potential customers play crucial roles in advertising success. |
| Approach: | They propose to use a paraphrase dataset to explore linguistic features of ad texts that influence human preferences to maximize the potential success of advertisements. |
| Outcome: | The proposed model improves the attractiveness of ad texts by focusing on human preferences. |
Token Weighting for Long-Range Language Modeling (2025.findings-naacl)
Copied to clipboard
| Challenge: | Many applications of large language models (LLMs) require long-context understanding, but models still struggle with such tasks. |
| Approach: | They propose token-weighting schemes that assign different weights to each training token in the loss, generalizing existing works. |
| Outcome: | The proposed methods compare confidences of a long-context and short-concept model and show that non-uniform loss weights improve the long-constability of LLMs. |
Learning to Explore and Select for Coverage-Conditioned Retrieval-Augmented Generation (2025.findings-naacl)
Copied to clipboard
| Challenge: | Recent advances of large language models (LLMs) have enabled them to provide long and detailed responses by leveraging their parametric knowledge. |
| Approach: | They introduce a dataset of 10K hierarchical sets of information-seeking subqueries that define structured boundaries for outline creation and evaluation in C2 scenarios. |
| Outcome: | The proposed model generates higher-quality outlines that better fulfill diverse user needs when trained with alignment techniques like DPO. |
LayAlign: Enhancing Multilingual Reasoning in Large Language Models via Layer-Wise Adaptive Fusion and Alignment Strategy (2025.findings-naacl)
Copied to clipboard
| Challenge: | Large language models (LLMs) are pretrained on multilingual corpora but exhibit suboptimal performance on low-resource languages. |
| Approach: | They propose a framework that integrates representations from all encoder layers and an adaptive fusion-enhanced attention mechanism to enable layer-wise interaction between the LLM and the multilingual encoder. |
| Outcome: | Experiments on multilingual reasoning tasks show that the proposed framework outperforms baselines. |
On the Impacts of Contexts on Repository-Level Code Generation (2025.findings-naacl)
Copied to clipboard
| Challenge: | CodeLLMs are widely used for code generation, but their ability to handle repository-level dependencies remains underexplored. |
| Approach: | They propose a benchmark for evaluating repository-level code generation based on dependency contexts. |
| Outcome: | The proposed model improves dependency handling and introduces a new metric, Dependency Invocation Rate (DIR), to measure context utilization. |
From Argumentation to Deliberation: Perspectivized Stance Vectors for Fine-grained (Dis)agreement Analysis (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing methods to identify conflict resolution points require a deeper analysis of arguments and the perspectives they are grounded in. |
| Approach: | They propose a framework for a deliberative analysis of arguments in a computational argumentation setup. |
| Outcome: | The proposed framework allows us to identify actionable options for conflict resolution, as a first step towards deliberation. |
LVLM-Compress-Bench: Benchmarking the Broader Impact of Large Vision-Language Model Compression (2025.findings-naacl)
Copied to clipboard
Souvik Kundu, Anahita Bhiwandiwalla, Sungduk Yu, Phillip Howard, Tiep Le, Sharath Nittur Sridhar, David Cobbley, Hao Kang, Vasudev Lal
| Challenge: | LVLMs have been shown to perform well on simple uni-modal benchmarks, but their detailed study on multi-modal models is still lacking. |
| Approach: | They propose a framework to analyze the impact of compression on LVLMs on multi-modal input driven tasks. |
| Outcome: | The proposed framework analyzes the impact of compression on generative performance of large vision language models on multi-modal input driven tasks. |
Does Generative AI speak Nigerian-Pidgin?: Issues about Representativeness and Bias for Multilingualism in LLMs (2025.findings-naacl)
Copied to clipboard
| Challenge: | Nigeria is a multilingual country with 500+ languages. |
| Approach: | They propose to use a pidgin and a creole to analyze the pidgins of Nigeria . they also use machine translation to analyze their results . |
| Outcome: | The results show that the two pidgins do not represent each other and are hard to teach . the results show the pidgin varieties are underrepresented in Generative AI . |
A Guide To Effectively Leveraging LLMs for Low-Resource Text Summarization: Data Augmentation and Semi-supervised Approaches (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing approaches for low-resource text summarization use large language models (LLMs) but such models suffer from inconsistent outputs and are difficult to adapt to domain-specific data. |
| Approach: | They propose two methods to effectively utilize large language models for low-resource text summarization. |
| Outcome: | The proposed methods synthesize high-quality documents using LLaMA-3-70b-Instruct model . they achieve competitive ROUGE scores as a fully supervised method with 5% of the labeled data. |
Decoding Dark Matter: Specialized Sparse Autoencoders for Interpreting Rare Concepts in Foundation Models (2025.findings-naacl)
Copied to clipboard
| Challenge: | Sparse Autoencoders (SAEs) are a promising tool for disentangling FM representations, but they struggle to capture rare, yet crucial concepts in the data. |
| Approach: | They propose a technique to train Sparse Autoencoders to illuminate elusive dark matter features by focusing on specific subdomains. |
| Outcome: | The proposed method achieves 12.5% better classification accuracy than general-purpose SAEs when applied to remove spurious gender information. |
MAiDE-up: Multilingual Deception Detection of AI-generated Hotel Reviews (2025.findings-naacl)
Copied to clipboard
| Challenge: | Deceptive reviews are becoming more common, especially given the increase in performance and the prevalence of LLMs. |
| Approach: | They compile and make publicly available a dataset of 10,000 real and 10,000 AI-generated fake hotel reviews in ten languages. |
| Outcome: | The proposed model can detect real reviews and fake reviews in 10 languages. |
LeCoPCR: Legal Concept-guided Prior Case Retrieval for European Court of Human Rights cases (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing approaches overlook the underlying semantic intent in determining relevance with respect to a query case. |
| Approach: | They propose a method that generates intents in the form of legal concepts from a query case facts and then augments the query with these concepts to enhance models understanding of semantic intent. |
| Outcome: | The proposed approach generates intents in the form of legal concepts and augments the query with these concepts to enhance models understanding of semantic intent that dictates relavance. |
How much do contextualized representations encode long-range context? (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing studies of contextualized representations focus on short sequences of tens to hundreds of tokens, whereas modern language models handle hundreds of thousands of token in a single context window. |
| Approach: | They use a perturbation setup and a metric to capture contextualization of long-range patterns from the perspective of representation geometry. |
| Outcome: | The proposed model can encode long-range contexts, but it's not fully recurrent, the authors say . their results suggest improvements in existing language models . |
Data Poisoning for In-context Learning (2025.findings-naacl)
Copied to clipboard
| Challenge: | In-context learning (ICL) has emerged as a capability of large language models (LLMs) but there is limited understanding of its vulnerability against data poisoning attacks. |
| Approach: | They propose an attack method that exploits ICL’s unique learning mechanisms by identifying discrete text perturbations that influence LLM hidden states. |
| Outcome: | The proposed attack method exploits ICL’s learning mechanisms by identifying discrete text perturbations that influence LLM hidden states. |
Synthetic Audio Helps for Cognitive State Tasks (2025.findings-naacl)
Copied to clipboard
| Challenge: | Prior work in NLP focuses on tasks that involve extracting information about the cognitive states of human entities from text. |
| Approach: | They propose a framework for learning to add synthetic audio to text-only corpora and a system that automatically tracks audio signals to produce naturalistic audio. |
| Outcome: | The proposed framework improves on 7 cognitive state modeling tasks on text and synthetic audio data from an off-the-shelf TTS system. |
BioEL: A Comprehensive Python Package for Biomedical Entity Linking (2025.findings-naacl)
Copied to clipboard
Prasanth Bathala, Christophe Ye, Batuhan Nursal, Shubham Lohiya, David Kartchner, Cassie S. Mitchell
| Challenge: | Entity Linking in biomedical literature is a critical task that enhances the extraction and integration of information from diverse scientific literature. |
| Approach: | They propose a Python package that allows for better Entity Linking in biomedical literature . the package includes four components: Ontology Object, Dataset Object and Evaluation Framework . |
| Outcome: | The proposed open-source package enables the implementation and comparison of biomedical entity linking tasks. |
PairScale: Analyzing Attitude Change with Pairwise Comparisons (2025.findings-naacl)
Copied to clipboard
| Challenge: | a text-based framework for measuring attitudes in communities is proposed . the framework uses both implicit and explicit evidence in language to characterize attitudes . |
| Approach: | They propose a text-based framework for measuring attitudes in communities toward issues of interest using language. |
| Outcome: | The proposed framework is validated by examining attitudes on two high-profile issues in the u.s. |
Semantic Consistency-Based Uncertainty Quantification for Factuality in Radiology Report Generation (2025.findings-naacl)
Copied to clipboard
| Challenge: | Radiology report generation has shown great potential in assisting radiologists . generative medical Vision Large Language Models (VLLMs) are prone to hallucinations and can produce inaccurate diagnostic information. |
| Approach: | They propose a framework that provides both report-level and sentence-level uncertainties. |
| Outcome: | The proposed method improves factuality scores by 10% by rejecting 20% of reports on the MIMIC-CXR dataset. |
RewardBench: Evaluating Reward Models for Language Modeling (2025.findings-naacl)
Copied to clipboard
Nathan Lambert, Valentina Pyatkin, Jacob Morrison, Lester James Validad Miranda, Bill Yuchen Lin, Khyathi Chandu, Nouha Dziri, Sachin Kumar, Tom Zick, Yejin Choi, Noah A. Smith, Hannaneh Hajishirzi
| Challenge: | Evaluating reward models presents an opportunity to understand the opaque technologies used for alignment of language models. |
| Approach: | They present a benchmark dataset and code-base for evaluation of reward models . they use prompt-chosen-rejected trios to benchmark how they perform on queries . |
| Outcome: | The proposed dataset compares RMs with other models on a set of questions. |
Evaluating Vision-Language Models for Emotion Recognition (2025.findings-naacl)
Copied to clipboard
| Challenge: | Large Vision-Language Models (VLMs) have been used for objective multimodal reasoning tasks for decades. |
| Approach: | They present a comprehensive evaluation of large vision-language models for recognizing evoked emotions from images. |
| Outcome: | The proposed model performs well in evoked emotion recognition task and is robust to human errors. |
Tomato, Tomahto, Tomate: Do Multilingual Language Models Understand Based on Subword-Level Semantic Concepts? (2025.findings-naacl)
Copied to clipboard
| Challenge: | a recent study shows that human understanding of text depends on general semantic concepts of words that are robust to their superficial forms. |
| Approach: | They evaluate the accuracy of multilingual multilingual language models based on subword-level semantics . they form "semantic tokens" by merging semantically similar subwords and embeddings based upon the results . |
| Outcome: | The proposed models are able to make predictions on multilingual tasks with different tokenizers and model sizes. |
Open Domain Question Answering with Conflicting Contexts (2025.findings-naacl)
Copied to clipboard
Siyi Liu, Qiang Ning, Kishaloy Halder, Zheng Qi, Wei Xiao, Phu Mon Htut, Yi Zhang, Neha Anna John, Bonan Min, Yassine Benajiba, Dan Roth
| Challenge: | Open domain question answering systems often rely on information retrieved from large collections of text to answer questions. |
| Approach: | They evaluate and benchmark three powerful Large Language Models with a dataset . they find that 25% of unambiguous open domain questions can lead to conflicting contexts . |
| Outcome: | The proposed model can't be used to answer questions with conflicting contexts . it can be fine tuned to provide richer information into the model's training . |
The Geometry of Prompting: Unveiling Distinct Mechanisms of Task Adaptation in Language Models (2025.findings-naacl)
Copied to clipboard
| Challenge: | Recent studies have focused on the inputoutput behavior of LMs, leaving the internal mechanisms behind ICL largely unexplored. |
| Approach: | They investigate how different prompting methods modify internal representations in pre-trained language models. |
| Outcome: | The proposed model can be programmed with natural language to perform a wide array of tasks without expensive fine-tuning. |
Biases in Opinion Dynamics in Multi-Agent Systems of Large Language Models: A Case Study on Funding Allocation (2025.findings-naacl)
Copied to clipboard
| Challenge: | We identify biases that drive the exchange of opinions based on the LLM’s tendency to find consensus with the other LLM's opinion, display caution when specifying funding, and consider ethical concerns in its opinion. |
| Approach: | They identify biases that drive the exchange of opinions based on the LLM’s tendency to find consensus with the other LLM's opinion, display caution when specifying funding, and consider ethical concerns in its opinion. |
| Outcome: | The proposed model can be used to predict the evolution of opinions in large language models. |
CaseSumm: A Large-Scale Dataset for Long-Context Summarization from U.S. Supreme Court Opinions (2025.findings-naacl)
Copied to clipboard
| Challenge: | CaseSumm is a dataset for long-context summarization in the legal domain . human groundtruth summaries are often not available for legal summarizing . |
| Approach: | They propose a dataset for long-context summarization that includes SCOTUS opinions and their official summaries. |
| Outcome: | The proposed dataset is the largest open legal case summarization dataset . it outperforms larger models on automatic metrics and human evaluation . |
Chasing Random: Instruction Selection Strategies Fail to Generalize (2025.findings-naacl)
Copied to clipboard
| Challenge: | Prior work has shown that language models can be tuned to follow user instructions using only a small set of high-quality instructions. |
| Approach: | They analyze popular selection strategies across different datasets and benchmarks to find out whether they generalize poorly. |
| Outcome: | The proposed methods outperform random baselines and cost-performance trade-offs on the full dataset and a random subset. |
Can’t Hide Behind the API: Stealing Black-Box Commercial Embedding Models (2025.findings-naacl)
Copied to clipboard
| Challenge: | a new study examines the retrieval effectiveness of commercial embedding models . robert mcgahey: can commercial embeds be "stolen" using distillation techniques? he says stealing models can offer benefits to different actors, including reduced costs and security . |
| Approach: | They propose to "steal" embedding models by training thief models on text–embedding pairs . they replicate retrieval effectiveness of commercial embeddable models with a cost of under $300 . |
| Outcome: | The proposed methods replicate retrieval effectiveness of commercial embedding models with under $300 . authors suggest measures to mitigate risk of model theft. |
CAMEL-Bench: A Comprehensive Arabic LMM Benchmark (2025.findings-naacl)
Copied to clipboard
Sara Ghaboura, Ahmed Heakl, Omkar Thawakar, Ali Husain Salem Abdulla Alharthi, Ines Riahi, Abduljalil Radman, Jorma Laaksonen, Fahad Shahbaz Khan, Salman Khan, Rao Muhammad Anwer
| Challenge: | Recent years have witnessed a significant interest in developing large multimodal models capable of performing various visual reasoning and understanding tasks. |
| Approach: | They propose to use Arabic as a language to evaluate large multi-modal models capable of performing visual reasoning and understanding tasks. |
| Outcome: | The proposed benchmark comprises eight diverse domains and 38 sub-domains to represent a large population of over 400 million speakers. |
ProxyLM: Predicting Language Model Performance on Multilingual Tasks via Proxy Models (2025.findings-naacl)
Copied to clipboard
| Challenge: | Performance prediction is a method to estimate the performance of Language Models (LMs) on various Natural Language Processing (NLP) tasks. |
| Approach: | They propose a task- and language-agnostic framework to predict the performance of Language Models (LMs) using proxy models. |
| Outcome: | The proposed framework outperforms the state-of-the-art in root-mean-square error (RMSE) and other robustness tests on multilingual NLP tasks. |
SimSMoE: Toward Efficient Training Mixture of Experts via Solving Representational Collapse (2025.findings-naacl)
Copied to clipboard
| Challenge: | Sparse mixture of experts (SMoE) is an effective approach for scaling large language models with a constant computational cost. |
| Approach: | They propose a similarity-based sparse mixture of experts (SimSMoE) that guarantees a solution to address the representation collapse problem between experts given a fixed FLOPs budget. |
| Outcome: | The proposed method outperforms existing methods and outperformed other SMoE routing methods in performance. |
UniRAG: Universal Retrieval Augmentation for Large Vision Language Models (2025.findings-naacl)
Copied to clipboard
| Challenge: | Large Vision Language Models (LVLMs) have unlocked many complex use cases that require Multi-Modal (MM) understanding and MM generation. |
| Approach: | They propose a plug-and-play technique that adds relevant retrieved information to prompts as few-shot examples during inference. |
| Outcome: | The proposed method significantly improves the output quality of large vision language models when input prompts are augmented with relevant information retrieved by Vision-Language retrievers like UniRAG. |
Evaluating the Performance of Large Language Models via Debates (2025.findings-naacl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) are evolving and impacting various fields . current methods for evaluation are based on fixed, domain-specific questions or rely on human input, making them unscalable. |
| Approach: | They propose a benchmarking framework based on debates between LLMs, judged by another LLM. |
| Outcome: | The proposed framework achieves rankings that align closely with popular rankings based on human input eliminating the need for costly crowdsourcing. |
CausalGraph2LLM: Evaluating LLMs for Causal Queries (2025.findings-naacl)
Copied to clipboard
| Challenge: | Recent advances in Large Language Models (LLMs) have opened up new avenues for their use beyond standard Natural Language Processing tasks. |
| Approach: | They propose a benchmark to evaluate the capabilities of Large Language Models (LLMs) they use over 700k queries to compare their encoding capabilities. |
| Outcome: | The proposed benchmark compared LLMs on graph-level and node-level queries and open-sourced and closed models. |
PuzzleGPT: Emulating Human Puzzle-Solving Ability for Time and Location Prediction (2025.findings-naacl)
Copied to clipboard
| Challenge: | PuzzleGPT is a modular and iterative puzzlesolving method for predicting time and location from images. |
| Approach: | They propose to formalize this ability into core skills and implement it using different modules in an expert pipeline called PuzzleGPT. |
| Outcome: | The proposed method outperforms large VLMs and finetuned models on TARA and WikiTilo and rivals or surpasses finetuned models. |
SAFR: Neuron Redistribution for Interpretability (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing studies on controlling neuron distribution for interpretability have focused on focusing on monosemanticity instead of focusing solely on feature interactions. |
| Approach: | They propose a method to regularize feature superposition by encoding representations of multiple features within a single neuron. |
| Outcome: | The proposed method improves model interpretability without compromising prediction performance. |
GPT-4V Cannot Generate Radiology Reports Yet (2025.findings-naacl)
Copied to clipboard
| Challenge: | Large language models (LLMs) are becoming multimodal, and GPT-4 models are supposed to possess advanced skills across a wide range of domains, including high-stakes scenarios such as medicine. |
| Approach: | They perform a systematic evaluation of GPT-4 in generating radiology reports across three chest X-ray report benchmarks: MIMIC-CXR, CheXpert Plus, and IU X ray. |
| Outcome: | The proposed model fails in lexical and clinical efficacy metrics . the distributions of model-predicted labels remain constant regardless of groundtruth conditions on the image, suggesting that the model is not interpreting chest X-rays meaningfully. |
Is Semantic Chunking Worth the Computational Cost? (2025.findings-naacl)
Copied to clipboard
| Challenge: | Recent advances in Retrieval-Augmented Generation (RAG) systems have popularized semantic chunking. |
| Approach: | They evaluate the effectiveness of semantic chunking using three common retrieval tasks . they find that the computational costs associated with semantic chunks are not justified by consistent performance gains. |
| Outcome: | The proposed semantic chunking approach is not able to deliver consistent performance gains in three retrieval-related tasks. |
On Using Arabic Language Dialects in Recommendation Systems (2025.findings-naacl)
Copied to clipboard
| Challenge: | Using natural language processing (NLP) to analyze user reviews in recommendation systems is unexplored. |
| Approach: | They propose to integrate Arabic dialects as a signal in recommendation systems by using explicit and implicit approaches. |
| Outcome: | The proposed approach improves recommendation performance and encourages further research in the Arab multicultural world. |
Assessing LLMs for Zero-shot Abstractive Summarization Through the Lens of Relevance Paraphrasing (2025.findings-naacl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have achieved state-of-the-art performance at zero-shot summarization of abstractive summaries for given articles, but little is known about their robustness at this task. |
| Approach: | They propose a strategy that uses the most relevant sentences to generate an ideal summary and then paraphrases them to obtain a minimally perturbed dataset. |
| Outcome: | The proposed approach can be used to measure the robustness of LLMs as summarizers on a minimally perturbed dataset. |
Beyond Silent Letters: Amplifying LLMs in Emotion Recognition with Vocal Nuances (2025.findings-naacl)
Copied to clipboard
| Challenge: | Recent studies have demonstrated that Large Language Models possess a form of emotional intelligence, capable of interpreting emotional stimuli in text. |
| Approach: | They propose a method that translates speech characteristics into natural language descriptions and integrates them into LLMs to perform multimodal emotion analysis via text prompts. |
| Outcome: | The proposed method outperforms baseline models that require structural modifications on two datasets showing significant improvements in emotion recognition accuracy. |
DomainSum: A Hierarchical Benchmark for Fine-Grained Domain Shift in Abstractive Text Summarization (2025.findings-naacl)
Copied to clipboard
| Challenge: | Abstractive summarization is a crucial task in natural language processing . current research focuses on summarizing specific types of documents . domain shifts between documents affect summarisation performance . |
| Approach: | They propose a hierarchical benchmark to capture fine-grained domain shifts in abstractive summarization. |
| Outcome: | The proposed benchmark measures the generalization capabilities of pre-trained language models and large language models in in-domain and cross-domain settings. |
Test-time Backdoor Mitigation for Black-Box Large Language Models with Defensive Demonstrations (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing studies on backdoor defense have focused on training phase, overlooking critical aspect of testing time defense. |
| Approach: | They propose to use demonstrations as a defense mechanism against backdoor attacks in black-box LLMs. |
| Outcome: | The proposed method outperforms existing defense baselines across most evaluation scenarios. |
“All that Glitters”: Techniques for Evaluations with Unreliable Model and Human Annotations (2025.findings-naacl)
Copied to clipboard
| Challenge: | Using standard metrics in the presence of poor labels masks label and model quality . evaluation techniques accounting for unreliable labels reveal important flaws, including spurious correlations and nonrandom racial biases . |
| Approach: | They analyze human labels, GPT model ratings, and transformer encoder model ratings . they show that standard metrics in the presence of poor labels mask label and model quality . |
| Outcome: | The proposed methods mask label and model quality even in the presence of poor models. |
KwaiChat: A Large-Scale Video-Driven Multilingual Mixed-Type Dialogue Corpus (2025.findings-naacl)
Copied to clipboard
Xiaoming Shi, Zeming Liu, Yiming Lei, Chenkai Zhang, Haitao Leng, Chuan Wang, Qingjie Liu, Wanxiang Che, Yunhong Wang
| Challenge: | Currently, video-based dialogue systems rely on a single dialogue type, hindering their versatility in practical applications. |
| Approach: | They propose to generate video-driven multilingual mixed-type dialogues using KwaiChat . they propose to create a video-based multilingual mix of 4 dialogue types, 30 domains, 4 languages, 13 topics . |
| Outcome: | The proposed model performs best on KwaiChat but is not perfect in this situation. |
GenEOL: Harnessing the Generative Power of LLMs for Training-Free Sentence Embeddings (2025.findings-naacl)
Copied to clipboard
| Challenge: | Training-free embedding methods focus on optimizing embeddable prompts . previous methods have overlooked the benefits of utilizing generative abilities of LLMs - GenEOL . |
| Approach: | They propose a method that leverages pretrained large language models to embed text . they propose generating diverse transformations of a sentence that preserve its meaning . |
| Outcome: | The proposed method outperforms existing training-free embedding methods by 2.85 points on the sentence semantic text similarity (STS) benchmark. |
Attention Tracker: Detecting Prompt Injection Attacks in LLMs (2025.findings-naacl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) are vulnerable to prompt injection attacks, where malicious inputs manipulate the model into ignoring original instructions and executing designated actions. |
| Approach: | They propose a training-free method that tracks attention patterns on instruction to detect prompt injection attacks without additional inference. |
| Outcome: | The proposed method shows an improvement of up to 10.0% over existing methods and performs well even on small LLMs. |
Unsupervised Speech-text word-level alignment with Dynamic Programming (2025.findings-naacl)
Copied to clipboard
| Challenge: | Word-level alignment in speech-text pretraining models is limited by word-level annotated data . authors propose an iterative training method for USDP that reduces the dependency on scarce annotation resources. |
| Approach: | They propose an Unsupervised Speech-text word-level alignment with Dynamic Programming (USDP) this method uses Dynamic programming principles to iteratively refine temporal alignment predictions . |
| Outcome: | The proposed method significantly improves on speech-text pretraining tasks compared to existing methods. |
SciAssess: Benchmarking LLM Proficiency in Scientific Literature Analysis (2025.findings-naacl)
Copied to clipboard
Hengxing Cai, Xiaochen Cai, Junhan Chang, Sihang Li, Lin Yao, Wang Changxin, Zhifeng Gao, Hongshuai Wang, Li Yongge, Mujie Lin, Shuwen Yang, Jiankun Wang, Mingjun Xu, Jin Huang, Xi Fang, Jiaxi Zhuang, Yuqi Yin, Yaqi Li, Changhong Chen, Zheng Cheng, Zifeng Zhao, Linfeng Zhang, Guolin Ke
| Challenge: | Existing benchmarks fail to adequately evaluate the proficiency of Large Language Models (LLMs) Existing standards do not cover the skills needed to evaluate LLMs in scientific literature analysis. |
| Approach: | They propose a benchmark to evaluate the proficiency of large language models in scientific literature analysis. |
| Outcome: | SciAssess evaluates 11 LLMs on multiple tasks across scientific fields. |
Towards Understanding the Fragility of Multilingual LLMs against Fine-Tuning Attacks (2025.findings-naacl)
Copied to clipboard
| Challenge: | Recent advances in Large Language Models have sparked concerns about their safety. |
| Approach: | They propose a method to identify safety-related information in the model parameter space . they propose to use a few adversarially chosen examples to fine-tune LLMs . |
| Outcome: | The proposed method can break safety alignment in multilingual LLMs using a few examples . it also shows that the proposed method jailbreaks LLM models adapted to new languages . |
MASSW: A New Dataset and Benchmark Tasks for AI-Assisted Scientific Workflows (2025.findings-naacl)
Copied to clipboard
Xingjian Zhang, Yutong Xie, Jin Huang, Jinge Ma, Zhaoying Pan, Qijia Liu, Ziyang Xiong, Tolga Ergen, Dongsub Shim, Honglak Lee, Qiaozhu Mei
| Challenge: | Scientific innovation is driven by detailed workflows, which include critical steps such as contextualizing literature, generating ideas, validating ideas, and planning new research. |
| Approach: | They propose to use large language models to extract five key aspects from scientific publications to optimize scientific workflows. |
| Outcome: | The proposed dataset includes more than 152,000 peer-reviewed publications from 17 leading computer science conferences spanning the past 50 years. |
Neuro-symbolic Training for Reasoning over Spatial Language (2025.findings-naacl)
Copied to clipboard
| Challenge: | Spatial reasoning is essential for everyday human tasks and is crucial for robots to interact with their environment in a human-like manner. |
| Approach: | They propose to train language models to adhere to spatial reasoning rules as constraints . this allows them to capture the necessary level of abstraction for spatial reasoning . |
| Outcome: | The proposed technique improves language models in multi-hop spatial reasoning over text . it achieves higher accuracy than other competitive Spatial Question-answering benchmarks . |
On Localizing and Deleting Toxic Memories in Large Language Models (2025.findings-naacl)
Copied to clipboard
Anubrata Das, Manoj Kumar, Ninareh Mehrabi, Anil Ramakrishna, Anna Rumshisky, Kai-Wei Chang, Aram Galstyan, Morteza Ziyadi, Rahul Gupta
| Challenge: | Existing methods to reduce toxic generation in large language models are not fully understood. |
| Approach: | They propose to understand the mechanisms that drive toxic generation in large language models by using memory localization to reduce toxic generation. |
| Outcome: | The proposed method reduces toxic generation from 62.86% to 28.61%, but it also improves generation quality. |
DiVISe: Direct Visual-Input Speech Synthesis Preserving Speaker Characteristics And Intelligibility (2025.findings-naacl)
Copied to clipboard
| Challenge: | Video-to-speech (V2S) synthesis requires acoustic hints to accurately reconstruct both speech content and speaker characteristics from video clips alone. |
| Approach: | They propose a video-to-speech (V2S) model that predicts Mel-spectrograms directly from video frames. |
| Outcome: | The proposed model outperforms existing models in acoustic intelligibility and preserves speaker-specific characteristics. |
GraphICL: Unlocking Graph Learning Potential in LLMs through Structured Prompt Design (2025.findings-naacl)
Copied to clipboard
| Challenge: | Text-Attributed Graphs (TAGs) are a powerful tool for understanding complex systems and relationships. |
| Approach: | They propose a benchmark to evaluate large language models for graph-structured data using prompts. |
| Outcome: | The proposed benchmark outperforms state-of-the-art graph LLMs and graph neural networks on graph learning tasks without training. |
FIDELITY: Fine-grained Interpretable Distillation for Effective Language Insights and Topic Yielding (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing methods for topic modeling generate contextually specific and semantically intuitive topics, especially in dynamic environments and low-resource languages. |
| Approach: | They propose a hybrid method that combines topic modeling and text summarization to produce fine-grained, semantically rich, and contextually relevant output. |
| Outcome: | FIDELITY outperforms traditional models in topic diversity, similarity, and ability to process new, unseen documents. |
Classic4Children: Adapting Chinese Literary Classics for Children with Large Language Model (2025.findings-naacl)
Copied to clipboard
| Challenge: | Recent large language models (LLMs) overlook children’s reading preferences, which poses challenges in CLA. |
| Approach: | They propose a method that augments large language models with children's reading preferences for adaptation by obtaining characters' personalities and narrative structure as additional information for fine-grained instruction tuning. |
| Outcome: | The proposed method significantly improves performance in automatic and human evaluation. |
Considering Length Diversity in Retrieval-Augmented Summarization (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing methods that require exhaustive exemplar-exemplar relevance comparisons do not consider summary lengths. |
| Approach: | They propose a Diverse Length-aware Maximal Marginal Relevance algorithm to better control summary lengths. |
| Outcome: | The proposed algorithm reduces the computational cost and memory consumption while maintaining the same level of informativeness. |
LMOD: A Large Multimodal Ophthalmology Dataset and Benchmark for Large Vision-Language Models (2025.findings-naacl)
Copied to clipboard
Zhenyue Qin, Yu Yin, Dylan Campbell, Xuansheng Wu, Ke Zou, Ninghao Liu, Yih Chung Tham, Xiuzhen Zhang, Qingyu Chen
| Challenge: | Existing benchmarks for large vision-language models (LVLMs) are limited to ophthalmology-specific applications. |
| Approach: | They introduce a large-scale multimodal ophthalmology benchmark consisting of 21,993 instances across five ocular imaging modalities and 13 state-of-the-art LVLM representatives from closed-source, open-source and medical domains. |
| Outcome: | The proposed model shows significant performance drop in ophthalmology compared to other domains. |
Syntriever: How to Train Your Retriever with Synthetic Data from LLMs (2025.findings-naacl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have become a core technology in various NLP applications such as chatbots and coding assistants. |
| Approach: | They propose a training framework for retrievers using synthetic data from black-box LLMs. |
| Outcome: | The proposed training framework achieves state-of-the-art on benchmark datasets from various domains in nDCG@K. |
DynClean: Training Dynamics-based Label Cleaning for Distantly-Supervised Named Entity Recognition (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing methods to identify entities using distant annotations are expensive and time-consuming. |
| Approach: | They propose a training dynamics-based label cleaning approach to characterize distant annotations and an automatic threshold estimation strategy to locate errors in distant labels. |
| Outcome: | The proposed method outperforms several advanced DS-NER approaches across four datasets. |
An Efficient Rehearsal Scheme for Catastrophic Forgetting Mitigation during Multi-stage Fine-tuning (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing approaches to fine-tune foundational models on new tasks or domains are costly and time-consuming. |
| Approach: | They propose a sampling scheme that prioritizes rehearsal of "collateral damage" samples . the scheme is computationally efficient and easy to implement, they say . |
| Outcome: | a new approach prioritizes rehearsal of “collateral damage” samples outperforms other continual learning methods. |
COAST: Enhancing the Code Debugging Ability of LLMs through Communicative Agent Based Data Synthesis (2025.findings-naacl)
Copied to clipboard
Weiqing Yang, Hanbin Wang, Zhenghao Liu, Xinze Li, Yukun Yan, Shuo Wang, Yu Gu, Minghe Yu, Zhiyuan Liu, Ge Yu
| Challenge: | Existing code debugging benchmarks focus on the Code Repair stage of the code generation process. |
| Approach: | They propose a framework to evaluate the debugging abilities of large language models by emulating the human debug process. |
| Outcome: | The proposed framework outperforms human-curated and GPT-4-generated training data, enabling 7B-scale LLMs to achieve comparable debugging performance to GPT-3.5. |
Chain-of-Probe: Examining the Necessity and Accuracy of CoT Step-by-Step (2025.findings-naacl)
Copied to clipboard
Zezhong Wang, Xingshan Zeng, Weiwen Liu, Yufei Wang, Liangyou Li, Yasheng Wang, Lifeng Shang, Xin Jiang, Qun Liu, Kam-Fai Wong
| Challenge: | Current research found the issue of Early Answering in large language models where the models already have an answer before generating the Chain-of-Thought (CoT). |
| Approach: | They propose a method to probe changes in confidence during the model’s reasoning and prioritize answers with correct reasoning among multiple candidates. |
| Outcome: | The proposed method reveals that in a significant number of question-answer cases, CoT appears to be unnecessary and this necessity correlates with the simplicity of the task, defined by the reasoning steps required. |
INDIC QA BENCHMARK: A Multilingual Benchmark to Evaluate Question Answering capability of LLMs for Indic Languages (2025.findings-naacl)
Copied to clipboard
Abhishek Kumar Singh, Vishwajeet Kumar, Rudra Murthy, Jaydeep Sen, Ashish Mittal, Ganesh Ramakrishnan
| Challenge: | Large Language Models perform well on unseen tasks in English, but their abilities in non-English languages are less explored due to limited benchmarks and training data. |
| Approach: | They propose to release a large dataset for context-grounded question answering in 11 major Indian languages. |
| Outcome: | The Indic-QA Benchmark compared large datasets of large LLMs on extractive and abstractive tasks in 11 major Indian languages. |
Learning with Less: Knowledge Distillation from Large Language Models via Unlabeled Data (2025.findings-naacl)
Copied to clipboard
Juanhui Li, Sreyashi Nag, Hui Liu, Xianfeng Tang, Sheikh Muhammad Sarwar, Limeng Cui, Hansu Gu, Suhang Wang, Qi He, Jiliang Tang
| Challenge: | Large Language Models (LLMs) have demonstrated superior language understanding abilities in many real-world NLP applications. |
| Approach: | They propose a learning-based sample selection method that incorporates signals from both teacher and student to enhance model performance. |
| Outcome: | The proposed method improves model performance across datasets with higher data efficiency. |
LSDC: An Efficient and Effective Large-Scale Data Compression Method for Supervised Fine-tuning of Large Language Models (2025.findings-naacl)
Copied to clipboard
Zhaoguang Long, Yuhao Zhou, Shangqing Zhao, Yupei Ren, Li Cai, Chenghao Jia, Zhe Chen, Zhe Fang, Yuxiang Song, Man Lan
| Challenge: | Large Language Models (LLMs) are expanding in scale and size, increasing computational costs . large-scale data compression techniques can reduce the size of training datasets while maintaining data integrity. |
| Approach: | They propose a large-scale data compression method to reduce the size of training data . they use a bifurcated quantization strategy to maximize the diversity of samples . |
| Outcome: | The proposed method significantly reduces the size of training data while maximizing the submodular gain. |
What Is Missing in Multilingual Visual Reasoning and How to Fix It (2025.findings-naacl)
Copied to clipboard
| Challenge: | NLP models today strive for supporting multiple languages and modalities, improving accessibility for diverse users. |
| Approach: | They propose a translation-test approach to tackle multilinguality, visual programming approach to break down complex reasoning, and a method that leverages image captioning to address multimodality. |
| Outcome: | The proposed interventions boost open models LLaVA-v1.5-13B by 13.4%, LLva-v1.6-34B by 20.3%, and Qwen-VL by 16.7% while minorly improving GPT-4V’s performance. |
Enhancing the Prototype Network with Local-to-Global Optimization for Few-Shot Relation Extraction (2025.findings-naacl)
Copied to clipboard
| Challenge: | Relation Extraction (RE) is a task that aims to extract semantic relationships from unstructured text. |
| Approach: | They propose a local optimization strategy that indirectly optimizes the prototypical networks by optimizing the other information contained within the prototypes. |
| Outcome: | The proposed model improves on the FewRel 1.0 and FewRela 2.0 datasets. |
LLMs for Mathematical Modeling: Towards Bridging the Gap between Natural and Mathematical Languages (2025.findings-naacl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have demonstrated strong performance across various natural language processing tasks, but their proficiency in mathematical reasoning remains a key challenge. |
| Approach: | They propose a process-oriented framework to evaluate LLMs' ability to construct mathematical models, using solvers to compare outputs with ground truth. |
| Outcome: | The proposed framework evaluates LLMs' ability to construct mathematical models, using solvers to compare outputs with ground truth. |
Advancing Persian LLM Evaluation (2025.findings-naacl)
Copied to clipboard
Sara Bourbour Hosseinbeigi, Behnam Rohani, Mostafa Masoudi, Mehrnoush Shamsfard, Zahra Saaberi, Mostafa Karimi Manesh, Mohammad Amin Abbasi
| Challenge: | Existing evaluation approaches for large language models in low-resource languages like Persian lack comprehensive frameworks, limiting their ability to assess models’ performance over a wide range of tasks requiring considerable cultural and contextual knowledge. |
| Approach: | They propose to provide two new benchmarks to assess models' performance over a wide range of tasks requiring considerable cultural and contextual knowledge. |
| Outcome: | The proposed benchmarks challenge the current state-of-the-art models’ abilities in a variety of Persian language comprehension tasks while reducing data contamination while providing an accurate assessment of Persian LLMs. |
Supportiveness-based Knowledge Rewriting for Retrieval-augmented Language Modeling (2025.findings-naacl)
Copied to clipboard
| Challenge: | Recent advances in large language models (LLMs) have significantly enhanced their performance in various natural language processing tasks. |
| Approach: | They propose a robust and pluggable knowledge rewriter that is optimized for LLM generation by supporting the model's supportiveness. |
| Outcome: | The proposed model can be used to rewrite knowledge in a supervised manner. |
Evaluating Self-Generated Documents for Enhancing Retrieval-Augmented Generation with Large Language Models (2025.findings-naacl)
Copied to clipboard
| Challenge: | Recent research focuses on optimizing the use of Self-Docs with their inherent properties remaining underexplored. |
| Approach: | They develop a taxonomy to compare the effectiveness of different types of Self-Docs and explore strategies for combining them with external sources. |
| Outcome: | The proposed model can supplement retrieved content and provide a powerful way to improve knowledge-intensive question answering tasks. |
PREMISE: Matching-based Prediction for Accurate Review Recommendation (2025.findings-naacl)
Copied to clipboard
| Challenge: | Experimental results show that PREMISE achieves promising performance with less computational cost. |
| Approach: | They propose a new architecture for matching-based learning in multimodal fields for the MRHP task. |
| Outcome: | The proposed architecture significantly boosts performance on multimodal tasks with less computational cost compared to the state-of-the-art fusion-based methods. |
Semi-supervised Fine-tuning for Large Language Models (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing LLMs require labeled data, which can be costly in real-world applications. |
| Approach: | They propose a framework that can fully exploit labeled and unlabeled data for LLM fine-tuning . they conducted experiments using GPT-4o-mini and Llama-3.1 on seven general or domain-specific datasets . |
| Outcome: | The proposed framework can fully exploit labeled and unlabeled data for LLM alignment from a propagate-and-select manner. |
CALM: Unleashing the Cross-Lingual Self-Aligning Ability of Language Model Question Answering (2025.findings-naacl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) are pre-trained on extensive multilingual corpora to acquire both language-specific cultural knowledge and general knowledge. |
| Approach: | They propose to use the **C**ross-Lingual Self-**Aligning ability of **L**anguage **M**odels to align knowledge across languages. |
| Outcome: | The proposed model performs well in both zero-shot and retrieval-augmented settings. |
Towards Prompt Generalization: Grammar-aware Cross-Prompt Automated Essay Scoring (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing approaches to score essays on unseen prompts are challenging to use in educational situations. |
| Approach: | They propose a grammar-aware cross-prompt trait scoring model which internally captures prompt-independent syntactic aspects to learn generic essay representation. |
| Outcome: | Empirical results show that the proposed model improves prompt-independent and grammar-related traits and achieves notable QWK gains in the most challenging cross-prompt scenario. |
MedEureka: A Medical Domain Benchmark for Multi-Granularity and Multi-Data-Type Embedding-Based Retrieval (2025.findings-naacl)
Copied to clipboard
| Challenge: | Embedding-based retrieval (EBR) is a mainstream approach in information retrieval. |
| Approach: | They propose an enriched benchmark to evaluate retrieval capabilities of embedding models . they use four levels of granularity and six types of medical texts to prompt instruction-fine-tuned embeddable models. |
| Outcome: | The proposed benchmark evaluates the retrieval capabilities of embedding models with multi-granularity and multi-data types. |
A Federated Framework for LLM-based Recommendation (2025.findings-naacl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have demonstrated potential in building generative recommendation systems through fine-tuning user behavior data. |
| Approach: | They propose a federated framework for LLM-based recommendation that combines dynamic parameter aggregation and learning speed for different clients. |
| Outcome: | The proposed framework achieves a more balanced client performance and improved overall performance in a computational and storage-efficient way while safeguarding user privacy well. |
WaterSeeker: Pioneering Efficient Detection of Watermarked Segments in Large Documents (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing methods focus on distinguishing fully watermarked text from non-watermarked text, overlooking real-world scenarios where LLMs generate only brief segments within longer documents. |
| Approach: | They propose a method to detect watermarked segments in large documents using an anomaly extraction method and a local traversal. |
| Outcome: | The proposed method achieves a superior balance between detection accuracy and computational efficiency. |
MIRAGE: A Metric-Intensive Benchmark for Retrieval-Augmented Generation Evaluation (2025.findings-naacl)
Copied to clipboard
| Challenge: | Retrieval-Augmented Generation (RAG) systems are limited in their evaluation due to the intricate interplay between retrieval and generation components. |
| Approach: | They propose a Question Answering Question Answerer dataset specifically designed for RAG evaluation that integrates external, non-parametric knowledge retrieved by a retrieval pool of 37,800 entries. |
| Outcome: | The proposed dataset consists of 7,560 curated instances mapped to a retrieval pool of 37,800 entries, enabling an efficient evaluation of both retrieval and generation tasks. |
FIRE: Fact-checking with Iterative Retrieval and Verification (2025.findings-naacl)
Copied to clipboard
Zhuohan Xie, Rui Xing, Yuxia Wang, Jiahui Geng, Hasan Iqbal, Dhruv Sahnan, Iryna Gurevych, Preslav Nakov
| Challenge: | Fact-checking long-form text is challenging, and breaking it down into multiple atomic claims is not cost-effective. |
| Approach: | They propose a novel agent-based framework that integrates evidence retrieval and claim verification in an iterative manner. |
| Outcome: | The proposed framework reduces large language model (LLM) costs by an average of 7.6 times and search costs by 16.5 times while retaining the same performance. |
Lessons from a User Experience Evaluation of NLP Interfaces (2025.findings-naacl)
Copied to clipboard
| Challenge: | Increasingly, questions are being asked on whether evaluations are reproducible and repeatable. |
| Approach: | They propose to design user interfaces that are more consistent and reproducible . only a minority of published experiments can be reproduced due to non-working code or resource limits . |
| Outcome: | The proposed UIs are based on standardized human-centered interaction principles and are evaluated by four experts. |
TrendSim: Simulating Trending Topics in Social Media Under Poisoning Attacks with LLM-based Multi-agent System (2025.findings-naacl)
Copied to clipboard
Zeyu Zhang, Jianxun Lian, Chen Ma, Yaning Qu, Ye Luo, Lei Wang, Rui Li, Xu Chen, Yankai Lin, Le Wu, Xing Xie, Ji-Rong Wen
| Challenge: | Trending topics bring in a new channel for poisoning attacks, resulting in negative impacts on society. |
| Approach: | They propose an LLM-based multi-agent system to simulate trending topics in social media . they propose a time-aware interaction mechanism, centralized message dissemination, and an interactive system . |
| Outcome: | The proposed system simulates trending topics under poisoning attacks on social media platforms. |
ASRank: Zero-Shot Re-Ranking with Answer Scent for Document Retrieval (2025.findings-naacl)
Copied to clipboard
| Challenge: | Retrieval-Augmented Generation models fail to rank the most relevant documents at the top . conventional retrieval methods fail to find the most important documents . |
| Approach: | They propose a new method for scoring retrieved documents using zero-shot answer scent based on a pre-trained large language model to compute the likelihood of document-derived answers aligning with the answer scent. |
| Outcome: | The proposed method improves top-1 retrieval accuracy on NQ, TriviaQA, WebQA, ArchivalQA, HotpotQA, and Entity Questions. |
DSQG-Syn: Synthesizing High-quality Data for Text-to-SQL Parsing by Domain Specific Question Generation (2025.findings-naacl)
Copied to clipboard
Shaoming Duan, Youxuan Wu, Chuanyi Liu, Yuhao Zhang, Zirui Wang, Peiyi Han, Shengyuan Yu, Liang Yan, Yingwei Liang
| Challenge: | Existing methods for generating SQL queries using natural language questions produce inconsistent NLQ-SQL pairs. |
| Approach: | They propose a text-to-SQL data synthesis framework that generates domain-relevant questions . they synthesize NLQ-SqL pairs that are domain-specific and intent-consistent . |
| Outcome: | The proposed method outperforms closed-source LLMs on the Text-to-SQL task. |
EgoSpeak: Learning When to Speak for Egocentric Conversational Agents in the Wild (2025.findings-naacl)
Copied to clipboard
Junhyeok Kim, Min Soo Kim, Jiwan Chung, Jungbin Cho, Jisoo Kim, Sungwoong Kim, Gyeongbo Sim, Youngjae Yu
| Challenge: | EgoSpeak predicts when an agent should begin speaking based on egocentric streaming video. |
| Approach: | They propose a framework for real-time speech initiation prediction in egocentric streaming video by modeling the conversation from the camera wearer's first-person perspective. |
| Outcome: | The proposed framework outperforms random and silence-based baselines in real time and highlights the importance of multimodal input and context length in effectively deciding when to speak. |
Plot2Code: A Comprehensive Benchmark for Evaluating Multi-modal Large Language Models in Code Generation from Scientific Plots (2025.findings-naacl)
Copied to clipboard
| Challenge: | Multi-modal Large Language Models have shown remarkable progress in visual contexts, yet their ability to convert visual figures into executable code remains underexplored. |
| Approach: | They propose to use a set of visual coding metrics to assess MLLMs' visual . pass rate, text-match ratio, and GPT-4V rating judgement to assess the quality of generated code and rendered images. |
| Outcome: | The proposed benchmark includes 132 high-quality matplotlib plots across six plot types, as well as 150 and 86 plots from Python’s and R’s plotly libraries respectively, totaling 368 plots. |
FunnelRAG: A Coarse-to-Fine Progressive Retrieval Paradigm for RAG (2025.findings-naacl)
Copied to clipboard
| Challenge: | Retrieval-Augmented Generation (RAG) is widely adopted in Large Language Models, but is flat and has limitations such as a significant burden on one retriever and constant granularity limits the ceiling of retrieval performance. |
| Approach: | They propose a progressive retrieval paradigm with coarse-to-fine granularity for RAG, termed FunnelRAG, so as to balance effectiveness and efficiency. |
| Outcome: | The proposed paradigm achieves comparable retrieval performance while the time overhead is reduced by nearly 40%. |
The Power of Bullet Lists: A Simple Yet Effective Prompting Approach to Enhancing Spatial Reasoning in Large Language Models (2025.findings-naacl)
Copied to clipboard
| Challenge: | Large language models (LLMs) are currently dominating the field of natural language processing, but spatial reasoning ability is lacking in LLMs. |
| Approach: | They propose a prompting technique that integrates bullet lists, coordinates, and visualizations into the reasoning process and integrates them into planning tasks. |
| Outcome: | The proposed technique boosts LLMs' spatial reasoning abilities compared to previous prompting techniques. |
Overcoming both Domain Shift and Label Shift for Referring Video Segmentation (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing methods to improve the robustness of open-set domain generalization can only recognize seen objects and mark all unseen objects as “unknown” categories . |
| Approach: | They propose a method to make the model maintain good segmentation ability for unknown objects . they propose CLIP-based Reasoning Prompt which can combine text and visual prompts . |
| Outcome: | The proposed method can bridge the gap caused by label shift by combining text and visual prompts to improve text-object matching ability. |
Language Modeling with Editable External Knowledge (2025.findings-naacl)
Copied to clipboard
| Challenge: | Continual learning methods tackle the problem of a changing world by incrementally training on new information. |
| Approach: | They propose a retrieval-augmented generation approach that incrementally deletes or rewrites other entries in the knowledge base each time a document is added. |
| Outcome: | The proposed model improves accuracy relative to conventional retrieval-augmented generation by 7-13% and 6-10% absolute. |
Beyond Excess and Deficiency: Adaptive Length Bias Mitigation in Reward Models for RLHF (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing efforts to mitigate length bias in reward models have decreased accuracy . achieving an automatic proxy that perfectly replicates human judgment is challenging in practice. |
| Approach: | They propose an adaptive approach that dynamically adjusts the influence of response length in reward evaluations according to the context of the query. |
| Outcome: | The proposed approach reduces unnecessary verbosity while improving overall response quality. |
Neuroplasticity and Corruption in Model Mechanisms: A Case Study Of Indirect Object Identification (2025.findings-naacl)
Copied to clipboard
| Challenge: | Recent advances in transformer-based language modelling have garnered attention in widespread applications. |
| Approach: | They investigate the effects of fine-tuning on poisoned data and analyze the changes after retraining a corrupted model on the original dataset and observe neuroplasticity behaviors. |
| Outcome: | The proposed model corruption mechanisms can be generalized to longer epochs and model reforming can be performed on clean datasets. |
VANE-Bench: Video Anomaly Evaluation Benchmark for Conversational LMMs (2025.findings-naacl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have greatly influenced the development of Large Multi-modal Video Models. |
| Approach: | They propose a benchmark to assess the proficiency of Large Multi-modal Video Models (LMMs) in detecting and localizing anomalies and inconsistencies in videos. |
| Outcome: | The proposed benchmark assesses the proficiency of Video-LMMs in detecting and localizing anomalies and inconsistencies in videos. |
Jailbreaking Prompt Attack: A Controllable Adversarial Attack against Diffusion Models (2025.findings-naacl)
Copied to clipboard
| Challenge: | Text-to-image (T2I) models can be used to generate harmful content such as sexually explicit, unfaithful, and misleading or Not-Safe-for-Work (NSFW) images. |
| Approach: | They propose a more practical and universal attack that does not require the presence of a target model. |
| Outcome: | The proposed attack bypasses both text and image safety checkers while preserving high semantic alignment with the target prompt. |
Emo3D: Metric and Benchmarking Dataset for 3D Facial Expression Generation from Emotion Description (2025.findings-naacl)
Copied to clipboard
Mahshid Dehghani, Amirahmad Shafiee, Ali Shafiei, Neda Fallah, Farahmand Alizadeh, Mohammad Mehdi Gholinejad, Hamid Behroozi, Jafar Habibi, Ehsaneddin Asgari
| Challenge: | Existing 3D facial emotion modeling models are constrained by limited emotion classes and insufficient datasets. |
| Approach: | They propose a 3D facial emotion modeling dataset that spans a wide spectrum of human emotions . they use large language models to generate a diverse array of textual descriptions . |
| Outcome: | Emo3D is an extensive dataset that spans human emotions with images and 3D blendshapes. |
Task-wrapped Continual Learning in Task-Oriented Dialogue Systems (2025.findings-naacl)
Copied to clipboard
| Challenge: | Continual learning is vital for task-oriented dialogue systems (ToDs), but its performance is limited by training separate adapters for each task, preventing global knowledge sharing. |
| Approach: | They propose a framework that employs task-wrapped Adapters to learn global and task-specific information through parameter sharing. |
| Outcome: | The proposed framework outperforms AdapterCL in 37 domains while using only 46% of the parameters. |
Untangling Hate Speech Definitions: A Semantic Componential Analysis Across Cultures and Domains (2025.findings-naacl)
Copied to clipboard
| Challenge: | a new framework for analyzing hate speech definitions is proposed to address cultural differences in interpretations . a dataset of 493 definitions from more than 100 cultures is used to analyze hate speech . |
| Approach: | They propose a framework for a cross-cultural and cross-domain analysis of hate speech definitions . they use open-source LLMs to analyze the impact of different definitions on hate speech detection . |
| Outcome: | The proposed framework enables cross-cultural and cross-domain analysis of hate speech definitions . it reveals that many domains borrow definitions from one another without taking into account target culture . |
CodeRAG-Bench: Can Retrieval Augment Code Generation? (2025.findings-naacl)
Copied to clipboard
Zora Zhiruo Wang, Akari Asai, Xinyan Velocity Yu, Frank F. Xu, Yiqing Xie, Graham Neubig, Daniel Fried
| Challenge: | Language models excel at generating code, but many programs are difficult to generate using only parametric knowledge. |
| Approach: | They propose a retrieval-augmented code generation benchmark that provides reproducible evaluations on retrieval and end-to-end code generation performance. |
| Outcome: | The proposed benchmark covers programming, open-domain, and repository-level tasks and provides reproducible evaluations on retrieval and end-to-end code generation performance. |
Multi-Condition Guided Diffusion Network for Multimodal Emotion Recognition in Conversation (2025.findings-naacl)
Copied to clipboard
| Challenge: | Current research emphasizes contextual factors, the speaker’s influence, and extracting complementary information across different modalities. |
| Approach: | They propose a diffusion-based approach to address the challenges posed by redundant information and redundant information at the semantic level while robustly capturing shared semantics. |
| Outcome: | The proposed model outperforms existing state-of-the-art models on two multimodal datasets and is generalizable and effective. |
Thank You, Stingray: Multilingual Large Language Models Can Not (Yet) Disambiguate Cross-Lingual Word Senses (2025.findings-naacl)
Copied to clipboard
Samuel Cahyawijaya, Ruochen Zhang, Jan Christian Blaise Cruz, Holy Lovenia, Elisa Gilbert, Hiroki Nomoto, Alham Fikri Aji
| Challenge: | Existing studies on multilingual large language models have raised concerns about their reliability beyond English. |
| Approach: | They propose a benchmark for cross-lingual sense disambiguation that uses false friends to identify the limitation of cross-linguistic sense disembarrassment in LLMs. |
| Outcome: | The proposed benchmark pinpoints the limitation of cross-lingual sense disambiguation in LLMs by using false friends in four languages. |
Atoxia: Red-teaming Large Language Models with Target Toxic Answers (2025.findings-naacl)
Copied to clipboard
| Challenge: | Large language models (LLMs) are still vulnerable to generation safety vulnerabilities. |
| Approach: | They propose a method that A**tacks LLMs with target "toxi" given a particular harmful answer, the method generates a user query and a misleading answer opening to examine the internal defects of a given LLM. |
| Outcome: | The proposed method detects safety risks in open-source models and state-of-the-art models such as GPT-4o. |
A Practical Method for Generating String Counterfactuals (2025.findings-naacl)
Copied to clipboard
| Challenge: | Interventions targeting the representation space of language models (LMs) have emerged as an effective means to influence model behavior. |
| Approach: | They propose a method to convert representation counterfactuals into string counterf actuals and analyze the linguistic alterations corresponding to the intervention. |
| Outcome: | The proposed method analyzes linguistic alterations corresponding to a given representation space intervention and interprets features utilized to encode a specific concept. |
Probing-RAG: Self-Probing to Guide Language Models in Selective Document Retrieval (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing methods to optimize retrieve-and-generate processes for real-world scenarios may not be optimal for large language models. |
| Approach: | They propose a Probing-RAG which utilizes hidden state representations from the intermediate layers of language models to adaptively determine the necessity of additional retrievals for a given query. |
| Outcome: | The proposed method outperforms previous methods while reducing the number of redundant retrieval steps. |
Extracting Military Event Temporal Relations via Relative Event Time Prediction and Virtual Adversarial Training (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing models for extracting event temporal relations typically compare the relative times of events directly, neglecting the contextual information between event pairs. |
| Approach: | They propose a temporal relationship extraction model based on relative event time prediction and virtual adversarial training, MFRV. |
| Outcome: | The proposed model can capture and infer temporal relationships and can be generalized by generating adversarial samples. |
Unlocking the Planning Capabilities of Large Language Models with Maximum Diversity Fine-tuning (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing studies have shown that LLMs struggle to generate valid plans in the automated planning domain due to weak System 2 competencies. |
| Approach: | They propose a method which uses a random sampling method to select diverse and representative data to enhance sample efficiency and the model’s generalization capability. |
| Outcome: | The proposed method outperforms baseline methods across scales and multiple benchmark domains. |
Continuous Speech Tokenizer in Text To Speech (2025.findings-naacl)
Copied to clipboard
| Challenge: | Autoregressive modeling is a common method for processing language sequences and is effective in token prediction. |
| Approach: | They propose a text-to-speech model based on continuous speech tokens and a continuous tokenizer for speech compression. |
| Outcome: | The proposed model has better continuity and higher estimated Mean Opinion Scores (MoS) this is attributed to better information preservation rate across low and high frequencies in the frequency domain. |
Efficient Annotator Reliability Assessment and Sample Weighting for Knowledge-Based Misinformation Detection on Social Media (2025.findings-naacl)
Copied to clipboard
Owen Cook, Charlie Grimshaw, Ben Peng Wu, Sophie Dillon, Jack Hicks, Luke Jones, Thomas Smith, Matyas Szert, Xingyi Song
| Challenge: | Misinformation spreads rapidly on social media, confusing the truth and targeting potentially vulnerable people. |
| Approach: | They propose to use inter- and intra-annotator agreement to understand the reliability of each annotator and influence the training of large language models based on annotators reliability. |
| Outcome: | The proposed framework utilises inter- and intra-annotator agreement to understand the reliability of each annotator and influence the training of large language models based on annotators reliability. |
Challenges in Trustworthy Human Evaluation of Chatbots (2025.findings-naacl)
Copied to clipboard
| Challenge: | apathetic or adversarial annotators can corrupt the reliability of open leaderboard rankings . human annotation is widely accepted as the gold standard for open-ended text generation tasks . |
| Approach: | They show that bad annotations can corrupt the reliability of open leaderboard rankings . they argue that human annotation is widely accepted as the gold standard . |
| Outcome: | The proposed algorithm can corrupt the reliability of open leaderboard rankings by up to 5 places. |
RATSD: Retrieval Augmented Truthfulness Stance Detection from Social Media Posts Toward Factual Claims (2025.findings-naacl)
Copied to clipboard
| Challenge: | Social media provides a valuable lens for assessing public perceptions and opinions. |
| Approach: | They propose a method that leverages large language models with retrieval-augmented generation to analyze tweets in relation to claims. |
| Outcome: | The proposed method outperforms state-of-the-art methods on a new dataset . it shows that it outperformed existing methods and achieves a significant increase in Macro-F1 score on TSD-CT. |
FACT: Examining the Effectiveness of Iterative Context Rewriting for Multi-fact Retrieval (2025.findings-naacl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) are proficient at retrieving single facts from extended contexts, but struggle with tasks requiring simultaneous retrieval of multiple facts. |
| Approach: | They propose a method that refines context through successive rounds of rewriting to address this problem by finding all Crucial Texts (FACT) |
| Outcome: | The proposed method improves multi-fact retrieval performance across tasks, though improvements are less notable in general-purpose QA scenarios. |
Temporal Working Memory: Query-Guided Segment Refinement for Enhanced Multimodal Understanding (2025.findings-naacl)
Copied to clipboard
Xingjian Diao, Chunhui Zhang, Weiyi Wu, Zhongyu Ouyang, Peijun Qing, Ming Cheng, Soroush Vosoughi, Jiang Gui
| Challenge: | Multimodal foundation models have demonstrated significant success in tasks such as visual captioning, question answering, and image-text retrieval. |
| Approach: | They propose a specialized cognitive module, temporal working memory, which selectively retains task-relevant information across temporal dimensions. |
| Outcome: | The module retains task-relevant information across temporal dimensions, ensuring that critical details are preserved throughout the processing of video and audio content. |
Investigating the Transferability of Code Repair for Low-Resource Programming Languages (2025.findings-naacl)
Copied to clipboard
| Challenge: | Large language models (LLMs) have shown remarkable performance on code generation tasks. |
| Approach: | They investigate the benefits of distilling the ability to repair code for both high and low resource languages to determine if the techniques are also applicable in low resource settings. |
| Outcome: | The proposed techniques are effective in high- and low-resource languages, but weak in low-level languages. |
Multilingual Blending: Large Language Model Safety Alignment Evaluation with Language Mixture (2025.findings-naacl)
Copied to clipboard
| Challenge: | a range of representative Large Language Models have exhibited remarkable generalization capabilities across numerous downstream tasks. |
| Approach: | They propose a query-response scheme to evaluate the safety alignment of LLMs . they found that multilingual query-responding significantly amplifies the detriment of malicious queries . |
| Outcome: | The proposed scheme improves the safety alignment of state-of-the-art LLMs under multilingual conditions. |
Mitigating Hallucinations in Multimodal Spatial Relations through Constraint-Aware Prompting (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing research has explored methods to enhance the performance of large vision-language models in spatial relations. |
| Approach: | They propose a constraint-aware prompting framework to reduce spatial relation hallucinations by incorporating two types of constraints into the prompt. |
| Outcome: | The proposed framework improves on three widely-used spatial relation datasets. |
Concise and Organized Perception Facilitates Reasoning in Large Language Models (2025.findings-naacl)
Copied to clipboard
| Challenge: | Extensive experimental results on several popular logical benchmarks (ProofWriter, PrOntoQA, PrONtoQA-OOD, and FOLIO) and mathematical benchmark (DI-GSM) show that COP significantly outperforms previous state-of-the-art methods. |
| Approach: | They propose a reasoning approach called Concise and Organized Perception (COP) that carefully analyzes the given statements to identify the most pertinent information while eliminating redundancy efficiently. |
| Outcome: | The proposed approach outperforms state-of-the-art methods on several popular logical benchmarks and mathematical benchmarks. |
Verifiable Format Control for Large Language Model Generations (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing methods focus on benchmarking general instruction following while overlooking how to improve specific format following ability for small LLMs. |
| Approach: | They propose to synthesize massive datasets to improve LLMs' format following abilities by using a verifiable format following feature. |
| Outcome: | The proposed method improves the format following ability of small LLMs with about 7B parameters. |
Taxonomy and Analysis of Sensitive User Queries in Generative AI Search System (2025.findings-naacl)
Copied to clipboard
Hwiyeol Jo, Taiwoo Park, Hyunwoo Lee, Nayoung Choi, Changbong Kim, Ohjoon Kwon, Donghyeon Jeon, Eui Hyeon Lee, Kyoungho Shin, Lim Sun Suk, Kyungmi Kim, Lee Jihye, Sun Kim
| Challenge: | generative LLMs have been used by industries for various purposes, but limited resources and limited experience hinder their deployment and maintenance. |
| Approach: | They propose a taxonomy for sensitive search queries and outline approaches to generating generative LLMs. |
| Outcome: | The proposed model can be used to analyze sensitive queries from real users. |
SynGhost: Invisible and Universal Task-agnostic Backdoor Attack via Syntactic Transfer (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing attacks are classified into end-to-end and pre-training types based on the attack phase . Existing backdoor attacks are based upon perplexity, fine-pruning, and maxEntropy. |
| Approach: | They propose an entropy-based poisoning filter that mitigates backdoor attacks . they propose an invisible and universal task-agnostic backdoor attack via syntactic transfer . |
| Outcome: | The proposed attack can transfer backdoors to various downstream tasks while preserving pre-trained language models' pre-training capabilities. |
TESTEVAL: Benchmarking Large Language Models for Test Case Generation (2025.findings-naacl)
Copied to clipboard
Wenhan Wang, Chenyuan Yang, Zhijie Wang, Yuheng Huang, Zhaoyang Chu, Da Song, Lingming Zhang, An Ran Chen, Lei Ma
| Challenge: | Existing methods to generate test cases using large language models are limited in their ability to generate unit test cases. |
| Approach: | They propose a test case generation benchmark that uses large language models to generate unit test cases. |
| Outcome: | The proposed test case generation benchmarks compare LLMs with commercial and open-source LLM platforms and find that they lack the ability to comprehend program logic and execution paths. |
Safe Inputs but Unsafe Output: Benchmarking Cross-modality Safety Alignment of Large Vision-Language Models (2025.findings-naacl)
Copied to clipboard
Siyin Wang, Xingsong Ye, Qinyuan Cheng, Junwen Duan, Shimin Li, Jinlan Fu, Xipeng Qiu, Xuanjing Huang
| Challenge: | Recent studies focus on single-modality threats, but this approach fails to address cross-modal safety alignment. |
| Approach: | They propose a safety alignment challenge to evaluate cross-modality safety alignment . they propose 'Safe Inputs but Unsafe Output' to consider safety of single modalities . |
| Outcome: | The proposed safety alignment challenge examines cases where modalities are safe independently but could lead to unsafe outputs when combined. |
FLEX: A Benchmark for Evaluating Robustness of Fairness in Large Language Models (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing safety evaluations may overlook the inherent weaknesses of Large Language Models, despite their benefits. |
| Approach: | They propose a benchmark to evaluate the robustness of Large Language Models under extreme conditions. |
| Outcome: | The proposed approach evaluates the fairness of large language models under extreme conditions. |
When and How to Augment Your Input: Question Routing Helps Balance the Accuracy and Efficiency of Large Language Models (2025.findings-naacl)
Copied to clipboard
| Challenge: | augmented generation of knowledge-based long-tail questions can be useful for large language models, but can cause significant latency. |
| Approach: | They propose an adaptive question routing framework that uses a query router to augment input to the right time. |
| Outcome: | The proposed framework surpasses existing approaches in accuracy and efficiency on benchmarks such as AmbigNQ, HotpotQA, MMLU-STEM, and PopQA. |
GraPPI: A Retrieve-Divide-Solve GraphRAG Framework for Large-scale Protein-protein Interaction Exploration (2025.findings-naacl)
Copied to clipboard
| Challenge: | Large Language Models and Retrieval-Augmented Generation frameworks have accelerated drug discovery, but integrating models into workflows remains challenging. |
| Approach: | They propose a large-scale knowledge graph-based retrieve-divide-solve agent pipeline RAG framework to support large-level PPI signaling pathway exploration. |
| Outcome: | The proposed framework is based on large-scale knowledge graphs and can be used to analyze protein-protein interactions. |
From Curiosity to Clarity : Exploring the Impact of Consecutive Why-Questions (2025.findings-naacl)
Copied to clipboard
| Challenge: | a recent study has demonstrated the utility of consecutive why-questions in everyday life. |
| Approach: | They used a WHY-Chain dataset to construct a model that asked a why-questions question . they also used objectives that capture the 'consecutive' characteristic of the data . |
| Outcome: | The proposed model performed better on downstream tasks that require commonsense reasoning . the model was validated by ablation studies and the validity check . |
CollabStory: Multi-LLM Collaborative Story Generation and Authorship Analysis (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing studies on LLM-LLM collaboration for open-ended tasks have focused on human-LLm interaction. |
| Approach: | They propose to generate a dataset exclusively for LLMs to explore multi-LLM collaboration scenarios . they extend their authorship-related tasks for multi-llm settings and extend their baselines . |
| Outcome: | The authors extend authorship-related tasks for multi-LLM settings and present baselines for LLM-LLMS collaboration. |
NTSEBENCH: Cognitive Reasoning Benchmark for Vision Language Models (2025.findings-naacl)
Copied to clipboard
| Challenge: | Recent advances in large language models have demonstrated their strong performance on IQ test questions, achieving high scores across many languages. |
| Approach: | They propose a dataset to evaluate cognitive multimodal reasoning and problem-solving skills of large models. |
| Outcome: | The proposed dataset contains 2,728 multiple-choice questions and 4,642 images spanning 26 categories. |
KnowAgent: Knowledge-Augmented Planning for LLM-Based Agents (2025.findings-naacl)
Copied to clipboard
Yuqi Zhu, Shuofei Qiao, Yixin Ou, Shumin Deng, Shiwei Lyu, Yue Shen, Lei Liang, Jinjie Gu, Huajun Chen, Ningyu Zhang
| Challenge: | Large Language Models (LLMs) fail to effectively guide the planning trajectories during task solving and result in planning hallucinations. |
| Approach: | They propose a novel approach to enhance the planning capabilities of large language models by incorporating explicit action knowledge. |
| Outcome: | The proposed approach can achieve comparable or superior performance to existing baselines on HotpotQA and ALFWorld. |
SWITCH: Studying with Teacher for Knowledge Distillation of Large Language Models (2025.findings-naacl)
Copied to clipboard
| Challenge: | Knowledge Distillation (KD) has emerged as a popular method for compressing large language models due to high inference costs and memory requirements. |
| Approach: | They propose a method that integrates the teacher model during the student's sequence generation to reduce misguidance from the teacher. |
| Outcome: | Experiments on three model families and five instruction-following datasets show that SWITCH surpasses traditional methods, especially in the generation of long sequential data. |
Thought2Text: Text Generation from EEG Signal using Large Language Models (LLMs) (2025.findings-naacl)
Copied to clipboard
| Challenge: | Recent advances in NLP driven by powerful Large Language Models such as Ope-nAI GPT-4 have been demonstrated in ALS and stroke patients. |
| Approach: | They propose to use instruction-tuned Large Language Models (LLMs) with EEG data to decode and express brain activity in a comprehensible form. |
| Outcome: | The proposed approach enables multimodal description generation from EEG data and further refinement on embeddings to generate text directly from EMG during inference. |
A Comprehensive Survey of Contemporary Arabic Sentiment Analysis: Methods, Challenges, and Future Directions (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing literature on Arabic sentiment analysis is limited, compared to high-resourced languages such as English and French. |
| Approach: | They present a systematic review of existing literature on Arabic sentiment analysis focusing on research utilizing deep learning. |
| Outcome: | The proposed methods highlight gaps in the literature on Arabic sentiment analysis and outline promising directions for future research. |
Towards Cross-Lingual Explanation of Artwork in Large-scale Vision Language Models (2025.findings-naacl)
Copied to clipboard
| Challenge: | LVLMs are increasingly capable of responding in multiple languages . however, there is a lack of evaluation tools for LVLs that handle multiple languages. |
| Approach: | They used an extended dataset in multiple languages to evaluate LVLMs' ability to generate explanations in multiple language combinations. |
| Outcome: | The proposed dataset in multiple languages evaluates LVLMs' ability to generate explanations in other languages. |
Large Language Models are Easily Confused: A Quantitative Metric, Security Implications and Typological Analysis (2025.findings-naacl)
Copied to clipboard
| Challenge: | Language Confusion is a phenomenon where Large Language Models (LLMs) generate text that is neither in the desired language, nor in a contextually appropriate one. |
| Approach: | They propose a metric to measure and quantify language confusion in Large Language Models (LLMs) they link language confusion to LLM security and find patterns in the case of multilingual embedding inversion attacks. |
| Outcome: | The proposed metric reveals language confusion across LLMs and link it to LLM security and embedding inversion attacks. |
Huatuo-26M, a Large-scale Chinese Medical QA Dataset (2025.findings-naacl)
Copied to clipboard
Xidong Wang, Jianquan Li, Shunian Chen, Yuxuan Zhu, Xiangbo Wu, Zhiyi Zhang, Xiaolong Xu, Junying Chen, Jie Fu, Xiang Wan, Anningzhe Gao, Benyou Wang
| Challenge: | Large Language Models are a powerful tool for medical research, but the data is a bottleneck. |
| Approach: | They propose to use the largest ever medical Question Answering dataset with 26 Million QA pairs as a fine-tuning data for training large language models. |
| Outcome: | The proposed dataset demonstrates that it can be used to train large language models and improves zero-shot performance on other datasets. |
SEP-MLDC: A Simple and Effective Paradigm for Multi-Label Document Classification (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing methods focus on optimizing document features, overlooking the potential of high-quality label features to enhance classification performance. |
| Approach: | They propose a multi-label document classification paradigm that utilizes large language models to expand the label content and generate pseudo-samples for the tail categories. |
| Outcome: | The proposed method significantly outperforms state-of-the-art models. |
Improving Pre-trained Language Models with Knowledge Enhancement and Filtering Framework (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing knowledge enhancement techniques for pre-trained language models (PLMs) introduce noisy entity representations. |
| Approach: | They propose a knowledge enhancement filter that integrates external knowledge bases to enhance PLMs' ability to capture entity knowledge. |
| Outcome: | The proposed method achieves the highest F1-score and accuracy while reducing the computational cost by 1.7-2.5x. |
Using Review Combination and Pseudo-Tokens for Aspect Sentiment Quad Prediction (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing models confuse implicit and explicit sentiment, making it difficult to extract quadruples effectively. |
| Approach: | They propose a framework that leverages distinct labeled features from diverse reviews and incorporates pseudo-token prompts to harness the semantic knowledge of pre-trained models. |
| Outcome: | The proposed framework improves over four public datasets, averaging 1.99% F1 improvement, particularly in instances involving implicit sentiment. |
DDGIP: Radiology Report Generation Through Disease Description Graph and Informed Prompting (2025.findings-naacl)
Copied to clipboard
| Challenge: | Automatic radiology report generation is challenging due to inherent biases in medical imaging data. |
| Approach: | They propose a disease description graph that encapsulates comprehensive and pertinent disease information. |
| Outcome: | The proposed model outperforms state-of-the-art models on two widely-used datasets . the proposed model is based on a three-layer decoder and improves on existing models . |
Lossless Acceleration of Large Language Models with Hierarchical Drafting based on Temporal Locality in Speculative Decoding (2025.findings-naacl)
Copied to clipboard
Sukmin Cho, Sangjin Choi, Taeho Hwang, Jeongyeon Seo, Soyeong Jeong, Huije Lee, Hoyun Song, Jong C. Park, Youngjin Kwon
| Challenge: | Existing methods for drafting and verifying tokens require significant fine-tuning or have inconsistent performance across tasks. |
| Approach: | They propose a lossless drafting approach that organizes various token sources into multiple databases in a hierarchical framework based on temporal locality. |
| Outcome: | The proposed method outperforms existing database drafting methods on Spec-Bench using 7B and 13B parameters. |
Improve Decoding Factuality by Token-wise Cross Layer Entropy of Large Language Models (2025.findings-naacl)
Copied to clipboard
| Challenge: | Large language models (LLMs) often struggle with the issue of generating inaccurate or fabricated content even when they possess correct knowledge. |
| Approach: | They propose a decoding method that mitigates hallucinations without extra training . they propose entropy eNhanced decoding that leverages inner probability changes . |
| Outcome: | The proposed method improves the truthfulness and informativeness of generation while maintaining robust QA accuracy. |
TEaR: Improving LLM-based Machine Translation with Systematic Self-Refinement (2025.findings-naacl)
Copied to clipboard
Zhaopeng Feng, Yan Zhang, Hao Li, Bei Wu, Jiayu Liao, Wenqiang Liu, Jun Lang, Yang Feng, Jian Wu, Zuozhu Liu
| Challenge: | Large Language Models (LLMs) have achieved impressive results in Machine Translation (MT). human evaluations reveal that LLM-generated translations still contain various errors. |
| Approach: | They propose a LLM-based self-refinement framework that feeds error information back into LLMs to facilitate self-finement, leading to enhanced translation quality. |
| Outcome: | The proposed framework outperforms internal refinement and feedback methods while ensuring a robust translation quality baseline. |
Vulnerability of Large Language Models to Output Prefix Jailbreaks: Impact of Positions on Safety (2025.findings-naacl)
Copied to clipboard
| Challenge: | Previous research on jailbreak attacks has focused on optimizing the adversarial snippet content injected into input prompts to expose LLM security vulnerabilities. |
| Approach: | They propose to use a simple adversarial snippet at the beginning of output to expose LLM security vulnerabilities. |
| Outcome: | The proposed approach exposes LLM security vulnerabilities much faster than input suffix attacks or prompt-based output jailbreaks. |
ImaRA: An Imaginative Frame Augmented Method for Low-Resource Multimodal Metaphor Detection and Explanation (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing methods for multimodal metaphor detection neglect cross-domain and attribute similarity characteristics underlying multimodal understanding. |
| Approach: | They propose an Imaginative FRame Augmented method for multimodal metaphor detection and explanation . they use a cross-modal imagination dataset rich in multimodal multimodal expressions . |
| Outcome: | The proposed method outperforms existing methods with training data on two datasets. |
XAMPLER: Learning to Retrieve Cross-Lingual In-Context Examples (2025.findings-naacl)
Copied to clipboard
| Challenge: | XAMPLER: Cross-Lingual Example Retrieval is a cross-lingual example retrieval method . large language models (LLMs) have emerged as effective in-context learning methods . |
| Approach: | They propose a method to train a multilingual model with annotated English examples . they use annotized English data to train the model and use it to train other languages . |
| Outcome: | XAMPLER: Cross-Lingual Example Retrieval improves in-context learning in English . it trains a retriever based on a multilingual small language model using annotated English examples . |
Evaluating Cultural and Social Awareness of LLM Web Agents (2025.findings-naacl)
Copied to clipboard
Haoyi Qiu, Alexander Fabbri, Divyansh Agarwal, Kung-Hsiang Huang, Sarah Tan, Nanyun Peng, Chien-Sheng Wu
| Challenge: | Existing benchmarks often overlook cultural and social awareness . current evaluations focus on task completion, often ignoring the diverse cultural and socio-cultural backgrounds. |
| Approach: | They propose a benchmark to assess LLM agents’ sensitivity to cultural and social norms across two web-based tasks: online shopping and social discussion forums. |
| Outcome: | The proposed framework evaluates LLM agents’ ability to detect and appropriately respond to norm-violating user queries and observations across two web-based tasks. |
GRAIT: Gradient-Driven Refusal-Aware Instruction Tuning for Effective Hallucination Mitigation (2025.findings-naacl)
Copied to clipboard
Runchuan Zhu, Xinke Jiang, Jiang Wu, Zhipeng Ma, Jiahe Song, Fengshuo Bai, Dahua Lin, Lijun Wu, Conghui He
| Challenge: | Experimental evaluations on open-ended and multiple-choice questions demonstrate GRAIT significantly outperforms existing RAIT methods in the overall performance. |
| Approach: | They propose a framework to reduce the risk of over-refusal and reduce hallucinations by rejecting unknown questions to minimize hallucinism and ensuring correct answers are not rejected. |
| Outcome: | The proposed framework outperforms existing methods on open-ended and multiple-choice questions. |
Entity Pair-guided Relation Summarization and Retrieval in LLMs for Document-level Relation Extraction (2025.findings-naacl)
Copied to clipboard
| Challenge: | Document-level relation extraction (DocRE) aims to extract relations between entities in a document. |
| Approach: | They propose an entity pair-guided relation summarization and retrieval model for DocRE . the model uses entity pairs to guide relation summaries and retrievals . |
| Outcome: | The proposed model achieves state-of-the-art (SOTA) performance on three datasets. |
A Recipe of Parallel Corpora Exploitation for Multilingual Large Language Models (2025.findings-naacl)
Copied to clipboard
| Challenge: | Recent studies have highlighted the potential of exploiting parallel corpora to enhance multilingual large language models. |
| Approach: | They investigate the impact of parallel corpora quality and quantity, training objectives, and model size on performance of multilingual large language models enhanced with parallel corporeal. |
| Outcome: | The proposed approach improves performance in bilingual and general-purpose tasks. |
Omni-Chart-600K: A Comprehensive Dataset of Chart Types for Chart Understanding (2025.findings-naacl)
Copied to clipboard
Shulei Wang, Shuai Yang, Wang Lin, Zirun Guo, Sihang Cai, Hai Huang, Ye Wang, Jingyuan Chen, Tao Jin
| Challenge: | Existing chart-related training methods lack capabilities in information extraction, mathematical reasoning, and understanding of multiple chart types. |
| Approach: | They propose a two-stage training strategy and method for jointly training a vision encoder tailored for multi-type charts to address the deficiencies in chart types and limited scope of chart tasks in existing datasets. |
| Outcome: | The proposed dataset includes 21 diverse chart types and tasks, including data retrieval and mathematical reasoning. |
Comprehensive Layer-wise Analysis of SSL Models for Audio Deepfake Detection (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing algorithms for audio deepfake detection are based on layer-wise analysis of self-supervised learning (SSL) models. |
| Approach: | They conduct a layer-wise analysis of self-supervised learning (SSL) models for audio deepfake detection across diverse contexts. |
| Outcome: | The proposed models achieve competitive equal error rate (EER) scores even when employing a reduced number of layers. |
Attention on Multiword Expressions: A Multilingual Study of BERT-based Models with Regard to Idiomaticity and Microsyntax (2025.findings-naacl)
Copied to clipboard
| Challenge: | Specifically, models fine-tuned on semantic tasks tend to distribute attention to idiomatic expressions more evenly across layers. |
| Approach: | They analyze attention patterns of encoder-only models towards two distinct types of Multiword Expressions (MWEs) idioms present challenges in semantic non-compositionality, while MSUs demonstrate unconventional syntactic behavior that does not conform to standard grammatical categorizations. |
| Outcome: | The proposed models show that fine-tuned models allocate attention to idiomatic expressions more evenly across layers. |
Perception Compressor: A Training-Free Prompt Compression Framework in Long Context Scenarios (2025.findings-naacl)
Copied to clipboard
| Challenge: | Long prompts contain redundant information and are sensitive to the position of key information in long context scenarios. |
| Approach: | They propose a training-free prompt compression framework that retains key information at token level while removing distracting tokens. |
| Outcome: | The proposed framework outperforms existing methods on long context benchmarks. |
MojoBench: Language Modeling and Benchmarks for Mojo (2025.findings-naacl)
Copied to clipboard
| Challenge: | Mojo is a programming language that has been praised for its speed and performance over Python. |
| Approach: | They propose a framework for Mojo code generation that evaluates code Large Language Models (LLMs) they propose 'mojo-Coder' which is the first LLM pretrained and fine-tuned for MoJO code generation . |
| Outcome: | MojoBench is the first framework for mojo code generation . it achieves a 30-35% performance improvement over leading models like GPT-4o and Claude-3.5-Sonnet . |
VLind-Bench: Measuring Language Priors in Large Vision-Language Models (2025.findings-naacl)
Copied to clipboard
| Challenge: | Large Vision-Language Models suffer from a problem known as language prior . such language priors can lead to undesirable biases and hallucinations when dealing with images that are out of distribution. |
| Approach: | They propose a benchmark to measure the language priors of Large Vision-Language Models. |
| Outcome: | The proposed benchmark is the first specifically designed to measure the language priors, or blindness, of LVLMs. |
GRAG: Graph Retrieval-Augmented Generation (2025.findings-naacl)
Copied to clipboard
| Challenge: | Naive Retrieval-Augmented Generation (RAG) focuses on individual documents during retrieval and is not suitable for networked documents. |
| Approach: | They propose a novel divide-and-conquer strategy that retrieves optimal subgraph structure in linear time. |
| Outcome: | The proposed approach outperforms current state-of-the-art methods on graph reasoning benchmarks. |
Sequence-level Large Language Model Training with Contrastive Preference Optimization (2025.findings-naacl)
Copied to clipboard
| Challenge: | a new method to improve the performance of large language models requires a small computational cost. |
| Approach: | They propose a CPO procedure that can inject sequence-level information into the model at any training stage without expensive human labeled data. |
| Outcome: | The proposed objective surpasses the next token prediction in terms of win rate in instruction-following and text generation tasks. |
Scaling Up Membership Inference: When and How Attacks Succeed on Large Language Models (2025.findings-naacl)
Copied to clipboard
| Challenge: | Membership inference attacks (MIAs) attempt to verify the membership of a data sample in the training set for a model. |
| Approach: | They propose to use membership inference attacks to verify the membership of a given data sample in a model training set. |
| Outcome: | The proposed methods are based on a new benchmark that measures the performance of membership inference attacks on large language models at a continuous scale. |
Mitigating Hallucinations in Large Vision-Language Models via Summary-Guided Decoding (2025.findings-naacl)
Copied to clipboard
| Challenge: | Large Vision-Language Models (LVLMs) generate detailed and coherent responses from visual inputs but are prone to generate hallucinations due to an over-reliance on language priors. |
| Approach: | They propose a method that reduces the text context and controls only the image-related POS tokens to maintain text quality by reducing the text contextualization. |
| Outcome: | The proposed method achieves state-of-the-art performance on object hallucination benchmarks and achieves Pareto optimality among the existing methods. |
Exploring Hybrid Sampling Inference for Aspect-based Sentiment Analysis (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing methods for inference require multiple sampling with preset size . however, it is a high-cost method that requires multiple sampling . |
| Approach: | They propose a method that combines multiple and single sampling to greatly reduce the cost of multiple sampling without sacrificing performance. |
| Outcome: | The proposed method greatly reduces the cost of multiple sampling without sacrificing performance. |
FeRG-LLM : Feature Engineering by Reason Generation Large Language Models (2025.findings-naacl)
Copied to clipboard
| Challenge: | FeRG-LLM is a large language model that performs feature engineering at an 8billion-parameter scale. |
| Approach: | They propose a framework to perform feature engineering at an 8billion-parameter scale using conversational dialogues. |
| Outcome: | The proposed framework outperforms Llama 3.1 70B and Llma 3.2 on most datasets while using fewer resources and achieving reduced inference time. |
Effective Self-Mining of In-Context Examples for Unsupervised Machine Translation with LLMs (2025.findings-naacl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have demonstrated impressive performance on a wide range of natural language processing tasks. |
| Approach: | They propose an unsupervised approach to mine in-context examples for machine translation (MT) they use word-level mining to acquire word translations that are then used to perform sentence-level mines . |
| Outcome: | The proposed approach outperforms state-of-the-art methods on 288 directions on 287 languages and is based on word-level mining and sentence-level extraction. |
GPT-NER: Named Entity Recognition via Large Language Models (2025.findings-naacl)
Copied to clipboard
Shuhe Wang, Xiaofei Sun, Xiaoya Li, Rongbin Ouyang, Fei Wu, Tianwei Zhang, Jiwei Li, Guoyin Wang, Chen Guo
| Challenge: | Large-scale language models (LLMs) have shown impressive ability for in-context learning with limited training data. |
| Approach: | They propose a novel sequence labeling task that transforms a sequence labeled as a text-generation task into a self-verification task that LLMs can adapt to. |
| Outcome: | The proposed model performs better on NER than supervised models on a variety of tasks . the proposed model can be easily adapted by LLMs to generate a text sequence . |
QPruner: Probabilistic Decision Quantization for Structured Pruning in Large Language Models (2025.findings-naacl)
Copied to clipboard
| Challenge: | Structured pruning can reduce model size but results in significant accuracy degradation . quantization and pruning increase the difficulty of fine-tuning, requiring a more refined quantization scheme. |
| Approach: | They propose a structured pruning framework followed by a layer-wise mixed-precision quantization scheme to reduce model memory consumption during fine-tuning and inference. |
| Outcome: | Experiments on benchmark datasets show that QPruner outperforms existing methods in memory savings while maintaining or improving model performance. |
MES-RAG: Bringing Multi-modal, Entity-Storage, and Secure Enhancements to RAG (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing approaches to retrieve entity information are limited by document level retrieval and intermingled storage of information from different entities. |
| Approach: | They propose a framework that enhances entity-specific query handling . MES-RAG introduces proactive security measures that ensure system integrity . |
| Outcome: | Experimental results show that MES-RAG improves accuracy and recall . the framework can be integrated into existing RAG architectures . |
LVPruning: An Effective yet Simple Language-Guided Vision Token Pruning Approach for Multi-modal Large Language Models (2025.findings-naacl)
Copied to clipboard
| Challenge: | Multi-modal Large Language Models (MLLMs) incur significant computational overhead due to the large number of vision tokens processed, limiting their practicality in resource-constrained environments. |
| Approach: | They propose a language-guided vision token pruning method that can be integrated into existing MLLMs with minimal architectural changes. |
| Outcome: | The proposed method reduces vision tokens by 90% and preserves model performance. |
How Much Knowledge Can You Pack into a LoRA Adapter without Harming LLM? (2025.findings-naacl)
Copied to clipboard
Sergey Pletenev, Maria Marina, Daniil Moskovskiy, Vasily Konovalov, Pavel Braslavski, Alexander Panchenko, Mikhail Salnikov
| Challenge: | Low-rank adaptation (LoRA) is a popular training technique for updating or domain-specific adaptation of Large Language Models (LLMs). |
| Approach: | They propose to use low-rank adaptation to incorporate new facts into the LLM without compromising previously learned knowledge. |
| Outcome: | The proposed approach is harmful because the model's performance declines after such fine-tuning. |
TART: An Open-Source Tool-Augmented Framework for Explainable Table-based Reasoning (2025.findings-naacl)
Copied to clipboard
| Challenge: | Current Large Language Models lack ability to understand table structures and apply precise numerical reasoning. |
| Approach: | They propose a tool-augmented reasoning framework for table-based tasks that integrates LLMs with specialized tools. |
| Outcome: | The proposed framework improves on the TOOLTAB dataset, a benchmark for LLMs in table–tool integration. |
Enhancing Text-to-SQL with Question Classification and Multi-Agent Collaboration (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing research focuses on the optimization of prompts and improvements in workflow, with few studies delving into the exploration of the questions. |
| Approach: | They propose a text-to-SQL framework based on question classification and multi-agent collaboration (QCMA-Sql) they employ multiple cross-attention mechanisms to train a schema selector to classify questions and select the most suitable database schema. |
| Outcome: | The proposed framework outperforms state-of-the-art methods on the Spider dataset and achieves 87.4% execution accuracy. |
Efficient Nearest Neighbor based Uncertainty Estimation for Natural Language Processing Tasks (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing methods for uncertainty estimation are inadequate for safety-critical applications. |
| Approach: | They propose a method that uses the distances from neighbors and the ratio of labels in neighbors to estimate uncertainty. |
| Outcome: | The proposed method outperforms baseline and density-based methods in calibration and uncertainty metrics. |
BitAbuse: A Dataset of Visually Perturbed Texts for Defending Phishing Attacks (2025.findings-naacl)
Copied to clipboard
| Challenge: | Phishing often targets victims through visually perturbed texts to bypass security systems . previous studies have used synthetic datasets that do not contain real-world phishing cases . |
| Approach: | They propose a visual perturbation dataset to target phishing attacks using visual perturbations. |
| Outcome: | The proposed dataset includes real-world phishing cases annotated with visual perturbations . language models trained on the proposed dataset achieved an accuracy of approximately 96% . |
Unfolding the Headline: Iterative Self-Questioning for News Retrieval and Timeline Summarization (2025.findings-naacl)
Copied to clipboard
| Challenge: | a new approach to timeline summarization is proposed for open-domain news content . large language models (LLMs) can be used to extract and organize news events from multiple documents . |
| Approach: | They propose a method to integrate Large Language Models into news timeline summarization by iterating on how events are linked and posing new questions. |
| Outcome: | The proposed system is able to generate and refresh chronological summaries based on documents retrieved in each round. |
RetrieverGuard: Empowering Information Retrieval to Combat LLM-Generated Misinformation (2025.findings-naacl)
Copied to clipboard
| Challenge: | Large language models (LLMs) have demonstrated impressive capabilities in generating human-like text and can store factual knowledge within their extensive parameters. |
| Approach: | They propose a self-supervised training method that captures textual rules and styles of false information from the corpus without human-labelled data. |
| Outcome: | The proposed method can capture rules and styles of false information from the corpus without human-labelled data, achieving higher accuracy and robustness in identifying misleading and highly deceptive AI-generated content. |
Unified Automated Essay Scoring and Grammatical Error Correction (2025.findings-naacl)
Copied to clipboard
| Challenge: | a new study explores the integration of automated writing evaluation and grammatical error correction through multitask learning. |
| Approach: | They propose a system that integrates automated writing evaluation and grammatical error correction through multitask learning by leveraging a shared learning framework. |
| Outcome: | The proposed system outperforms models trained on AWE and GEC, the authors show . their study demonstrates that the proposed system improves writing assessment accuracy and accuracy . |
A Closer Look into Mixture-of-Experts in Large Language Models (2025.findings-naacl)
Copied to clipboard
| Challenge: | Mixture-of-experts (MoE) architectures are gaining increasing attention for their unique properties and remarkable performance. |
| Approach: | They propose a mixture-of-experts architecture that allows for model scaling without sacrificing computational efficiency. |
| Outcome: | The proposed model increases model size without sacrificing computational efficiency . the proposed model is modular and can be used by a broad spectrum of practitioners . |
CDB: A Unified Framework for Hope Speech Detection Through Counterfactual, Desire and Belief (2025.findings-naacl)
Copied to clipboard
Tulio Ferreira Leite Da Silva, Gonzalo Freijedo Aduna, Farah Benamara, Alda Mari, Zongmin Li, Li Yue, Jian Su
| Challenge: | Using algorithms to model user-generated desires on social media, we propose a new approach to understanding and detection of hope speech. |
| Approach: | They propose a language-driven decomposition of the notional category hope and its automatic detection in a unified setting. |
| Outcome: | The proposed model captures future-oriented hopes through desires and beliefs and the counterfactuality of past unfulfilled wishes and regrets. |
How Well Do LLMs Handle Cantonese? Benchmarking Cantonese Capabilities of Large Language Models (2025.findings-naacl)
Copied to clipboard
| Challenge: | Cantonese has scant representation in NLP research, especially compared to other languages from similarly developed regions. |
| Approach: | They propose to evaluate Cantonese LLM performance in factual generation, mathematical logic, complex reasoning, and general knowledge in Cantonesian. |
| Outcome: | The proposed models will evaluate Cantonese's performance in factual generation, mathematical logic, complex reasoning, and general knowledge in Cantone. |
Improving Reward Models with Synthetic Critiques (2025.findings-naacl)
Copied to clipboard
| Challenge: | a recent study shows that reward models overfit on superficial features, hindering generalization performance . prevailing approach to training preference-based reward models presents several challenges . |
| Approach: | They propose a method that uses synthetic natural language critiques to provide additional feedback to large language models. |
| Outcome: | The proposed approach improves performance and data efficiency of RMs initialized from different pretrained models, reducing the reliance on costly human annotations. |
Rethinking Smoothness for Fast and Adaptable Entity Alignment Decoding (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing methods for integrating knowledge graphs rely on entity and relation embeddings . Fig. 1 shows how to decode knowledge graph in under 6 seconds . |
| Approach: | They propose a framework that only utilizes entity embeddings to decode knowledge graphs. |
| Outcome: | The proposed framework reconstructs KG representation by maximizing smoothness of entity embeddings. |
Lost in the Distance: Large Language Models Struggle to Capture Long-Distance Relational Knowledge (2025.findings-naacl)
Copied to clipboard
| Challenge: | Recent large language models have demonstrated impressive capabilities in handling long contexts . however, as context length increases, LLMs struggle more with filtering out irrelevant information . |
| Approach: | They propose to use unrelated sentences to capture relational knowledge over long contexts . they find that LLMs can handle edge noise with little impact, but can reason about distant relationships . |
| Outcome: | The proposed model can handle edge noise with little impact, but its ability to reason about distant relationships declines as the noise grows. |
FinNLI: Novel Dataset for Multi-Genre Financial Natural Language Inference Benchmarking (2025.findings-naacl)
Copied to clipboard
| Challenge: | FinNLI is a benchmark dataset for Financial Natural Language Inference (NLI) across diverse financial texts like SEC Filings, Annual Reports, and Earnings Call transcripts. |
| Approach: | They propose to use FinNLI to evaluate financial natural language inference models across diverse financial texts like SEC Filings, Annual Reports, and Earnings Call transcripts. |
| Outcome: | The proposed dataset is based on a high-quality test set of 3,304 instances annotated by finance experts. |
Music for All: Representational Bias and Cross-Cultural Adaptability of Music Generation Models (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing music generation models are limited in their coverage of the musical genres and cultures of the world. |
| Approach: | They propose to use parametric fine tuning techniques to mitigat the bias in existing music datasets. |
| Outcome: | The proposed models are able to perform well across genres and cultures. |
SFMSS: Service Flow aware Medical Scenario Simulation for Conversational Data Generation (2025.findings-naacl)
Copied to clipboard
| Challenge: | Medical-specific Large Language Models (LLMs) have demonstrated impressive performance on medical-related exams and tasks. |
| Approach: | They propose a framework for medical conversational data generation that uses Authentic Seed Data to ensure quality of the data. |
| Outcome: | The proposed model outperforms all baselines and human evaluations, and aligns with human preferences and clinical demands. |
Re-evaluating Automatic LLM System Ranking for Alignment with Human Preference (2025.findings-naacl)
Copied to clipboard
| Challenge: | Evaluating and ranking the capabilities of different LLMs is crucial for understanding their performance and alignment with human preferences. |
| Approach: | They propose a system-level evaluation framework that ranks LLMs based on their alignment with human preferences. |
| Outcome: | The proposed framework aims to rank LLMs based on their performance and alignment with human preferences. |
GuideQ: Framework for Guided Questioning for progressive informational collection and classification (2025.findings-naacl)
Copied to clipboard
| Challenge: | Using a new multilingual dataset, we examine how LLMs can be used to represent factual knowledge across languages. |
| Approach: | They propose a methodology to measure the extent of representation sharing across languages by repurposing knowledge editing methods. |
| Outcome: | The proposed model can answer a question consistently across languages and can store the answers in a shared representation for several languages. |
Richer Output for Richer Countries: Uncovering Geographical Disparities in Generated Stories and Travel Recommendations (2025.findings-naacl)
Copied to clipboard
| Challenge: | a large body of work examines language models for biases concerning gender, race, occupation and religion . however, the impact of the encoded geographical knowledge on real-world applications has not been documented . |
| Approach: | They examine large language models for two common scenarios that require geographical knowledge: travel recommendations and geo-anchored story generation. |
| Outcome: | The results show that the language models are biased against poorer countries and poorer socioeconomic conditions. |
Swan and ArabicMTEB: Dialect-Aware, Arabic-Centric, Cross-Lingual, and Cross-Cultural Embedding Models and Benchmarks (2025.findings-naacl)
Copied to clipboard
Gagan Bhatia, El Moatez Billah Nagoudi, Abdellah El Mekki, Fakhraddin Alwajih, Muhammad Abdul-Mageed
| Challenge: | In this paper, we introduce a family of embedding models addressing both small-scale and large-scale use cases. |
| Approach: | They propose to use ArabicMTEB to evaluate Arabic text embedding models . they propose to build a benchmark suite that assesses cross-lingual, multi-dialectal, multidomain, and multi-cultural Arabic text embedded models. |
| Outcome: | The proposed models outperform Multilingual-E5-large and Swan-Large in most Arabic tasks while remaining dialectally and culturally aware. |
TAGCOS: Task-agnostic Gradient Clustered Coreset Selection for Instruction Tuning Data (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing methods for instruction tuning are limited due to the increasing volume of instruction datasets and the increased computational costs. |
| Approach: | They propose to extract a small and highly informative subset of training samples from a large dataset that achieves comparable performance to the full dataset. |
| Outcome: | The proposed algorithm outperforms other unsupervised methods and achieves comparable performance to the full dataset. |
From Text to Emoji: How PEFT-Driven Personality Manipulation Unleashes the Emoji Potential in LLMs (2025.findings-naacl)
Copied to clipboard
Navya Jain, Zekun Wu, Cristian Enrique Munoz Villalobos, Airlie Hilliard, Xin Guan, Adriano Koshiyama, Emre Kazim, Philip Colin Treleaven
| Challenge: | Methods like prompt-based In-Context Knowledge Editing and gradient-based Model Editor Networks (MEND) show irregularity and variability; IKE depends on the prompt, leading to variability and sensitivity; MEND yields inconsistent and gibberish outputs. |
| Approach: | They employ Opinion QA Based Parameter-Efficient Fine-Tuning (PEFT) to manipulate the Big Five personality traits: Openness, Conscientiousness, Extraversion, Agreeableness, and Neuroticism. |
| Outcome: | The proposed methods show that they are more accurate than prompt-based IKE and gradient-based MEND outputs. |
Decoding Fatphobia: Examining Anti-Fat and Pro-Thin Bias in AI-Generated Images (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing studies have shown that AI-generated images reinforce social biases, including those related to race and gender. |
| Approach: | They use DALL-E 3 to examine weight bias in AI-generated images . authors discuss findings and their impact on existing research on weight bias . |
| Outcome: | The study examines stereotypical associations between moral character and body weight . it finds that fatphobia is a feature of social systems that rank fatter bodies inferior to thinner bodies . |
MMAU: A Holistic Benchmark of Agent Capabilities Across Diverse Domains (2025.findings-naacl)
Copied to clipboard
Guoli Yin, Haoping Bai, Shuang Ma, Feng Nan, Yanchao Sun, Zhaoyang Xu, Shen Ma, Jiarui Lu, Xiang Kong, Aonan Zhang, Dian Ang Yap, Yizhe Zhang, Karsten Ahnert, Vik Kamath, Mathias Berglund, Dominic Walsh, Tobias Gindele, Juergen Wiest, Zhengfeng Lai, Xiaoming Simon Wang, Jiulong Shan, Meng Cao, Ruoming Pang, Zirui Wang
| Challenge: | Existing benchmarks focus on specific application scenarios, emphasizing task completion but failing to dissect the underlying skills that drive these outcomes. |
| Approach: | They propose a Massive Multitask Agent Understanding benchmark that evaluates LLMs across five domains and offline tasks. |
| Outcome: | The Massive Multitask Agent Understanding (MMAU) benchmark evaluates models across five domains including Tool-use, Directed Acyclic Graph (DAG) QA, Data Science and Machine Learning coding, Contest-level programming and Mathematics. |
Improving Consistency in LLM Inference using Probabilistic Tokenization (2025.findings-naacl)
Copied to clipboard
| Challenge: | Prior work has shown that probabilistic tokenizations can generate multiple tokenization of the same input string. |
| Approach: | They propose a method to leverage the multiple tokenization capabilities of modern LLM tokenizers. |
| Outcome: | The proposed method improves the self-consistency of large language models by generating multiple tokenizations. |
WordGame: Efficient & Effective LLM Jailbreak via Simultaneous Obfuscation in Query and Response (2025.findings-naacl)
Copied to clipboard
| Challenge: | Recent advances in large language models have raised concerns about their susceptibility to jailbreaking attacks, which leads to harmful content inadvertently. |
| Approach: | They propose to exploit the safety alignment patterns of LLMs by simultaneous obfuscation in queries and responses to break down adversarial intent of query. |
| Outcome: | The proposed attack breaks down adversarial intent of query and encourages benign content regarding the games to precede anticipated harmful content in the response. |
Human and LLM-Based Resume Matching: An Observational Study (2025.findings-naacl)
Copied to clipboard
Swanand Vaishampayan, Hunter Leary, Yoseph Berhanu Alebachew, Louis Hickman, Brent A. Stevenor, Weston Beck, Chris Brown
| Challenge: | Resume matching assesses the extent to which candidates qualify for jobs based on the content of resumes. |
| Approach: | They compare GPT-4 and human ratings for resumes submitted to job openings from diverse fields using real-world evaluation criteria. |
| Outcome: | The proposed model improves the quality of LLM ratings and does not show bias. |
A Practical Examination of AI-Generated Text Detectors for Large Language Models (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing methods to detect large language models are prone to misuse, such as generating fake news articles, facilitating academic plagiarism or spamming. |
| Approach: | They evaluate several popular detectors to evaluate their effectiveness against a range of domains, datasets, and models. |
| Outcome: | The proposed methods perform poorly in certain settings, with TPR@.01 as low as 0%. |
Robust Bias Detection in MLMs and its Application to Human Trait Ratings (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing methods to assess demographic bias in MLMs ignore random variability of templates and target concepts, and neglect bias quantification. |
| Approach: | They propose a systematic statistical approach to assess bias in MLMs using mixed models to account for random effects, pseudo-perplexity weights for sentences derived from templates and quantify bias using statistical effect sizes. |
| Outcome: | The proposed method matches on bias scores in magnitude and direction with small to medium effect sizes. |
How Inclusively do LMs Perceive Social and Moral Norms? (2025.findings-naacl)
Copied to clipboard
| Challenge: | Language models (LMs) are used in decision-making systems and as interactive assistants. |
| Approach: | They propose to prompt 11 LMs on rules-of-thumb and compare their outputs with 100 human annotators. |
| Outcome: | The proposed model is compared with 100 human annotators to find out if they are inclusive of diverse human values. |
Jailbreaking with Universal Multi-Prompts (2025.findings-naacl)
Copied to clipboard
| Challenge: | Large language models (LLMs) have seen rapid development in recent years, but ethical concerns and new types of attacks have emerged. |
| Approach: | They propose a prompt-based method to jailbreak large language models using universal multi-prompts and an approach for defense that outperforms existing techniques. |
| Outcome: | The proposed method outperforms existing techniques for jailbreaking LLMs using universal multi-prompts. |
Echoes of Discord: Forecasting Hater Reactions to Counterspeech (2025.findings-naacl)
Copied to clipboard
| Challenge: | Hate speech (HS) online causes increased prejudice and discrimination, fostering an environment of hostility and social division. |
| Approach: | They analyze the Reddit Echoes of Hate dataset to assess the impact of counterspeech from the hater's perspective and focus on whether the counterspeak leads the reentry to be hateful. |
| Outcome: | The proposed model outperforms the two-stage reaction predictor and the three-way classifier to predict haters' reactions to the reentry of the conversation and determines the type of resentment. |
Contextual Metric Meta-Evaluation by Measuring Local Metric Accuracy (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing approaches to metric meta-evaluation focus on general statements about absolute and relative quality of metrics across arbitrary system outputs, but in practice, metrics are applied in highly contextual settings. |
| Approach: | They propose a method for contextual metric meta-evaluation by comparing local metric accuracy. |
| Outcome: | The proposed method compares the local metric accuracy of evaluation metrics across translation, speech recognition, and ranking tasks. |
Advocating Character Error Rate for Multilingual ASR Evaluation (2025.findings-naacl)
Copied to clipboard
| Challenge: | Word error rate (WER) has been used for automatic speech recognition (ASR) evaluations for English datasets for many years. |
| Approach: | They propose to use the character error rate as the primary metric in multilingual ASR evaluation to account for morphologically complex languages. |
| Outcome: | The character error rate (CER) is the primary evaluation metric in multilingual ASR evaluation. |
Enhancing Temporal Understanding in LLMs for Semi-structured Tables (2025.findings-naacl)
Copied to clipboard
| Challenge: | Temporal reasoning over tabular data presents significant challenges for large language models (LLMs), as evidenced by recent research. |
| Approach: | They propose a method that enhances LLMs' temporal reasoning over tabular data by using standard prompts and introduce a novel approach, C.L.E.A.R. |
| Outcome: | The proposed method improves evidence-based reasoning across models and indirect supervision with auxiliary unstructured data significantly boosts model performance in these tasks. |
BnTTS: Few-Shot Speaker Adaptation in Low-Resource Setting (2025.findings-naacl)
Copied to clipboard
Mohammad Jahid Ibna Basher, Md Kowsher, Md Saiful Islam, Rabindra Nath Nandi, Nusrat Jahan Prottasha, Mehadi Hasan Menon, Tareq Al Muntasir, Shammur Absar Chowdhury, Firoj Alam, Niloofar Yousefi, Ozlem Garibay
| Challenge: | Empirical evaluations in few-shot settings show that BnTTS significantly improves the naturalness, intelligibility, and speaker fidelity of synthesized Bangla speech. |
| Approach: | They propose to integrate Bangla into a multilingual TTS pipeline with modifications to account for the phonetic and linguistic characteristics of the language. |
| Outcome: | The proposed framework improves the naturalness, intelligibility, and speaker fidelity of synthesized Bangla speech compared to state-of-the-art systems. |
Playing with Voices: Tabletop Role-Playing Game Recordings as a Diarization Challenge (2025.findings-naacl)
Copied to clipboard
| Challenge: | Using a small dataset, we propose that audio of tabletop role-playing games (TTRPGs) could serve as a challenge for speaker diarization systems. |
| Approach: | They propose that audio of tabletop role-playing games (TTRPGs) could serve as a challenge for speaker diarization systems. |
| Outcome: | The proposed system can pick the speaker and determine that impersonating is just that. |
Causally Testing Gender Bias in LLMs: A Case Study on Occupational Bias (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing studies have shown that large language models can cause harmful, human-like biases against various demographics. |
| Approach: | They propose a causal formulation for bias measurement in generative language models based on a list of desiderata for designing robust bias benchmarks and a bias-measuring procedure to investigate occupational gender bias. |
| Outcome: | The proposed framework is generalizable and can be extended to include other datasets. |
OLMES: A Standard for Language Model Evaluations (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing models claim to perform better on tasks measuring model capabilities, but there is no standard setup for reproducible evaluations. |
| Approach: | They propose a document that is documented and practical for reproducible LLM evaluations and includes recommendations from existing literature and new experiments. |
| Outcome: | The proposed standard identifies and reviews the varying factors in evaluation practices adopted by the community, such as prompt formatting, choice of in-context examples, probability normalizations, and task formulation. |
Induction Heads as an Essential Mechanism for Pattern Matching in In-context Learning (2025.findings-naacl)
Copied to clipboard
| Challenge: | Large language models have shown a remarkable ability to learn and perform complex tasks through in-context learning (ICL). |
| Approach: | They analyse two state-of-the-art models, Llama-3-8B and InternLM2-20B on abstract pattern recognition and NLP tasks. |
| Outcome: | The proposed model can perform up to 32% better than previous models on abstract pattern recognition and NLP tasks. |
MoLA: MoE LoRA with Layer-wise Expert Allocation (2025.findings-naacl)
Copied to clipboard
Chongyang Gao, Kezhen Chen, Jinmeng Rao, Ruibo Liu, Baochen Sun, Yawen Zhang, Daiyi Peng, Xiaoyuan Guo, Vs Subrahmanian
| Challenge: | Recent efforts to integrate low-rank adaptation (LoRA) with the Mixture-of-Experts (MoE) have achieved performance comparable to full-parameter fine-tuning by tuning much fewer parameters. |
| Approach: | They propose a parameter-efficient MoE method for low-rank adaptation with the Mixture-of-Experts (MoE) they use layers of LoRA experts to allocate more LoRA expert to middle layers . |
| Outcome: | The proposed method outperforms baseline models on six well-known NLP and commonsense QA benchmarks on LLAMA-2, Mistral, and Gemma. |
CodeSim: Multi-Agent Code Generation and Problem Solving through Simulation-Driven Planning and Debugging (2025.findings-naacl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have made significant strides in code generation and problem solving. |
| Approach: | They propose a multi-agent code generation framework that integrates human-like perception to address the stages of program synthesis. |
| Outcome: | The proposed framework achieves state-of-the-art (pass@1) results and shows potential for even greater enhancement when cascaded with external debuggers. |
On the Feasibility of In-Context Probing for Data Attribution (2025.findings-naacl)
Copied to clipboard
| Challenge: | In-context probing (ICP) can be used to identify training data that contributes to model outputs, but many data attribution methods, such as influence functions, use model gradients and are computationally expensive. |
| Approach: | They propose to use in-context probing (ICP) to proxy for gradient-based data attribution for data selection under conditions contingent on data similarity. |
| Outcome: | The proposed method can be used to identify training data that contribute to model outputs and fine tune models on training data. |
Evaluation of Multilingual Image Captioning: How far can we get with CLIP models? (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing approaches to evaluate image captions are English-centric, despite improvements in the CLIPScore metric . however, there are no available benchmarks for multilingual captioning evaluation . |
| Approach: | They propose to use machine-translated and machine-repurposed datasets to evaluate CLIPScore variants in multilingual settings. |
| Outcome: | The proposed evaluation strategies are based on machine-translated and human judgements. |
Avoiding Copyright Infringement via Large Language Model Unlearning (2025.findings-naacl)
Copied to clipboard
| Challenge: | Pre-trained Large Language Models (LLMs) have demonstrated remarkable capabilities but also pose significant legal and ethical concerns. |
| Approach: | They propose a framework that unlearns copyrighted content from large language models over multiple time steps by identifying and removing specific weight updates in the model’s parameters that correspond to copyright content. |
| Outcome: | The proposed framework achieves an effective trade-off between unlearning efficacy and general-purpose language abilities, outperforming baselines. |
A Context-Aware Contrastive Learning Framework for Hateful Meme Detection and Segmentation (2025.findings-naacl)
Copied to clipboard
| Challenge: | Empirical experiments show HateSieve surpasses existing LMMs in performance with fewer trainable parameters . |
| Approach: | They propose a framework to enhance detection and segmentation of hateful elements in memes by creating a triplet dataset and an Image-Text Alignment module. |
| Outcome: | HateSieve features a new framework that creates semantically correlated memes and generates contextual embeddings for accurate meme segmentation. |
LLM-Generated Passphrases That Are Secure and Easy to Remember (2025.findings-naacl)
Copied to clipboard
| Challenge: | Automated password generation is a cornerstone of IT security, but passphrases are often hard to remember and see limited adoption. |
| Approach: | They propose to use large language models to generate passphrases with sufficient entropy . they also use prompt construction to generate a grammatical model with sufficient output entropic . |
| Outcome: | The proposed methods show that the outputs are highly reliable and easy to remember . the proposed methods also show that prompt construction is important in steering the output topic and grammatical structure. |
Does Data Contamination Detection Work (Well) for LLMs? A Survey and Evaluation on Detection Assumptions (2025.findings-naacl)
Copied to clipboard
| Challenge: | Large language models have demonstrated great performance across various benchmarks, but data contamination is a concern in their evaluation. |
| Approach: | They analyze 50 papers on data contamination detection and test three of them as case studies to identify the possibility of data contamination. |
| Outcome: | The proposed methods can detect membership inference attacks on instance-level data, and can perform similar to random guessing on LLM pretraining datasets. |
Representation-to-Creativity (R2C): Automated Holistic Scoring Model for Essay Creativity (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing studies on Automated Essay Scoring (AES) are limited. |
| Approach: | They propose a new essay rubric specifically designed for assessing creativity in essays . they use a ground truth data set to construct a self-supervised learning model . |
| Outcome: | The proposed model improves the assessment of creativity in essays by 58% compared to the current models. |
From Single to Multi: How LLMs Hallucinate in Multi-Document Summarization (2025.findings-naacl)
Copied to clipboard
| Challenge: | a recent study investigated hallucinations in multi-document summarization tasks . but, it is unclear how challenges arising from handling multiple documents affect outputs . |
| Approach: | They investigate how hallucinations manifest in large language models when summarizing topic-specific information from a set of documents. |
| Outcome: | The proposed benchmarks show that the models generate more hallucinations than baselines . the results highlight the need for more effective approaches to mitigate hallucinosity in MDS . |
Aligning to Constraints for Data-Efficient Language Model Customization (2025.findings-naacl)
Copied to clipboard
Fei Wang, Chao Shang, Shuai Wang, Sarthak Jain, Qiang Ning, Bonan Min, Vittorio Castelli, Yassine Benajiba, Dan Roth
| Challenge: | General-purpose language models (LMs) are aligned to diverse user intents, but fall short when it comes to specific applications. |
| Approach: | They propose a framework that uses constraints to automatically produce supervision signals for user alignment with constraints. |
| Outcome: | The proposed framework can produce supervision signals for user alignment with constraints. |
Where is this coming from? Making groundedness count in the evaluation of Document VQA models (2025.findings-naacl)
Copied to clipboard
| Challenge: | Document Visual Question Answering (VQA) models have come close to or matching human performance on some benchmarks. |
| Approach: | They propose a method that accounts for the semantic and multimodal groundedness of a model’s outputs and can be parameterized so that users can configure the score according to their preferences. |
| Outcome: | The proposed method produces scores that are a better indicator of a model’s robustness and tends to give higher rewards to better-calibrated answers. |
Transformer-based Causal Language Models Perform Clustering (2025.findings-naacl)
Copied to clipboard
| Challenge: | Recent studies have shown great improvements in instruction-following capability through additional training for instruction- following tasks. |
| Approach: | They propose to use a Transformer-based causal language model to study instruction-following capabilities. |
| Outcome: | The proposed model learns task-specific information by clustering data within its hidden space, with this clustering process evolving dynamically during learning. |
Towards Better Multi-task Learning: A Framework for Optimizing Dataset Combinations in Large Language Models (2025.findings-naacl)
Copied to clipboard
| Challenge: | Using a neural network, large language models can be trained on multiple tasks, allowing them to perform tasks efficiently. |
| Approach: | They propose a framework that leverages a neural network to select the best dataset combinations for enhancing multi-task learning (MTL) They propose to iteratively refine the selection, greatly improving efficiency while being model-, dataset-, and domain-independent. |
| Outcome: | The proposed framework iteratively refines the selection, greatly improving efficiency, while being model-, dataset-, and domain-independent. |
Gender Bias in Instruction-Guided Speech Synthesis Models (2025.findings-naacl)
Copied to clipboard
| Challenge: | Recent advances in controllable expressive speech synthesis have allowed for the generation of speech with specific styles guided by textual descriptions, known as style prompts. |
| Approach: | They examine whether models exhibit gender bias when interpreting occupation-related prompts. |
| Outcome: | The proposed models exhibit gender bias for certain occupations and different sizes show varying degrees of this bias across occupations. |
ResoFilter: Fine-grained Synthetic Data Filtering for Large Language Models through Data-Parameter Resonance Analysis (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing methods for enhancing large language models lack clear metrics for evaluating data characteristics. |
| Approach: | They propose a method that integrates models, data, and tasks to refine datasets. |
| Outcome: | The proposed method achieves comparable results to full-scale fine-tuning using only half the data in mathematical tasks and exhibits strong generalization across different models and domains. |
UCFE: A User-Centric Financial Expertise Benchmark for Large Language Models (2025.findings-naacl)
Copied to clipboard
Yuzhe Yang, Yifei Zhang, Yan Hu, Yilin Guo, Ruoli Gan, Yueru He, Mingcong Lei, Xiao Zhang, Haining Wang, Qianqian Xie, Jimin Huang, Honghai Yu, Benyou Wang
| Challenge: | Recent advances in large language models (LLMs) have expanded their potential applications in finance. |
| Approach: | They propose a framework to evaluate the ability of large language models to handle financial tasks using human expert evaluations and task-specific interactions. |
| Outcome: | The proposed framework evaluates the ability of large language models to handle complex financial tasks and combines human expert evaluations with dynamic, task-specific interactions to simulate the complexities of evolving financial scenarios. |
BRIEF: Bridging Retrieval and Inference for Multi-hop Reasoning via Compression (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing approaches to augment language models with external knowledge but they are limited by static nature of pre-training data. |
| Approach: | They propose a lightweight approach that compresses retrieved documents into highly dense textual summaries to integrate into in-context RAG. |
| Outcome: | The proposed approach reduces latency and costs while achieving high performance in open-domain questions. |
An Optimizable Suffix Is Worth A Thousand Templates: Efficient Black-box Jailbreaking without Affirmative Phrases via LLM as Optimizer (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing jailbreaking methods generate harmful and unethical content when subjected to jailbreaking attacks. |
| Approach: | They propose a black-box jailbreaking method with optimizable suffixes that translate jailbreaking objectives into natural language instructions. |
| Outcome: | The proposed method outperforms existing methods by 2.4 times in the ASR of three open-source LLMs and GPT-3.5-Turbo. |
Multi-Stage LLM Fine-Tuning with a Continual Learning Setting (2025.findings-naacl)
Copied to clipboard
Changhao Guan, Chao Huang, Hongliang Li, You Li, Ning Cheng, Zihe Liu, Yufeng Chen, Jinan Xu, Jian Liu
| Challenge: | Large language models (LLMs) have made significant progress in knowledge-intensive applications, but they may face a multi-stage continuous learning scenario. |
| Approach: | They propose a multi-stage continuous learning paradigm that includes a preference-based learning bias to identify potential knowledge conflicts and a self-distillation-based data augmentation strategy to expand and enrich the training corpus. |
| Outcome: | The proposed learning paradigm achieves a significant improvement in accuracy after 7 stages of fine-tuning compared to previous methods while preserving general knowledge. |
Constraining Sequential Model Editing with Editing Anchor Compression (2025.findings-naacl)
Copied to clipboard
| Challenge: | Large language models (LLMs) exhibit hallucinations due to incorrect or outdated knowledge embedded in their parameters. |
| Approach: | They propose a framework to constrain the deviation of the parameter matrix during sequential editing by selecting editing anchors that are important in encoding new relations without deviating too much from the original matrix. |
| Outcome: | The proposed framework minimizes deviations caused by model editing while retaining over 70% of the general abilities. |
MLKV: Multi-Layer Key-Value Heads for Memory Efficient Transformer Decoding (2025.findings-naacl)
Copied to clipboard
| Challenge: | Multi-Layer Key-Value (MLKV) sharing reduces memory usage by 6x compared to Multi-Query Attention and Grouped-Query Attributes. |
| Approach: | They propose a novel approach that extends KV sharing across transformer layers to reduce memory usage beyond what was possible with Multi-Query Attention and Grouped-Query Attributes. |
| Outcome: | The proposed approach reduces KV cache size by 6x with minimal performance loss and scales linearly with model size, batch size, and sequence length. |
Clarify When Necessary: Resolving Ambiguity Through Interaction with LMs (2025.findings-naacl)
Copied to clipboard
| Challenge: | Ambiguity is embedded throughout natural language, and even simple utterances can have multiple interpretations when read in isolation. |
| Approach: | They propose a task-agnostic framework for evaluating a system’s ability to determine when to ask for clarification. |
| Outcome: | The proposed framework outperforms existing uncertainty estimation approaches at identifying predictions that will benefit from clarification. |
DOLFIN - Document-Level Financial Test-Set for Machine Translation (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing document-level machine translation test-sets cover general domain but fall short on specialised domains, such as legal and financial. |
| Approach: | They propose to use a document-level machine translation test-set to replace perfectly aligned sentences by presenting data in units of sections rather than sentences. |
| Outcome: | The proposed dataset is built from specialised financial documents and it shows that it can discriminate between context-sensitive and context-agnostic models and shows the weaknesses when models fail to accurately translate financial texts. |
Are Large Language Models Effective in Clinical Trial Design? A Study on Baseline Feature Generation (2025.findings-naacl)
Copied to clipboard
Nafis Neehal, Bowen Wang, Shayom Debopadhaya, Corey Curran, Keerthiram Murugesan, Soham Dan, Vibha Anand, Kristin Bennett
| Challenge: | Clinical trials require baseline features to characterize participants and ensure accurate study outcomes. |
| Approach: | They evaluate LLMs' ability to generate appropriate baseline features for clinical trials . they use CT-Repo and CT-Pub datasets to generate features from clinical trials. |
| Outcome: | The proposed framework improves the performance of the baseline feature generation model on a clinical trial. |
Lightweight Contenders: Navigating Semi-Supervised Text Mining through Peer Collaboration and Self Transcendence (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing frameworks for semi-supervised text mining with lightweight models are limited by label data scarcity. |
| Approach: | They propose a framework for semi-supervised text mining with lightweight models . it incorporates online distillation to train lightweight student models by imitating the Teacher model . |
| Outcome: | The proposed framework exhibits notable performance enhancements over existing frameworks. |
Language-based Valence and Arousal Expressions between the United States and China: a Cross-Cultural Examination (2025.findings-naacl)
Copied to clipboard
Young Min Cho, Dandan Pang, Stuti Thapa, Garrick Sherman, Lyle Ungar, Louis Tay, Sharath Chandra Guntuku
| Challenge: | valence and arousal are functionally equivalent across social media platforms . americans display higher emotional intensity than Chinese users . |
| Approach: | They compare valence and arousal on Twitter/X and Sina Weibo in China . they use the NRC-VAD lexicon to measure valance and valency . |
| Outcome: | The results show that the valence and arousal of the two platforms differ across cultures . the analysis also shows that the US users display higher emotional intensity than Chinese users . |
Chain-of-Rank: Enhancing Large Language Models for Domain-Specific RAG in Edge Device (2025.findings-naacl)
Copied to clipboard
| Challenge: | Retrieval-augmented generation (RAG) is valuable in specialized domains where precision is critical. |
| Approach: | They propose a chain-of-rank algorithm which allows LLMs to access a target domain early via finetuning. |
| Outcome: | The proposed method achieves state-of-the-art in benchmarks and analyzes its efficacy. |
MALoRA: Mixture of Asymmetric Low-Rank Adaptation for Enhanced Multi-Task Learning (2025.findings-naacl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) can be fine-tuned to new tasks, but in multi-task scenarios, training imbalance and seesaw effect often arise. |
| Approach: | They propose a flexible fine-tuning framework that leverages asymmetric optimization among LoRA experts to reduce training imbalance and improve performance. |
| Outcome: | The proposed framework outperforms baseline methods in inter- and intra-task learning scenarios. |
LlamaLens: Specialized Multilingual LLM for Analyzing News and Social Media Content (2025.findings-naacl)
Copied to clipboard
Mohamed Bayan Kmainasi, Ali Ezzat Shahroor, Maram Hasanain, Sahinur Rahman Laskar, Naeemul Hassan, Firoj Alam
| Challenge: | Large Language Models (LLMs) have demonstrated remarkable success as general-purpose task solvers across various fields. |
| Approach: | They propose to develop a specialized LLM for analyzing news and social media content in a multilingual context. |
| Outcome: | The proposed model outperforms the current state-of-the-art on 23 testing sets and achieves comparable performance on 8 sets. |
LLMs are Biased Teachers: Evaluating LLM Bias in Personalized Education (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing studies have shown that relying on LLMs as information providers may hurt student learning. |
| Approach: | They introduce and apply two bias score metrics to evaluate LLMs for bias in the personalized educational setting, specifically on the models’ roles as “teachers.” |
| Outcome: | The proposed models harm student learning by perpetuating harmful stereotypes and reversing them. |
Preserving Zero-shot Capability in Supervised Fine-tuning for Multi-label Text Classification (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing methods that assume label descriptions ensure zero-shot capability lose their zero-shot capability during training. |
| Approach: | They propose a method that preserves the zero-shot capabilities of powerful dual encoders and label-wise attention networks by freezing the label encoder. |
| Outcome: | The proposed methods preserve the zero-shot capabilities of powerful dual encoder and label-wise attention network architectures by freezing the label encoder. |
Data-centric NLP Backdoor Defense from the Lens of Memorization (2025.findings-naacl)
Copied to clipboard
| Challenge: | Backdoor attacks pose a severe threat to the trustworthiness of DNN-based language models. |
| Approach: | They propose a data-centric defense that extends memorization definitions to fine-grained sentences . they find that duplicated sentence elements are necessary for successful backdoor attacks . |
| Outcome: | The proposed defense outperforms state-of-the-art defenses against backdoor attacks. |
Neuro-Symbolic Integration Brings Causal and Reliable Reasoning Proofs (2025.findings-naacl)
Copied to clipboard
| Challenge: | a new framework for complex reasoning with LLMs is developed to improve reasoning proof accuracy and interpretability. |
| Approach: | They propose to use LLMs to generate search logs that can be interpreted into human-readable reasoning proofs. |
| Outcome: | The proposed framework improves reasoning accuracy but lacks interpretability due to black-box nature of the solvers. |
Infogent: An Agent-Based Framework for Web Information Aggregation (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing web navigation tasks evaluate web agents on task completion basis . however, information aggregation tasks have received relatively little attention . |
| Approach: | They propose a web navigation framework that uses three components for web information aggregation. |
| Outcome: | The proposed framework beats existing SOTA search framework by 7% under Direct API-Driven Access on FRAMES and improves over an existing information-seeking web agent by 4.3% under Interactive Visual Access on AssistantBench. |
On the Role of Key Phrases in Argument Mining (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing approaches to argument mining often overlook crucial conceptual links between ACs and ARs. |
| Approach: | They propose a framework that extracts key phrases from AM benchmarks using an open-source Large Language Model. |
| Outcome: | The proposed framework surpasses baselines on three structurally distinct AM benchmarks by up to 9.5% F1 score, demonstrating its strong potential. |
TabComp: A Dataset for Visual Table Reading Comprehension (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing OCR-free models struggle with complex table layouts and formatting. |
| Approach: | They propose a dataset to evaluate OCR-free models in table reading comprehension using table images, questions, and generative answers. |
| Outcome: | The proposed model outperforms existing models in the visual table reading comprehension dataset and encourages the research community to work on improving them. |
RankAdaptor: Hierarchical Rank Allocation for Efficient Fine-Tuning Pruned LLMs via Performance Model (2025.findings-naacl)
Copied to clipboard
| Challenge: | Current compression techniques entail structural pruning and a recovery phase that leverages the Low-Rank Adaptation algorithm. |
| Approach: | They propose a hierarchical rank allocation method that enables efficient fine-tuning of pruned LLMs according to layerwise specific recovery requirements. |
| Outcome: | The proposed algorithm outperforms state-of-the-art methods across pruning settings and LLM architectures with improvements ranging from 0.7% to 5.5%. |
Rationale Behind Essay Scores: Enhancing S-LLM’s Multi-Trait Essay Scoring with Rationale Generated by LLMs (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing automated essay scoring relies on essay text without explanatory rationales for the scores. |
| Approach: | They propose a rationale-based multiple trait scoring approach that integrates large language models with a smaller large language model to generate trait-specific rationales. |
| Outcome: | The proposed approach outperforms state-of-the-art models and vanilla S-LLMs on benchmark datasets. |
MTPChat: A Multimodal Time-Aware Persona Dataset for Conversational Agents (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing time-aware datasets that focus on persona-grounded conversations focus on temporal dynamics, which narrows their scope and diminishes their complexity. |
| Approach: | They propose a multimodal, time-aware persona dialogue dataset that integrates linguistic, visual, and temporal elements within dialogue and persona memory. |
| Outcome: | The proposed framework integrates linguistic, visual, and temporal elements within dialogue and persona memory to assess a model’s ability to understand implicit temporal cues and dynamic interactions. |
MetaAlign: Align Large Language Models with Diverse Preferences during Inference Time (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing methods to align large language models with human preferences often result in a static alignment that cannot account for the diversity of human preferences in practical applications. |
| Approach: | They propose a method to help large language models dynamically align with various explicit or implicit preferences specified at inference time. |
| Outcome: | The proposed method can help LLMs dynamically align with various explicit or implicit preferences specified at the inference stage, validating the feasibility of MetaAlign. |
MAQA: Evaluating Uncertainty Quantification in LLMs Regarding Data Uncertainty (2025.findings-naacl)
Copied to clipboard
| Challenge: | despite advances in large language models, they still produce false but incorrect responses. |
| Approach: | They propose a new benchmark for large language models that requires more than two unambiguous answers . they also assess 5 different uncertainty quantification methods in the presence of data uncertainty. |
| Outcome: | The proposed method fails in multi-answer question answering tasks compared to single-answered questions . entropy- and consistency-based methods effectively estimate model uncertainty, the authors show . |
Tuning-Free Personalized Alignment via Trial-Error-Explain In-Context Learning (2025.findings-naacl)
Copied to clipboard
Hyundong Justin Cho, Karishma Sharma, Nicolaas Paul Jedema, Leonardo F. R. Ribeiro, Jonathan May, Alessandro Moschitti
| Challenge: | Language models are biased towards generic outputs as they are trained to align to an aggregate preference to be generally useful. |
| Approach: | They propose a tuning-free method that personalizes language models for text generation tasks with fewer than 10 examples per user. |
| Outcome: | The proposed method achieves favorable win rates on pairwise comparisons with the previous state-of-the-art and outperforms competitive tuning-free baselines for personalized alignment tasks of writing emails, essays and news articles. |
Causal Inference with Large Language Model: A Survey (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing causal inference frameworks do not match human judgment in several key areas, such as domain knowledge, logical inference, and cultural context. |
| Approach: | They propose to apply large language models to causal inference tasks . they summarize the main causal problems and approaches and compare their results . |
| Outcome: | The proposed methods are compared with traditional methods in healthcare, finance, and economics. |
Ask Optimal Questions: Aligning Large Language Models with Retriever’s Preference in Conversation (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing methods to perform conversational search are sub-optimal due to the limited ability to incorporate signals from the retrieval results. |
| Approach: | They propose to optimize a language model for reformulating search queries in line with retrievers’ preferences by combining a large-scale dataset with Retrievers’ Feedback. |
| Outcome: | The proposed framework outperforms existing methods on two benchmarks and surpasses the state-of-the-art methods. |
Systematic Knowledge Injection into Large Language Models via Diverse Augmentation for Domain-Specific RAG (2025.findings-naacl)
Copied to clipboard
Kushagra Bhushan, Yatin Nandwani, Dinesh Khandelwal, Sonam Gupta, Gaurav Pandey, Dinesh Raghu, Sachindra Joshi
| Challenge: | Retrieval-Augmented Generation (RAG) enhances response relevance by incorporating retrieved domain knowledge in the context, retrieval errors can still lead to hallucinations and incorrect answers. |
| Approach: | They propose a framework that augments the learning process by context augmentation and knowledge paraphrasing by incorporating retrieved domain knowledge into the context. |
| Outcome: | The proposed framework achieves 10% relative gain in token-level recall while preserving the LLM’s generalization capabilities. |
Find the Intention of Instruction: Comprehensive Evaluation of Instruction Understanding for Large Language Models (2025.findings-naacl)
Copied to clipboard
| Challenge: | LLMs are prone to generate responses to instruction-formatted statements in an instinctive manner, rather than comprehending the underlying user intention within the given instructions. |
| Approach: | They propose to use an instruction-following capability benchmark to evaluate LLMs' instruction understanding capability. |
| Outcome: | The proposed benchmark analyzes the instruction understanding capability of large language models with four instruction candidates and a single candidate. |
Long-Tail Crisis in Nearest Neighbor Language Models (2025.findings-naacl)
Copied to clipboard
| Challenge: | Prior studies have shown that kNN-LM can retrieve long-tail contexts, leaving the model’s performance underexplored in estimating the probabilities of long-tailed target tokens. |
| Approach: | They investigate the behavior of kNN-LM on low-frequency tokens, examining prediction probability, retrieval accuracy, and token distribution in the datastore. |
| Outcome: | The proposed model improves the perplexity of given text by directly accessing a large datastore built from any text data during inference. |
Keep Guessing? When Considering Inference Scaling, Mind the Baselines (2025.findings-naacl)
Copied to clipboard
| Challenge: | Experiments spanning mathematical reasoning and factual knowledge reveal that repeated sampling improves coverage in large language models. |
| Approach: | They define a baseline that enumerates answers according to their prevalence in the training set and then scale it to measure coverage. |
| Outcome: | The proposed model outperforms repeated model sampling for some LLMs while the coverage for others is on par with that of a mixture strategy that obtains k answers by using only 10 model samples and guessing the remaining k-10 attempts via enumeration. |
Large Language Models for Anomaly and Out-of-Distribution Detection: A Survey (2025.findings-naacl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have demonstrated their effectiveness in natural language processing but also in broader applications due to their advanced comprehension and generative capabilities. |
| Approach: | They propose a taxonomy to categorize existing approaches into two classes based on the role played by LLMs. |
| Outcome: | The proposed taxonomy categorizes existing approaches into two classes based on the role played by LLMs. |
Time-aware ReAct Agent for Temporal Knowledge Graph Question Answering (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing solutions for temporal knowledge graph question answering lack sufficient temporal constraints in retrieval process. |
| Approach: | They propose a temporal knowledge graph question answering framework that integrates temporal constraints into information retrieval. |
| Outcome: | The proposed framework achieves a 41.3% improvement over the baseline model and a 32.2% gain compared to the Abstract Reasoning Induction (ARI) method. |
SG-FSM: A Self-Guiding Zero-Shot Prompting Paradigm for Multi-Hop Question Answering Based on Finite State Machine (2025.findings-naacl)
Copied to clipboard
Xiaochen Wang, Junqing He, Liang Chen, Gholamreza Haffari, Yiru Wang, Zhe Yang, Xiangdi Meng, Kunhao Pan, Zhifang Sui
| Challenge: | Multi-hop Question Answering (MHQA) is a challenging task that requires models to answer multiple questions with multiple passages. |
| Approach: | They propose a self-guided prompting finite state machine to improve multi-hop reasoning abilities by iterating over multiple questions and correcting itself to improve accuracy. |
| Outcome: | The proposed approach outperforms baselines on Musique and other datasets. |
Dynamic Strategy Planning for Efficient Question Answering with Large Language Models (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing studies have shown that using a single fixed strategy for answering all different kinds of questions is sub-optimal in performance and inefficient in terms of generated tokens and retrievals. |
| Approach: | They propose a technique to induce a dynamic strategy selection process in Large Language Models (LLMs) by incorporating an initial decision step to select the most suitable strategy conditioned on the input question and guides the LLM’s response generation accordingly. |
| Outcome: | The proposed technique improves model performance by 7-13% while reducing the cost by 11-32% relative to the best baseline model. |
Can I Introduce My Boyfriend to My Grandmother? Evaluating Large Language Models Capabilities on Iranian Social Norm Classification (2025.findings-naacl)
Copied to clipboard
| Challenge: | Introducing the Iranian Social Norms dataset, a collection of 1,699 social norms, with Farsi adding linguistic complexity. |
| Approach: | They propose a collection of Iranian social norms with English translations and a novel Iranian dataset. |
| Outcome: | The Iranian Social Norms dataset is the first to be used in the Farsi language . it includes 1,699 social norms including environments, demographic features, and scope annotation, alongside English translations. |
PLD+: Accelerating LLM Inference by Leveraging Language Model Artifacts (2025.findings-naacl)
Copied to clipboard
| Challenge: | speculative decoding is a novel decoding paradigm for large language models . however, its use is limited by its computational resources and fine-tuning requirements . |
| Approach: | They propose a tuning-free approach that accelerates inference of large language models . they use draft and verify principle to accelerate inference process . |
| Outcome: | The proposed approach outperforms tuning-free approaches on input-guided tasks and outperformed state-of-the-art EAGLE on four of the tasks. |
Adapting LLM Agents with Universal Communication Feedback (2025.findings-naacl)
Copied to clipboard
| Challenge: | Recent advances in large language models (LLMs) have demonstrated potential for LLM agents. |
| Approach: | They propose a universal buffer and iterative pipeline to store feedback and itersative pipelines to enable LLM agents to explore and update their policy in an environment. |
| Outcome: | The proposed approach outperforms supervised instruction fine-tuning baselines on four datasets. |
Ignore the KL Penalty! Boosting Exploration on Critical Tokens to Enhance RL Fine-Tuning (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing studies show that large language models can learn to achieve long-term goals when fine-tuned with reinforcement learning. |
| Approach: | They propose a modification to the Kullback-Leibler penalty to favor exploration on critical tokens . they show how varying degrees of pre-training influence exploration . |
| Outcome: | The proposed model favors exploration on critical tokens, increasing the efficiency of the RL fine-tuning stage. |
SeaExam and SeaBench: Benchmarking LLMs with Local Multilingual Questions in Southeast Asia (2025.findings-naacl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have shown remarkable performance across various English benchmarks, including both human exam datasets such as MMLU and instruction-following datasets. |
| Approach: | They introduce two new benchmarks to evaluate the capabilities of Large Language Models in Southeast Asian (SEA) application scenarios. |
| Outcome: | The proposed benchmarks show that they can discern LLM performance on SEA language tasks compared to their translated benchmarks. |
Learning to Search Effective Example Sequences for In-Context Learning (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing methods address these factors in isolation, overlooking their interdependencies. Existing approaches focus on sequence selection, while focusing on the sequence of examples. |
| Approach: | They propose a method that considers key factors involved in sequence selection and incrementally builds the sequence. |
| Outcome: | Experiments across various datasets and language models show that the proposed method significantly reduces the search space and improves performance. |
From Intentions to Techniques: A Comprehensive Taxonomy and Challenges in Text Watermarking for Large Language Models (2025.findings-naacl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) are rapidly growing and allowing textual content to be protected against unauthorized use. |
| Approach: | They present a unified overview of different perspectives behind designing watermarking techniques through a comprehensive survey of the research literature. |
| Outcome: | The proposed methods are based on the evaluation datasets used and watermarking addition and removal methods to construct a taxonomy. |
M-IFEval: Multilingual Instruction-Following Evaluation (2025.findings-naacl)
Copied to clipboard
| Challenge: | Instruction following is a core capability of Large language models (LLMs), making evaluating this capability essential to understanding these models. |
| Approach: | They propose a multilingual instruction following evaluation benchmark that expands to other languages . they propose to use both general and language-specific instructions to evaluate LLMs . |
| Outcome: | The proposed benchmark is extended to French, Japanese, and Spanish . it shows that performance across languages and instruction types can vary widely . |
Automatic Annotation Augmentation Boosts Translation between Molecules and Natural Language (2025.findings-naacl)
Copied to clipboard
| Challenge: | Recent advances in AI for biological research focus on integrating molecular data with natural language to accelerate drug discovery. |
| Approach: | They propose a Language-based Automatic Annotation Augmentation framework that leverages large language models to augment existing datasets. |
| Outcome: | The proposed framework outperforms state-of-the-art models on text-based tasks and validates its versatility and utility. |
Let Modalities Teach Each Other: Modal-Collaborative Knowledge Extraction and Fusion for Multimodal Knowledge Graph Completion (2025.findings-naacl)
Copied to clipboard
| Challenge: | Recent studies have focused on missing triples in knowledge graphs, but lack correlation between modalities. |
| Approach: | They propose a framework to foster mutual guidance and collaboration in unimodal knowledge extraction and multimodal knowledge fusion. |
| Outcome: | Extensive experiments on three real-world datasets demonstrate advantages of Moodle over state-of-the-art methods. |
Modeling the Differential Prevalence of Online Supportive Interactions in Private Instant Messages of Adolescents (2025.findings-naacl)
Copied to clipboard
| Challenge: | Approximately two-thirds (68%) of American teenagers aged 13-17 have reported that social media make them feel as though they have people who will support them during challenging times. |
| Approach: | They propose to use the Social Support Behavioral Code to detect and model gender-based and pair-or-group disparities in online supportive interactions among adolescents. |
| Outcome: | The proposed model can be used to model gender-based and pair-or-group disparities in supportive interactions among adolescents. |
Dynamic Feature Fusion for Sign Language Translation Using HyperNetworks (2025.findings-naacl)
Copied to clipboard
| Challenge: | Using RGB and keypoint streams, sign language translation is highly dependent on the brain's ability to process color, shape, and motion simultaneously. |
| Approach: | They propose a hypernetwork-based fusion method that extracts salient features from RGB and keypoint streams and introduces self-distillation and SST contrastive learning to maintain feature advantages while aligning the global semantic space. |
| Outcome: | The proposed method achieves state-of-the-art performance on two public sign language datasets, reducing model parameters by about two-thirds. |
Selective Self-to-Supervised Fine-Tuning for Generalization in Large Language Models (2025.findings-naacl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) can be fine-tuned on task-specific data to improve performance on target tasks but can be overfitted resulting in a loss of generalization. |
| Approach: | They propose a method that uses the correct model responses from a training set to fine-tune the model using the correct response and the gold response for the remaining samples. |
| Outcome: | The proposed approach reduces model specialization during the fine-tuning stage while improving generalization. |
ProverbEval: Exploring LLM Evaluation Challenges for Low-resource Language Understanding (2025.findings-naacl)
Copied to clipboard
Israel Abebe Azime, Atnafu Lambebo Tonja, Tadesse Destaw Belay, Yonas Chanie, Bontu Fufa Balcha, Negasi Haile Abadi, Henok Biadglign Ademtew, Mulubrhan Abebe Nerea, Debela Desalegn Yadeta, Derartu Dagne Geremew, Assefa Atsbiha Tesfu, Philipp Slusallek, Thamar Solorio, Dietrich Klakow
| Challenge: | Large language models (LLMs) evaluation is gaining increasing attention as they are typically trained on general-domain datasets while demonstrating notable performance on tasks out of their training domains. |
| Approach: | They propose an LLM evaluation benchmark for low-resource languages that focuses on low-rsource language understanding in culture-specific scenarios. |
| Outcome: | The proposed benchmarks outperform monolingual evaluations on proverb generation tasks and native language proverb descriptions on multiple choice tasks. |
MRE-MI: A Multi-image Dataset for Multimodal Relation Extraction in Social Media Posts (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing approaches to Multimodal Relation Extraction focus on single image scenarios . current approaches focus on text paired with a single image, ignoring valuable insights provided by remaining images. |
| Approach: | They propose a human-annotated dataset that includes multi-image and single-image instances for relation extraction. |
| Outcome: | The proposed model significantly improves relation extraction in multi-image scenarios. |
Discrete Diffusion Language Model for Efficient Text Summarization (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing discrete diffusion models fail on conditional long-text generation due to incompatibility between the backbone architectures and the random noising process. |
| Approach: | They propose a semantic-aware noising process that enables Transformer backbones to handle long sequences effectively. |
| Outcome: | The proposed model outperforms existing models on three benchmark summarization datasets while achieving much faster inference speed compared to autoregressive models. |
CAPE: A Chinese Dataset for Appraisal-based Emotional Generation in Large Language Models (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing LLMs fail to capture the nuances of human emotions, making their interactions seem impersonal or inadequate. |
| Approach: | They propose a two-stage automatic data generation framework to generate a Chinese dataset called CAPE . their data is a cognitive appraisal theory-based Emotional corpus that accounts for personal and situational factors. |
| Outcome: | The proposed framework can generate human-like responses in conversation with large language models. |
Beyond Under-Alignment: Atomic Preference Enhanced Factuality Tuning for Large Language Models (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing work evaluates the factuality of large language models on in-domain (ID) datasets and the factuality on out-of-domain datasets. |
| Approach: | They propose a framework that enhances model’s awareness of factuality at the granularity of individual facts and propose 'Atomic Preference Enhanced Factuality Tuning' this framework enhances the model’ s awareness and accuracy of factual information at the level of individual factual facts. |
| Outcome: | The proposed framework improves model performance by an average of on ID and OOD datasets, which is highly effective. |
Weight-based Analysis of Detokenization in Language Models: Understanding the First Stage of Inference Without Inference (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing evidence for the stages-of-inference hypothesis is that early layers of language models map their subword tokenized input to more meaningful representations that form the model’s “inner vocabulary”. |
| Approach: | They propose an analytical decomposition of first-layer attention in language models that quantifies the relative contributions of position-related, token-related and mixed effects. |
| Outcome: | The proposed analysis yields interpretable terms that quantify the relative contributions of position-related, token-related and mixed effects. |
DiPT: Enhancing LLM Reasoning through Diversified Perspective-Taking (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing work on improving language model reasoning typically explores a single solution path, which can be prone to errors. |
| Approach: | They propose a novel approach that incorporates diversified viewpoints into existing reasoning methods to improve their reasoning performance. |
| Outcome: | The proposed approach can be flexibly integrated into existing models that focus on a single reasoning approach, enhancing their reasoning performance and stability when presented with paraphrased problems. |
SOLID: Self-seeding and Multi-intent Self-instructing LLMs for Generating Intent-aware Information-Seeking Dialogs (2025.findings-naacl)
Copied to clipboard
Arian Askari, Roxana Petcu, Chuan Meng, Mohammad Aliannejadi, Amin Abolghasemi, Evangelos Kanoulas, Suzan Verberne
| Challenge: | Existing methods for intent prediction rely on human feedback and are tailored to structured intents. |
| Approach: | They propose a method that generates dialogs turn-by-turn using self-seeding and multi-intent self-instructing strategies. |
| Outcome: | The proposed methods generate dialogs turn-by-turn using self-seeding and multi-intent self-instructing strategies. |
CollagePrompt: A Benchmark for Budget-Friendly Visual Recognition with GPT-4V (2025.findings-naacl)
Copied to clipboard
| Challenge: | Recent advances in generative AI have suggested that by taking visual prompts, GPT-4V can demonstrate significant proficiency in visual recognition tasks. |
| Approach: | They propose a collage prompting task that collages multiple images into a single visual prompt and makes GPT-4V perform visual recognition on several images simultaneously. |
| Outcome: | The proposed task reduces the cost associated with GPT-4V's visual recognition . the proposed task group images of the same category together leads to better visual recognition results . |
ARISE: Iterative Rule Induction and Synthetic Data Generation for Text Classification (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing frameworks for large language models (LLMs) generate high-quality synthetic data that can be used to supplement training data or surpass crowd-sourced annotations. |
| Approach: | They propose a framework that iteratively induces rules and generates synthetic data for text classification. |
| Outcome: | The proposed framework outperforms existing models on in-context learning and fine-tuning settings by using augmented data. |
Unleashing Multi-Hop Reasoning Potential in Large Language Models through Repetition of Misordered Context (2025.findings-naacl)
Copied to clipboard
| Challenge: | Multi-hop reasoning requires multi-step reasoning based on supporting documents within a given context. |
| Approach: | They propose a method that prompts the model by repeatedly presenting the context. |
| Outcome: | The proposed method improves the F1 score by 30%p on multi-hop QA tasks and increases accuracy by 70%p on a synthetic task. |
Text Annotation via Inductive Coding: Comparing Human Experts to LLMs in Qualitative Data Analysis (2025.findings-naacl)
Copied to clipboard
| Challenge: | Qualitative data analysis (QDA) is an important research method across fields such as marketing, media studies, social science, psychology, medical research, and others. |
| Approach: | They evaluate the performance of open-source LLMs by comparing them to human experts. |
| Outcome: | The proposed method is based on inductive coding using large language models. |
Investigating the Zone of Proximal Development of Language Models for In-Context Learning (2025.findings-naacl)
Copied to clipboard
| Challenge: | In-context learning is a dynamic and progressive process where learners integrate new information into their knowledge base through interactions with the environment. |
| Approach: | They propose a learning analytics framework to analyze the in-context learning behavior of large language models (LLMs) through the lens of the Zone of Proximal Development (ZPD), an established theory in educational psychology. |
| Outcome: | The proposed framework improves inference and fine-tuning scenarios by selectively applying it to queries that are most likely to benefit from demonstrations. |
Breaking ReAct Agents: Foot-in-the-Door Attack Will Get You In (2025.findings-naacl)
Copied to clipboard
| Challenge: | Indirect prompt injection attacks, prompted by harmless and unrelated requests, can significantly increase the likelihood of the agent performing subsequent malicious actions. |
| Approach: | They propose to implement a simple reflection mechanism that prompts the agent to reassess the safety of its actions during execution, which can help mitigate this vulnerability. |
| Outcome: | The proposed method reduces the success of such attacks by prompting the agent to reassess its actions during execution. |
As easy as PIE: understanding when pruning causes language models to disagree (2025.findings-naacl)
Copied to clipboard
| Challenge: | Language Model pruning reduces the model's efficiency by removing weights, nodes, or other parts of its architecture. |
| Approach: | They propose to prune Language Models (LMs) to produce smaller, hence more efficient models with small loss to their effectiveness. |
| Outcome: | The proposed pruning method hurts data points that matter the most when pruning . the proposed pruning technique is based on a new study of NLP datasets . |
Multi-Agent Simulator Drives Language Models for Legal Intensive Interaction (2025.findings-naacl)
Copied to clipboard
ShengbinYue ShengbinYue, Ting Huang, Zheng Jia, Siyuan Wang, Shujun Liu, Yun Song, Xuanjing Huang, Zhongyu Wei
| Challenge: | Large Language Models (LLMs) have advanced legal intelligence, but the scarcity of scenario data impedes the progress toward interactive legal scenarios. |
| Approach: | They propose a Multi-agent Legal Simulation Driver to generate synthetic data by simulating interactive legal scenarios. |
| Outcome: | The proposed framework ensures consistency of legal attributes between participants and introduces a supervisory mechanism to align participants’ characters and behaviors as well as addressing distractions. |
Exploring Backward Reasoning in Large Language Models (2025.findings-naacl)
Copied to clipboard
| Challenge: | Multi-step reasoning through in-context learning strategies have been extensively explored, highlighting the abilities of Large Language Models (LLMs) to solve problems in a step-wise manner. |
| Approach: | They propose to use Large Language Models to generate answers from step-by-step reasoning by re-constructing the original question that led to the final answer. |
| Outcome: | The proposed models show that they are able to reason about the conclusion and reconstruct the original question that led to the final answer. |
MMLF: Multi-query Multi-passage Late Fusion Retrieval (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing approaches to query expansion are limited in terms of lexical overlap. |
| Approach: | They propose a query expansion pipeline that generates sub-queries, expands them into pseudo-documents, retrieves them individually and aggregates results using reciprocal rank fusion. |
| Outcome: | The proposed pipeline improves on five BEIR benchmark datasets and achieves a maximum gain of up to 8%. |
Dynamic Guided and Domain Applicable Safeguards for Enhanced Security in Large Language Models (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing defense methods struggle with two key issues: inadequate defense capabilities and over-defensiveness. |
| Approach: | They propose a multi-agents-based framework that leverages accurate external information to provide an unbiased summary of user intentions and safety response guidance. |
| Outcome: | Experiments on popular jailbreak attacks and benign datasets show that the proposed framework can enhance LLM's robustness against jailbreaks without compromising its general functionality. |
kNN For Whisper And Its Effect On Bias And Speaker Adaptation (2025.findings-naacl)
Copied to clipboard
| Challenge: | Speech recognition performance varies by language, domain, speaker characteristics such as accent . fine-tuning models on any of these categories may lead to catastrophic forgetting . |
| Approach: | They propose a nonparametric method that adapts to any category without training the model. |
| Outcome: | The proposed method improves the performance of a transformer end-to-end speech model on transcribed audio in over 680,000 hours of training data on a wide variety of languages. |
VisualCoder: Guiding Large Language Models in Code Execution with Fine-grained Multimodal Chain-of-Thought Reasoning (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing approaches to enhance large language models' ability to predict program behavior struggle with dynamic reasoning tasks. |
| Approach: | They propose a visual control flow graph that integrates CoT reasoning with a control flow . they aim to improve performance in program behavior prediction, error detection and output generation . |
| Outcome: | The proposed approach improves program behavior prediction, error detection, and output generation. |
Optimizing LLMs for Italian: Reducing Token Fertility and Enhancing Efficiency Through Vocabulary Adaptation (2025.findings-naacl)
Copied to clipboard
Luca Moroni, Giovanni Puccetti, Pere-Lluís Huguet Cabot, Andrei Stefan Bejgu, Alessio Miaschi, Edoardo Barba, Felice Dell’Orletta, Andrea Esuli, Roberto Navigli
| Challenge: | Pretrained Large Language Models (LLMs) are mainly designed for the English language, but are not optimized for non-English languages due to language contamination or multilingual pretraining data. |
| Approach: | They propose a method that leverages neural mapping for vocabulary substitution to optimize LLMs for the Italian language. |
| Outcome: | The proposed method reduces token fertility by 25% and improves grounded alignment strategies. |
Beyond the Mode: Sequence-Level Distillation of Multilingual Translation Models for Low-Resource Language Pairs (2025.findings-naacl)
Copied to clipboard
Aarón Galiano-Jiménez, Juan Antonio Pérez-Ortiz, Felipe Sánchez-Martínez, Víctor M. Sánchez-Cartagena
| Challenge: | Existing multilingual pre-trained models for low-resource languages have outperformed those trained from scratch for low resources due to high hardware requirements. |
| Approach: | They propose to use beam search to decode the whole output distribution of the teacher to improve student learning. |
| Outcome: | The proposed methods improve student model performance and reduce gender bias amplification common to beam search based methods. |
LLMs for Extremely Low-Resource Finno-Ugric Languages (2025.findings-naacl)
Copied to clipboard
| Challenge: | Low-resource languages such as those in the Finno-Ugric family are underrepresented in large language models. |
| Approach: | They propose to develop large language models for extremely low-resource languages . they focus on Vro, Livonian, and Komi, which are underrepresented . |
| Outcome: | The proposed models cover almost the entire cycle of creation, from data collection to instruction tuning and evaluation. |
LOFT: Scalable and More Realistic Long-Context Evaluation (2025.findings-naacl)
Copied to clipboard
Jinhyuk Lee, Anthony Chen, Zhuyun Dai, Dheeru Dua, Devendra Singh Sachan, Michael Boratko, Yi Luan, Séb Arnold, Vincent Perot, Siddharth Dalmia, Hexiang Hu, Xudong Lin, Panupong Pasupat, Aida Amini, Jeremy R. Cole, Sebastian Riedel, Iftekhar Naim, Ming-Wei Chang, Kelvin Guu
| Challenge: | Long-context language models (LCLMs) can be used to perform tasks traditionally reliant on external tools like retrieval systems or databases. |
| Approach: | They propose a benchmark to evaluate LCLMs' performance on in-context retrieval and reasoning tasks using a set of tokens. |
| Outcome: | The proposed model outperforms state-of-the-art retrieval and RAG systems on in-context retrieval tasks while still requiring prompting strategies. |
On the Influence of Context Size and Model Choice in Retrieval-Augmented Generation Systems (2025.findings-naacl)
Copied to clipboard
| Challenge: | Retrieval-augmented generation (RAG) is an approach to augment large language models (LLMs) despite their impressive performance, LLMs can generate plausible sounding but factually incorrect responses (hallucinations) |
| Approach: | They propose to use BM25 and semantic search as retrievers to augment large language models by reducing their reliance on static knowledge and improving answer factuality. |
| Outcome: | The proposed approach improves QA performance on a biomedical task with up to 15 snippets but stagnates or declines beyond that. |
Aligning Black-box Language Models with Human Judgments (2025.findings-naacl)
Copied to clipboard
| Challenge: | Large language models (LLMs) are increasingly used as automated judges to evaluate recommendation systems, search engines, and other subjective tasks. |
| Approach: | They propose a framework to align LLM judgments with individual human evaluators or their aggregated judgments without retraining or fine-tuning the LLM. |
| Outcome: | The proposed framework achieves 142% improvement in agreement across 29 tasks and exceeds inter-human agreement on four out of six tasks. |
Guideline Compliance in Task-Oriented Dialogue: The Chained Prior Approach (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing solutions based on large language models cannot achieve strict guideline compliance . a novel TOD system is being developed to improve guideline adherence . |
| Approach: | They propose a task-oriented dialogue system that explicitly considers domain-specific guidelines by integrating a policy module. |
| Outcome: | The proposed system achieves 20% better guideline compliance than state-of-the-art solutions. |
AutoBreach: Universal and Adaptive Jailbreaking with Efficient Wordplay-Guided Optimization via Multi-LLMs (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing jailbreak research exhibits limitations in universality, validity, and efficiency . Existing methods for jailbreaking LLMs have limited validity and effectiveness . |
| Approach: | They propose a black-box approach that uses wordplay-guided mapping rule sampling to create universal adversarial prompts. |
| Outcome: | The proposed method efficiently identifies security vulnerabilities across various LLMs, achieving an average success rate of over 80% with fewer than 10 queries. |
𝒮2IT: Stepwise Syntax Integration Tuning for Large Language Models in Aspect Sentiment Quad Prediction (2025.findings-naacl)
Copied to clipboard
| Challenge: | Aspect Sentiment Quad Prediction (ASQP) is an extractive task that focuses on predicting tuples of sentiment-related elements from a given text. |
| Approach: | They propose a stepwise syntax integration tuning framework that integrates syntactic structure knowledge into LLMs through a multi-step tuning process. |
| Outcome: | The proposed framework integrates syntactic structure knowledge into large language models . it decomposes the quadruple generation task into two stages . the proposed framework significantly improves state-of-the-art performance across multiple datasets . |
BanNERD: A Benchmark Dataset and Context-Driven Approach for Bangla Named Entity Recognition (2025.findings-naacl)
Copied to clipboard
Md. Motahar Mahtab, Faisal Ahamed Khan, Md. Ekramul Islam, Md. Shahad Mahmud Chowdhury, Labib Imam Chowdhury, Sadia Afrin, Hazrat Ali, Mohammad Mamun Or Rashid, Nabeel Mohammed, Mohammad Ruhul Amin
| Challenge: | In a cross-dataset evaluation, models trained on BanNERD consistently outperformed those trained on four existing Bangla NER datasets. |
| Approach: | They propose to use Bangla as a language to create the most extensive human-annotated and validated Bangla NLP dataset. |
| Outcome: | The proposed method outperforms existing methods on Bangla NER datasets and performs competitively on English datasets. |
Large Language Models Reflect Human Citation Patterns with a Heightened Citation Bias (2025.findings-naacl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) can help in citation generation but can also amplify existing biases, such as the Matthew effect, and introduce new ones, potentially skewing scientific knowledge dissemination. |
| Approach: | They propose to use large language models to generate scholarly references for in-text citations in papers published after GPT-4's knowledge cut-off date. |
| Outcome: | The proposed model can generate scholarly references for in-text citations, but without the aid of web browsing or a search engine, the results show a similarity between human and LLM citation patterns, but with a more pronounced high citation bias. |
What can Large Language Models Capture about Code Functional Equivalence? (2025.findings-naacl)
Copied to clipboard
| Challenge: | SeqCoBench is a benchmark to assess how Code-LLMs can capture code semantics. |
| Approach: | They propose a benchmark to assess how Code-LLMs capture code semantics . they use seqCoBench to evaluate whether they can discern semantically equivalent or different pairs of programs . |
| Outcome: | The proposed benchmarks show that they can capture code semantics better than classical match-based retrieval scores. |
Make Every Penny Count: Difficulty-Adaptive Self-Consistency for Cost-Efficient Reasoning (2025.findings-naacl)
Copied to clipboard
Xinglin Wang, Shaoxiong Feng, Yiwei Li, Peiwen Yuan, Yueqi Zhang, Chuyi Tan, Boyuan Pan, Yao Hu, Kan Li
| Challenge: | Existing decoding strategies for chain-of-thought reasoning do not exploit prior information about question difficulty. |
| Approach: | They propose a decoding strategy called self-consistency to improve reasoning performance by adjusting the number of samples based on the posterior distribution of a set of pre-samples. |
| Outcome: | The proposed method outperforms baseline methods on arithmetic, commonsense and symbolic reasoning tasks while achieving comparable performance. |
Large Language Models Are Better Logical Fallacy Reasoners with Counterargument, Explanation, and Goal-Aware Prompt Formulation (2025.findings-naacl)
Copied to clipboard
| Challenge: | Recent large language models have demonstrated improved processing of complex language, but detecting logical fallacies remains a challenge. |
| Approach: | They propose a prompt formulation approach for logical fallacy detection that integrates contextual information into input text and queries for validity within the argument’s context. |
| Outcome: | The proposed approach improves over state-of-the-art models by 0.57 in F1-scores and 0.45 in fine-tuned models. |
MorphNLI: A Stepwise Approach to Natural Language Inference Using Text Morphing (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing models fail to capture important semantic features of logic such as monotonicity and negation. |
| Approach: | They propose a modular step-by-step approach to natural language inference . they use a language model to generate edits to incrementally transform the premise into the hypothesis . |
| Outcome: | The proposed method outperforms baseline models in realistic cross-domain settings with improvements up to 12.6% (relative). |
Unmasking Database Vulnerabilities: Zero-Knowledge Schema Inference Attacks in Text-to-SQL Systems (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing text-to-SQL systems rely on database schema information for SQL generation, exposing them to security vulnerabilities. |
| Approach: | They propose a framework for reconstructing database schema of text-to-SQL models without prior knowledge of the database. |
| Outcome: | The proposed framework achieves high accuracy in reconstructing table names, with F1 scores of up to .99 for generative models and .78 for fine-tuned models, underscoring the severity of schema leakage risks. |
Media of Langue: Exploring Word Translation Network (2025.findings-naacl)
Copied to clipboard
| Challenge: | a new interface for word translation is proposed to explore the semantic configurations of words in multiple languages at once. |
| Approach: | They propose a web interface for word translation that points to the semantic configurations of many words in multiple languages at once. |
| Outcome: | The proposed interface is available as a web application on seven language pairs . it points to the semantic configurations of many words in multiple languages at once . |
Tackling Social Bias against the Poor: a Dataset and a Taxonomy on Aporophobia (2025.findings-naacl)
Copied to clipboard
Georgina Curto, Svetlana Kiritchenko, Muhammad Hammad Fahim Siddiqui, Isar Nejadgholi, Kathleen C. Fraser
| Challenge: | Poverty is a multidimensional phenomenon that affects 712 million people worldwide . |
| Approach: | They propose to annotate a corpus of English tweets from five world regions for the presence of harmful beliefs and discriminative actions against poor people on social media. |
| Outcome: | The proposed model can be used to identify, track and mitigat aporophobia on social media at scale. |
The American Sign Language Knowledge Graph: Infusing ASL Models with Linguistic Knowledge (2025.findings-naacl)
Copied to clipboard
| Challenge: | Sign language models could make language technologies more accessible to deaf and hard-of-hearing signers, but the supply of accurately labeled data struggles to meet the demand associated with training large, end-to-end architectures. |
| Approach: | They construct an American Sign Language Knowledge Graph from 11 sources of linguistic knowledge and use it to train neuro-symbolic models on ASL video input tasks. |
| Outcome: | The proposed model achieves 91% accuracies for isolated sign recognition, 14% for predicting the semantic features of unseen signs, and 36% for classifying the topic of Youtube-ASL videos. |
Reinforcement Learning for Aligning Large Language Models Agents with Interactive Environments: Quantifying and Mitigating Prompt Overfitting (2025.findings-naacl)
Copied to clipboard
Mohamed Salim Aissi, Clément Romac, Thomas Carta, Sylvain Lamprier, Pierre-Yves Oudeyer, Olivier Sigaud, Laure Soulier, Nicolas Thome
| Challenge: | Reinforcement learning (RL) is a promising approach for aligning large language models knowledge with sequential decision-making tasks. |
| Approach: | They propose to use a contrastive loss framework to analyze the sensitivity of LLMs to prompt formulations following RL training in a textual environment. |
| Outcome: | The proposed framework improves the model's robustness and generalization capabilities by minimizing the model’s internal representations and salient tokens. |
An empirical study of validating synthetic data for formula generation (2025.findings-naacl)
Copied to clipboard
Usneek Singh, José Cambronero, Sumit Gulwani, Aditya Kanade, Anirudh Khatry, Vu Le, Mukul Singh, Gust Verbruggen
| Challenge: | Large language models (LLMs) can be leveraged to help write formulas in spreadsheets, but formula data resources are scarce, limiting the ability to fine-tune them. |
| Approach: | They validate a corpus of formulas with a model to generate synthetic natural language utterances for fine-tuning. |
| Outcome: | The proposed model generates synthetic natural language utterances with a model that is accurate enough to fine-tune them. |
TeCoFeS: Text Column Featurization using Semantic Analysis (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing methods for topic modeling and feature extraction are based on syntactic features and overlook the semantics. |
| Approach: | They propose a semantic text column featurization problem that extracts a small sample smartly using an LLM to label only the sample and then extends that labeling to the whole column using text embeddings. |
| Outcome: | The proposed approach performs better than baselines and naive use of LLMs. |
CA*: Addressing Evaluation Pitfalls in Computation-Aware Latency for Simultaneous Speech Translation (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing metrics for Simultaneous speech translation (SimulST) are inaccurately measuring latency in unsegmented streaming settings. |
| Approach: | They propose to modify existing metrics to correctly measure computation-aware latency for SimulST systems, addressing limitations present in existing metrics. |
| Outcome: | The proposed model is based on a real-time, lowlatency scenario where the model starts generating the textual translation before the entire audio input is processed. |
Augmented Adversarial Trigger Learning (2025.findings-naacl)
Copied to clipboard
| Challenge: | Gradient optimization-based adversarial attack methods can generate jailbreak prompts or leak system prompts. |
| Approach: | They propose an algorithm that enhances negative log-likelihood loss and augments it with auxiliary loss. |
| Outcome: | The proposed approach outperforms current state-of-the-art techniques in nearly 100% of attacks while requiring 80% fewer queries. |
Adaptive Attacks Break Defenses Against Indirect Prompt Injection Attacks on LLM Agents (2025.findings-naacl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) agents use external tools to interact with environments, but they also introduce security risks, such as indirect prompt injection (IPI) attacks. |
| Approach: | They evaluate eight different defenses and bypass all of them using adaptive attacks. |
| Outcome: | The proposed defenses achieve an attack success rate of over 50% and are robust against adaptive attacks. |
Flaming-hot Initiation with Regular Execution Sampling for Large Language Models (2025.findings-naacl)
Copied to clipboard
Weizhe Chen, Zhicheng Zhang, Guanlin Liu, Renjie Zheng, Wenlei Shi, Chen Dun, Zheng Wu, Xing Jin, Lin Yan
| Challenge: | Large language models (LLMs) have demonstrated remarkable capabilities across various domains since the release of ChatGPT . a key challenge in developing these general capabilities is efficiently sourcing diverse, high-quality data. |
| Approach: | They introduce Flaming-hot Initiation with Regular Execution (FIRE) sampling to efficiently find good responses by promoting diversity. |
| Outcome: | The proposed method enhances inference-time generation quality and benefits training in the alignment stage. |
HEISIR: Hierarchical Expansion of Inverted Semantic Indexing for Training-free Retrieval of Conversational Data using LLMs (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing methods for retrieving dialogue data often face challenges in capturing semantic intent or require extensive labeling and fine-tuning. |
| Approach: | They propose a framework that enhances semantic understanding in conversational data retrieval by optimizing data ingestion. |
| Outcome: | The proposed framework outperforms fine-tuned models in retrieval and intent analysis. |
“Women do not have heart attacks!” Gender Biases in Automatically Generated Clinical Cases in French (2025.findings-naacl)
Copied to clipboard
| Challenge: | Healthcare professionals are increasingly including Language Models (LMs) in clinical practice. |
| Approach: | They propose to use LMs to generate clinical cases in french and an automatic linguistic gender detection tool to measure gender biases. |
| Outcome: | The proposed model over-generates cases describing male patients, creating synthetic corpora that are not consistent with documented prevalence for these disorders. |
NOTA: Multimodal Music Notation Understanding for Visual Large Language Model (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing general-domain visual language models lack ability of music notation understanding . Symbolic music is represented in two distinct forms: auditory music and symbolic music . |
| Approach: | They propose to train a multimodal music notation model using a large-scale dataset . they use cross-modal alignment to train the model for music notations analysis . |
| Outcome: | The proposed model improves on music understanding by training with a multimodal music notation model. |
Exploring Large Language Models for Hate Speech Detection in Rioplatense Spanish (2025.findings-naacl)
Copied to clipboard
| Challenge: | Hate speech detection deals with many language variants, slang, nuances, and cultural nuances. |
| Approach: | They propose to use large language models to detect hate speech in Rioplatense Spanish . they compare their results to those of a state-of-the-art BERT classifier . |
| Outcome: | The proposed models show lower precision than the state-of-the-art classifier, but are sensitive to highly nuanced cases. |
An Annotated Dataset of Errors in Premodern Greek and Baselines for Detecting Them (2025.findings-naacl)
Copied to clipboard
Creston Brooks, Johannes Haubold, Charlie Cowen-Breen, Jay White, Desmond DeVaul, Frederick Riemenschneider, Karthik R Narasimhan, Barbara Graziosi
| Challenge: | a dataset of real errors in premodern Greek is presented to improve error detection methods . scribal errors are more difficult to detect than print or digitization errors. |
| Approach: | They propose to annotate 1,000 words more likely to contain errors and annotated them as errors or not . they propose to evaluate new error detection methods that outperform other methods . |
| Outcome: | The proposed method outperforms all other methods, improving true positive rate by 5%. |
WorldMedQA-V: a multilingual, multimodal medical examination dataset for multimodal language models evaluation (2025.findings-naacl)
Copied to clipboard
João Matos, Shan Chen, Siena Kathleen V. Placino, Yingya Li, Juan Carlos Climent Pardo, Daphna Idan, Takeshi Tohyama, David Restrepo, Luis Filipe Nakayama, José María Millet Pascual-Leone, Guergana K Savova, Hugo Aerts, Leo Anthony Celi, An-Kwok Ian Wong, Danielle Bitterman, Jack Gallifant
| Challenge: | Existing multiple-choice question and answer (QA) datasets are text-only and available in a limited subset of languages and countries. |
| Approach: | They propose a multilingual, multimodal benchmarking dataset to evaluate multimodal/vision language models in healthcare. |
| Outcome: | The WorldMedQA-V includes 568 labeled multiple-choice QAs paired with 568 medical images from four countries. |
BanTH: A Multi-label Hate Speech Detection Dataset for Transliterated Bangla (2025.findings-naacl)
Copied to clipboard
Fabiha Haider, Fariha Tanjim Shifat, Md Farhan Ishmam, Md Sakib Ul Rahman Sourove, Deeparghya Dutta Barua, Md Fahim, Md Farhad Alam Bhuiyan
| Challenge: | Existing work on monolingual or binary hate classification in Bangla has not addressed the challenge of multi-label hate speech classification in underrepresented languages. |
| Approach: | They propose a multi-label transliterated Bangla hate speech dataset that translates or transliterates under-resourced text to higher-resource text before classifying the hate group(s). |
| Outcome: | The proposed approach outperforms other methods in the zero-shot setting while achieving state-of-the-art performance. |
Mutual Reinforcement of LLM Dialogue Synthesis and Summarization Capabilities for Few-Shot Dialogue Summarization (2025.findings-naacl)
Copied to clipboard
Yen-Ju Lu, Ting-Yao Hu, Hema Swetha Koppula, Hadi Pouransari, Jen-Hao Rick Chang, Yin Xia, Xiang Kong, Qi Zhu, Xiaoming Simon Wang, Oncel Tuzel, Raviteja Vemulapalli
| Challenge: | Empirical results demonstrate that our method improves dialogue summarization, achieving a 1.5% increase in ROUGE scores and a 0.3% improvement in BERT scores in few-shot settings. |
| Approach: | They propose Mutual Reinforcing Data Synthesis (MRDS) within large language models to enhance few-shot dialogue summarization task. |
| Outcome: | Empirical results show that the proposed method improves dialogue summarization, achieving a 1.5% increase in ROUGE scores and a 0.3% improvement in BERT scores in few-shot settings. |
UNLEARN Efficient Removal of Knowledge in Large Language Models (2025.findings-naacl)
Copied to clipboard
| Challenge: | Large Language Models excel in many tasks but are outperformed by specialized tools for certain tasks. |
| Approach: | They propose a method that uses subspace techniques to selectively remove knowledge . they propose 'unlearn' method that can forget or unlear the knowledge without retraining . |
| Outcome: | The proposed method outperforms existing methods for forgetting target knowledge while preserving related knowledge. |
Adaptive Parameter Compression for Language Models (2025.findings-naacl)
Copied to clipboard
| Challenge: | Adaptive parameter compression is a new approach to improve NLP models . the current algorithm is based on a single parameter, but it is not scalable. |
| Approach: | They propose a hardware-independent compression strategy that extends the weight-squeezing approach by introducing compression biases and weights. |
| Outcome: | The proposed compression strategy outperforms DistilBERT base models while being significantly more efficient. |
Personalize Your LLM: Fake it then Align it (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing personalization methods require fine-tuning of large language models for each user, rendering them prohibitively expensive for widespread adoption. |
| Approach: | They propose a retrieval-based personalization approach that uses self-generated personal preference data and representation editing to enable quick and cost-effective personalization. |
| Outcome: | The proposed approach outperforms two personalization baselines by 40% on various tasks. |
A Survey to Recent Progress Towards Understanding In-Context Learning (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing research on In-Context Learning (ICL) is unclear, despite empirical success . a data generation perspective is used to interpret ICL . |
| Approach: | They propose to use data generation to reinterpret recent efforts from a systematic angle to demonstrate the potential broader usage of ICL. |
| Outcome: | The proposed model can learn from examples provided in the prompt, enabling downstream generalization without the need for gradient updates. |
Inference Scaling for Bridging Retrieval and Augmented Generation (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing work observed the generator bias, such that improving the retrieval results may negatively affect the outcome. |
| Approach: | They propose to use inference scaling to aggregate inference calls from the permuted order of retrieved contexts to create a new ranking. |
| Outcome: | The proposed approach improves ROUGE-L on MS MARCO and EM on HotpotQA benchmarks by 7 points. |
GeoCoder: Solving Geometry Problems by Generating Modular Code through Vision-Language Models (2025.findings-naacl)
Copied to clipboard
| Challenge: | Various vision-language models (VLMs) have made significant progress in multimodal tasks, but they still struggle with geometry problems. |
| Approach: | They propose a vision-language model that leverages modular code-finetuning to generate and execute code using a predefined geometry function library. |
| Outcome: | The proposed model improves geometric reasoning abilities by 16% on a GeomVerse dataset compared to other methods. |
SEEval: Advancing LLM Text Evaluation Efficiency and Accuracy through Self-Explanation Prompting (2025.findings-naacl)
Copied to clipboard
| Challenge: | Large language models (LLMs) have achieved remarkable success in various natural language generation tasks, but their performance in automatic text evaluation is not ready as human replacements. |
| Approach: | They propose a prompt-based text evaluator that incorporates self-explanation, a metacognitive strategy, to enhance automatic text evaluation. |
| Outcome: | The proposed method achieves competitive and often superior performance compared to the two state-of-the-art baselines – G-Eval and Analyze-Rate – and is 20 times more efficient in terms of run-time. |
When natural language is not enough: The limits of in-context learning demonstrations in multilingual reasoning (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing studies have demonstrated the effectiveness of reasoning methods in eliciting multi-step reasoned answers from Large Language Models (LLMs) by leveraging in-context demonstrations. |
| Approach: | They investigate how well CoT and PAL perform across languages for arithmetic and symbolic reasoning tasks. |
| Outcome: | The proposed methods perform well in monolingual contexts, primarily in English, but have been limited in other languages. |
Uncovering Latent Arguments in Social Media Messaging by Employing LLMs-in-the-Loop Strategy (2025.findings-naacl)
Copied to clipboard
| Challenge: | Supervised methods are adept at text categorization, but dynamic nature of social media debates pose challenges for them . traditional methods for extracting themes from public discourse often reveal overarching patterns that might not capture specific nuances. |
| Approach: | They propose a generic approach that leverages the advanced capabilities of Large Language Models to extract latent arguments from social media messaging. |
| Outcome: | The proposed approach leverages the advanced capabilities of Large Language Models (LLMs) to extract latent arguments from social media messaging. |
AcrosticSleuth: Probabilistic Identification and Ranking of Acrostics in Multilingual Corpora (2025.findings-naacl)
Copied to clipboard
| Challenge: | acrostics are hidden messages in which initial letters of consecutive lines or paragraphs form meaningful words or phrases. |
| Approach: | They propose a method to identify acrostics automatically and rank them by the probability that the corresponding sequence of characters does not occur by chance. |
| Outcome: | The proposed method achieves F1 scores of 0.39, 0.59, and 0.66 on the French, English, and Russian subdomains of WikiSource. |
MedThink: A Rationale-Guided Framework for Explaining Medical Visual Question Answering (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing models for medical visual question answering are limited in their interpretation and interpretation . a semi-automated annotation process is used to streamline data preparation and build new benchmark datasets . |
| Approach: | They propose a semi-automated annotation process to streamline data preparation and build new benchmark Med-VQA datasets. |
| Outcome: | The proposed method achieves an accuracy of 83.5% on R-RAD, 86.3% on RSLAKE and 87.2% on RPath. |
How to Learn in a Noisy World? Self-Correcting the Real-World Data Noise in Machine Translation (2025.findings-naacl)
Copied to clipboard
| Challenge: | Semantic misalignment, as the primary source of the noise, poses a challenge for training machine translation systems. |
| Approach: | They propose a process for simulating misalignment controlled by semantic similarity which closely resembles misaligned sentences in real-world web-crawled corpora. |
| Outcome: | The proposed model significantly improves translation performance in the presence of misalignment noise and when applied to real-world, noisy web-mined datasets, across a range of translation tasks. |
Rejected Dialects: Biases Against African American Language in Reward Models (2025.findings-naacl)
Copied to clipboard
| Challenge: | Preference alignment via reward models can introduce new biases, hindering reward models’ fairness and equity. |
| Approach: | They propose a framework for evaluating dialect biases in reward models and conduct a case study on biase . they compare reward models' preferences and behavior on paired White Mainstream English and machine-translated and human-written AAL corpora. |
| Outcome: | The proposed framework evaluates dialect biases in reward models and compares them with paired White Mainstream English (WME) and machine-translated and human-written AAL corpora. |
Do Large Language Models Align with Core Mental Health Counseling Competencies? (2025.findings-naacl)
Copied to clipboard
Viet Cuong Nguyen, Mohammad Taher, Dongwan Hong, Vinicius Konkolics Possobom, Vibha Thirunellayi Gopalakrishnan, Ekta Raj, Zihang Li, Heather J. Soled, Michael L. Birnbaum, Srijan Kumar, Munmun De Choudhury
| Challenge: | Large language models are promising for mental health, but their alignment with core counseling competencies remains underexplored. |
| Approach: | They propose a benchmark to evaluate 22 general-purpose and medical-finetuned LLMs across five key competencies. |
| Outcome: | The proposed model outperforms generalist models in Intake, Assessment & Diagnosis but struggles with core counseling attributes and professional practice & ethics. |
Uncertainty Quantification for Clinical Outcome Predictions with (Large) Language Models (2025.findings-naacl)
Copied to clipboard
| Challenge: | Language models (LMs) have significant potential for clinical prediction tasks . however, unreliable decisions can result in significant costs due to compromised patient safety and ethical concerns . |
| Approach: | They propose to combine ensembling and multi-tasking approaches to reduce uncertainty in EHRs by using multi-tapping methods. |
| Outcome: | The proposed framework reduces model uncertainty in white-box and black-box settings, and improves model transparency in both settings. |
Hypothesis Generation for Materials Discovery and Design Using Goal-Driven and Constraint-Guided LLM Agents (2025.findings-naacl)
Copied to clipboard
| Challenge: | Recent research has leveraged Large Language Models to accelerate materials discovery and design. |
| Approach: | They propose a dataset that features goals, constraints, and methods for designing real-world applications and a method that emulates the process a materials scientist would use to evaluate a hypothesis critically. |
| Outcome: | The proposed method emulates the process a materials scientist would use to evaluate a hypothesis critically. |
Aligning to What? Limits to RLHF Based Alignment (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing studies on RLHF and covert and overt biases in large language models are unclear . et al. analyzed off-the-shelf language models to evaluate their overt and cover racial biase . |
| Approach: | They evaluate the relationship between reinforcement learning from human feedback and biases in large language models. |
| Outcome: | The proposed approach can be used to mitigat covert biases, the authors show . they found that the RLHF approach calcifies model biase . |
Beyond Words: Exploring Cultural Value Sensitivity in Multimodal Models (2025.findings-naacl)
Copied to clipboard
| Challenge: | Using large vision-language models to understand cultural contexts is a critical area of research. |
| Approach: | They conduct a thorough evaluation of multimodal models at different scales, focusing on their alignment with cultural values. |
| Outcome: | The proposed models show that they exhibit sensitivity to cultural values but their performance is highly context-dependent. |
Features that Make a Difference: Leveraging Gradients for Improved Dictionary Learning (2025.findings-naacl)
Copied to clipboard
| Challenge: | Sparse Autoencoders (SAEs) are a promising approach for extracting neural network representations by learning a sparse and overcomplete decomposition of the network’s internal activations. |
| Approach: | They propose a method that learns a sparse and overcomplete decomposition of the network's internal activations and a gradient approach to learn latents. |
| Outcome: | The proposed algorithms improve the performance of the k-sparse autoencoder and the ability to learn latent features. |
Tooling or Not Tooling? The Impact of Tools on Language Agents for Chemistry Problem Solving (2025.findings-naacl)
Copied to clipboard
Botao Yu, Frazier N. Baker, Ziru Chen, Garrett Herb, Boyu Gou, Daniel Adu-Ampratwum, Xia Ning, Huan Sun
| Challenge: | Existing evaluations of large language models (LLMs) with tools are limited and qualitative . existing evaluations have been limited and only focus on 14 tasks focusing on compound synthesis. |
| Approach: | They propose to develop an enhanced chemistry agent over ChemCrow to improve chemistry problem solving by integrating tools into LLMs. |
| Outcome: | The proposed agent does not consistently outperform its base LLMs without tools on specialized chemistry tasks and general chemistry questions. |
RusCode: Russian Cultural Code Benchmark for Text-to-Image Generation (2025.findings-naacl)
Copied to clipboard
Viacheslav Vasilev, Julia Agafonova, Nikolai Gerasimenko, Alexander Kapitanov, Polina Mikhailova, Evelina Mironova, Denis Dimitrov
| Challenge: | Text-to-image generation models exhibit a strong bias toward English-speaking cultures, ignoring or misrepresenting the unique characteristics of other language groups, countries, and nationalities. |
| Approach: | They propose a RusCode benchmark to evaluate the quality of text-to-image generation containing elements of the Russian cultural code. |
| Outcome: | The proposed model is based on 1250 text prompts in Russian and their translations into English. |
Evaluation of LLMs-based Hidden States as Author Representations for Psychological Human-Centered NLP Tasks (2025.findings-naacl)
Copied to clipboard
| Challenge: | Many human-centered NLP tasks focus on assessing human-attributes of a user based on their language. |
| Approach: | They evaluate different ways of representing documents and users using different LM and HuLM architectures to predict task outcomes as dynamically changing states and averaged trait-like user-level attributes. |
| Outcome: | The proposed representations predict valence, arousal, empathy, and distress as well as trait-like user-level attributes. |
Large Language Models and Causal Inference in Collaboration: A Comprehensive Survey (2025.findings-naacl)
Copied to clipboard
Xiaoyu Liu, Paiheng Xu, Junda Wu, Jiaxin Yuan, Yifan Yang, Yuhang Zhou, Fuxiao Liu, Tianrui Guan, Haoliang Wang, Tong Yu, Julian McAuley, Wei Ai, Furong Huang
| Challenge: | Large Language Models (LLMs) have shown great potential to enhance Natural Language Processing (NLP) models in areas such as predictive accuracy, fairness, robustness, and explainability. |
| Approach: | They evaluate or improve generative Large Language Models from a causal perspective in areas such as reasoning capacity, fairness and safety issues, explainability, and handling multimodality. |
| Outcome: | The proposed models can be used to perform causal relationship discovery and causal effect estimation tasks. |
ThoughtSculpt: Reasoning with Intermediate Revision and Search (2025.findings-naacl)
Copied to clipboard
| Challenge: | THOUGHTSCULPT is a general reasoning and search method for tasks with outputs that can be decomposed into components. |
| Approach: | They propose a general reasoning and search method for tasks with outputs that can be decomposed into components. |
| Outcome: | THOUGHTSCULPT outperforms state-of-the-art reasoning methods on three tasks . authors show that distinct prompting strategies can influence the performance of LLMs . |
Optimizing Hidden Markov Language Models: An Empirical Study of Reparameterization and Initialization Techniques (2025.findings-naacl)
Copied to clipboard
| Challenge: | Recent work on scaling-up HMMs to perform competitively as language models has indicated that this challenge only increases with larger hidden state sizes. |
| Approach: | They propose two strategies that use neural reparameterization and neural initialization to enhance HMM optimization. |
| Outcome: | The proposed techniques work well for scaled HMM language modeling, and linear reparameterizations can be as effective as non-linear ones, and the strategies are complementary. |
Using Linguistic Entrainment to Evaluate Large Language Models for Use in Cognitive Behavioral Therapy (2025.findings-naacl)
Copied to clipboard
Mina Kian, Kaleen Shrestha, Katrin Fischer, Xiaoyuan Zhu, Jonathan Ong, Aryan Trehan, Jessica Wang, Gloria Chang, Séb Arnold, Maja Mataric
| Challenge: | Entrainment is a communication process that builds a strong relationship between a mental health therapist and their client. |
| Approach: | They evaluate the linguistic entrainment of an LLM in a mental health dialog setting and compare it to trained therapists and non-expert online peer supporters. |
| Outcome: | The proposed model outperforms humans in a cognitive behavioral therapy setting. |
Analysis of LLM as a grammatical feature tagger for African American English (2025.findings-naacl)
Copied to clipboard
| Challenge: | African American English (AAE) presents unique challenges in natural language processing (NLP). |
| Approach: | They evaluate the ability of different NLP systems to recognize distinctive AAE grammatical features by using sentence-level binary classification tasks using both zero-shot and fewshot strategies. |
| Outcome: | The evaluation involved sentence-level binary classification tasks, using both zero-shot and few-shot strategies. |
LLM-Microscope: Uncovering the Hidden Role of Punctuation in Context Memory of Transformers (2025.findings-naacl)
Copied to clipboard
Anton Razzhigaev, Matvey Mikhalchuk, Temurbek Rahmatullaev, Elizaveta Goncharova, Polina Druzhinina, Ivan Oseledets, Andrey Kuznetsov
| Challenge: | Large Language Models (LLMs) encode and store contextual information, but internal mechanisms are opaque. |
| Approach: | They propose a toolkit that assesses token-level nonlinearity, evaluates contextual memory, visualizes intermediate layer contributions and measures intrinsic dimensionality of representations. |
| Outcome: | The proposed framework assesses token-level nonlinearity, evaluates contextual memory, visualizes intermediate layer contributions, and measures the intrinsic dimensionality of representations. |
On A Scale From 1 to 5: Quantifying Hallucination in Faithfulness Evaluation (2025.findings-naacl)
Copied to clipboard
| Challenge: | Hallucination is a popular topic in natural language generation (NLG). |
| Approach: | They propose to use large language models to evaluate faithfulness of guided NLGs by a rubric template and large language inference models to score the generation on quantifiable scales. |
| Outcome: | The proposed system can provide accurate judgement and explain whether a source and generation are factually consistent. |
LITERA: An LLM Based Approach to Latin-to-English Translation (2025.findings-naacl)
Copied to clipboard
| Challenge: | Recent advances in large language models (LLMs) have shown promise in addressing these challenges across languages like Latin. |
| Approach: | They propose a Latin-to-English translation platform based on GPT-4o and GPT4o that combines a fine-tuned version of GPT-3o and a sophisticated algorithm to produce literal translations. |
| Outcome: | The model is based on two languages: Latin Interpretation and Translations into English for Research Assistance and GPT-4o. |
Investigating the Shortcomings of LLMs in Step-by-Step Legal Reasoning (2025.findings-naacl)
Copied to clipboard
Venkatesh Mishra, Bimsara Pathiraja, Mihir Parmar, Sat Chidananda, Jayanth Srinivasa, Gaowen Liu, Ali Payani, Chitta Baral
| Challenge: | Reasoning abilities of LLMs have been a key focus in recent years. |
| Approach: | They propose to use a college-level Multiple Choice Question-Answering task to identify LLM errors and evaluate their performance. |
| Outcome: | The proposed framework can be used in detailed error analysis of reasoning chains for logic-intensive complex tasks. |
Towards Long Context Hallucination Detection (2025.findings-naacl)
Copied to clipboard
Siyi Liu, Kishaloy Halder, Zheng Qi, Wei Xiao, Nikolaos Pappas, Phu Mon Htut, Neha Anna John, Yassine Benajiba, Dan Roth
| Challenge: | Large language models are prone to contextual hallucination, generating information that is either unsubstantiated or contradictory to the given context. |
| Approach: | They propose a dataset specifically designed for long-context hallucination detection. |
| Outcome: | The proposed architecture outperforms existing models while providing faster inference. |
How to Talk to Language Models: Serialization Strategies for Structured Entity Matching (2025.findings-naacl)
Copied to clipboard
| Challenge: | Entity matching (EM) identifies whether two data records refer to the same entity . however, its performance heavily depends on how structured entities are “talked” through serialized text. |
| Approach: | They propose a novel serialization scheme for entities with complex relations in knowledge graphs based on random walks and use open-source LLMs to encode sampled semantic walks for matching. |
| Outcome: | The proposed scheme achieves leading performance on EM in canonical and heterogeneous KGs. |
Accounting for Sycophancy in Language Model Uncertainty Estimation (2025.findings-naacl)
Copied to clipboard
| Challenge: | Effective human-machine collaboration requires machine learning models to externalize uncertainty. |
| Approach: | They propose a generalization of the definition of sycophancy bias and a new algorithm to account for scophancies in uncertainty estimation. |
| Outcome: | The proposed algorithm can account for sycophancy in uncertainty estimation process. |
Zero-Shot Keyphrase Generation: Investigating Specialized Instructions and Multi-sample Aggregation on Large Language Models (2025.findings-naacl)
Copied to clipboard
| Challenge: | Keyphrase generation is a long-standing NLP task for automatically generating keyphrases for a given document. |
| Approach: | They propose to use open-source instruction-tuned LLMs for keyphrase generation . they propose task-specific counterparts to self-consistency-style strategies for LLM . |
| Outcome: | The proposed model improves on existing models and shows significant benefits over baselines. |
Meta-Reasoning Improves Tool Use in Large Language Models (2025.findings-naacl)
Copied to clipboard
| Challenge: | External tools have proven effective at boosting the performance of large language models in knowledge-intensive tasks such as QA and math problem-solving. |
| Approach: | They propose a tool-based meta-reasONing system that makes a final decision based on a custom language modelling head. |
| Outcome: | The proposed system improves the performance of large language models in QA and math problem-solving tasks. |
CLERC: A Dataset for U. S. Legal Case Retrieval and Retrieval-Augmented Analysis Generation (2025.findings-naacl)
Copied to clipboard
Abe Bohan Hou, Orion Weller, Guanghui Qin, Eugene Yang, Dawn Lawrie, Nils Holzenberger, Andrew Blair-Stanek, Benjamin Van Durme
| Challenge: | a dataset of case law is used to train and evaluate models for writing legal analyses . current approaches struggle to find relevant cases and generate legal analyses, authors say . |
| Approach: | They build a dataset of case law to support information retrieval and retrieval-augmented generation. |
| Outcome: | The proposed dataset supports two important backbone tasks: retrieval (IR) and retrieval-augmented generation (RAG). |
GAIfE: Using GenAI to Improve Literacy in Low-resourced Settings (2025.findings-naacl)
Copied to clipboard
Allahsera Auguste Tapo, Nouhoum Coulibaly, Seydou Diallo, Sebastien Diarra, Christopher M Homan, Mamadou K. Keita, Michael Leventhal
| Challenge: | Illiteracy is a predictor of many negative social and personal outcomes in underresourced countries, where few books exist that are suitable for children to learn to read from. |
| Approach: | They propose to use generative AI to create culturally-engaging materials for learning in mali's vehicular language Bambara by multiplying the content by 10 times . authors propose to apply bias-aware tools to reduce illiteracy and improve learning outcomes through native language education. |
| Outcome: | The proposed toolchain and workflow can be adapted to address low literacy in mali using generative AI. |
Hard Emotion Test Evaluation Sets for Language Models (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing tests on emotion datasets do not show whether language models understand emotions or exploit supperficial lexical cues. |
| Approach: | They propose to use two existing emotion datasets to evaluate whether language models make inferential decisions for emotion detection. |
| Outcome: | The proposed test sets evaluate language models on emotion datasets. |
UCL-Bench: A Chinese User-Centric Legal Benchmark for Large Language Models (2025.findings-naacl)
Copied to clipboard
Ruoli Gan, Duanyu Feng, Chen Zhang, Zhihang Lin, Haochen Jia, Hao Wang, Zhenyang Cai, Lei Cui, Qianqian Xie, Jimin Huang, Benyou Wang
| Challenge: | Existing legal benchmarks focusing on knowledge and logic evaluate LLMs on various tasks in legal domain, but few have explored the practical application of LLM by actual users. |
| Approach: | They propose a Chinese user-centric legal benchmark that aims to assess the practical application of LLMs by real users. |
| Outcome: | The proposed model outperforms existing models on various tasks in legal domain but does not outperfect ChatGPT. |
MIDAS: Multi-level Intent, Domain, And Slot Knowledge Distillation for Multi-turn NLU (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing Large Language Models (LLMs) can generate coherent text, but they struggle to recognise user intent behind queries. |
| Approach: | They propose a novel approach leveraging multi-level intent, domain, and slot knowledge distillation for multi-turn NLU. |
| Outcome: | The proposed model improves multi-turn conversation understanding by integrating teacher teachers into a student model. |
A Practical Analysis of Human Alignment with *PO (2025.findings-naacl)
Copied to clipboard
| Challenge: | Prior research focused on identifying the best-performing method to varying hyperparameters . prior research focused primarily on a grid search, which can be impractical for general practitioners . |
| Approach: | They propose a preference optimization method that is more stable across hyperparameters and reduces the average response length. |
| Outcome: | The proposed method increases likelihood of achieving better results through various metrics, such as KL divergence and response length. |
Understanding Reference Policies in Direct Preference Optimization (2025.findings-naacl)
Copied to clipboard
| Challenge: | Direct Preference Optimization (DPO) is a widely used training method for the fine-tuning of large language models (LLMs). |
| Approach: | They explore the dependency on the reference model or policy in DPO and examine the necessity of the KL-constraint from the reference policies in D PO. |
| Outcome: | The proposed model is sensitive to the KL divergence constraint, and the reference policy can penalize deviations from the reference constraint. |
LLM-Coordination: Evaluating and Analyzing Multi-agent Coordination Abilities in Large Language Models (2025.findings-naacl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have demonstrated emergent common-sense reasoning and Theory of Mind (ToM) capabilities, making them promising candidates for developing coordination agents. |
| Approach: | They propose to use Large Language Models (LLMs) to analyze coordination models in Pure Coordination settings where agents must cooperate to maximize gains. |
| Outcome: | The proposed benchmark evaluates LLMs through two distinct tasks: Agentic Coordination and Coordination Question Answering. |
AssertionBench: A Benchmark to Evaluate Large-Language Models for Assertion Generation (2025.findings-naacl)
Copied to clipboard
| Challenge: | Assertions have been the de facto collateral for hardware for over a decade. |
| Approach: | They propose a benchmark to evaluate LLMs’ effectiveness for assertion generation quantitatively. |
| Outcome: | The proposed benchmark compares state-of-the-art LLMs with existing benchmarks and shows that they generate higher fractions of functionally correct assertions. |
On Reference (In-)Determinacy in Natural Language Inference (2025.findings-naacl)
Copied to clipboard
Sihao Chen, Chaitanya Malaviya, Alex Fabrikant, Hagai Taitelbaum, Tal Schuster, Senaka Buthpitiya, Dan Roth
| Challenge: | Using reference determinacy, models fail to recognize context mismatch in NLI examples . |
| Approach: | They propose a benchmark to identify reference ambiguity in NLI examples . they propose RD as a possible assumption for natural language inference . |
| Outcome: | The proposed benchmark identifies reference ambiguity in natural language inference examples . 80% false contradiction and >50% entailment predictions are found . |
DHP Benchmark: Are LLMs Good NLG Evaluators? (2025.findings-naacl)
Copied to clipboard
Yicheng Wang, Jiayi Yuan, Yu-Neng Chuang, Zhuoer Wang, Yingchi Liu, Mark Cusick, Param Kulkarni, Zhengping Ji, Yasser Ibrahim, Xia Hu
| Challenge: | Large Language Models (LLMs) are increasingly serving as evaluators in Natural Language Generation (NLG) tasks. |
| Approach: | They propose a framework that measures the discernment of Large Language Models (LLMs) across diverse NLG tasks. |
| Outcome: | The proposed framework provides quantitative discernment scores for LLMs across four NLG tasks. |
GraphEval36K: Benchmarking Coding and Reasoning Capabilities of Large Language Models on Graph Datasets (2025.findings-naacl)
Copied to clipboard
| Challenge: | Large language models (LLMs) have demonstrated significant capabilities in processing and understanding text data. |
| Approach: | They propose a structure-based instruction-based method to enhance LLM performance on complex graph tasks. |
| Outcome: | The proposed framework outperforms open-source models on graph problem-solving, but the gap is narrowing. |
SimulBench: Evaluating Language Models with Creative Simulation Tasks (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing benchmarks for large language models do not fully evaluate their potential for broad implementation. |
| Approach: | They propose to use a fixed LLM as a user agent to engage with an LLM to collect dialogues first under different tasks. |
| Outcome: | The proposed framework outperforms LLaMA-3-70b-Chat on 18.55% more cases. |
ReasoningRec: Bridging Personalized Recommendations and Human-Interpretable Explanations through LLM Reasoning (2025.findings-naacl)
Copied to clipboard
| Challenge: | Empirical evaluations demonstrate that ReasoningRec surpasses state-of-the-art methods by up to 12.5% in recommendation prediction while simultaneously providing human-intelligible explanations. |
| Approach: | They propose a reasoning-based recommendation framework that leverages Large Language Models to model users and items, focusing on preferences, aversions, and explanatory reasoning. |
| Outcome: | The proposed framework surpasses state-of-the-art methods by up to 12.5% in recommendation prediction while providing human-intelligible explanations. |
2D-DPO: Scaling Direct Preference Optimization with 2-Dimensional Supervision (2025.findings-naacl)
Copied to clipboard
Shilong Li, Yancheng He, Hui Huang, Xingyuan Bu, Jiaheng Liu, Hangyu Guo, Weixun Wang, Jihao Gu, Wenbo Su, Bo Zheng
| Challenge: | Existing methods that optimize for scalar scores or ranking reward ignore multi-dimensional nature of human preferences. |
| Approach: | They propose to extend the preference of Direct Preference Optimization to two dimensions: segments and aspects. |
| Outcome: | The proposed framework decomposes the overall objective into multi-segment and multi-aspect objectives. |
Demystifying the Power of Large Language Models in Graph Generation (2025.findings-naacl)
Copied to clipboard
Yu Wang, Ryan A. Rossi, Namyong Park, Nesreen K. Ahmed, Danai Koutra, Franck Dernoncourt, Tyler Derr
| Challenge: | Large Language Models (LLMs) have been used for graph discriminative tasks, but their potential for graph structure generation remains unexplored. |
| Approach: | They propose to use LLMs to generate graphs that optimize network properties by injecting domain expertise from network science into the code. |
| Outcome: | The proposed model generates graphs satisfying each property in different domains and compares it with established graph generative models across multiple domains. |
COIG-CQIA: Quality is All You Need for Chinese Instruction Fine-tuning (2025.findings-naacl)
Copied to clipboard
Yuelin Bai, Xeron Du, Yiming Liang, Leo Jin, Junting Zhou, Ziqiang Liu, Feiteng Fang, Mingshan Chang, Tianyu Zheng, Xincheng Zhang, Nuo Ma, Zekun Moore Wang, Ruibin Yuan, Haihong Wu, Hongquan Lin, Wenhao Huang, Jiajun Zhang, Chenghua Lin, Jie Fu, Min Yang, Shiwen Ni, Ge Zhang
| Challenge: | Existing datasets for Chinese instruction tuning are not well-aligned with Chinese users’ interaction patterns. |
| Approach: | They propose to use Chinese instruction tuning datasets to improve instruction fine-tuning for Chinese users. |
| Outcome: | The proposed dataset shows that Chinese models achieve competitive performance in diverse benchmarks. |
Gradient-guided Attention Map Editing: Towards Efficient Contextual Hallucination Mitigation (2025.findings-naacl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) often experience “contextual hallucination” where they prioritize self-generated content over input context, leading to a disregard for pertinent details. |
| Approach: | They propose a method that dynamically adjusts attention maps to enhance contextual relevance by using a trained classifier to identify attention maps likely to induce hallucinations. |
| Outcome: | The proposed approach reduces hallucinations across open-source models on summarization and open-book QA tasks. |
Alleviating Hallucinations of Large Language Models through Induced Hallucinations (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing studies have shown that large language models generate inaccurate or fabricated information, a phenomenon known as hallucinations. |
| Approach: | They propose a simple strategy to induce-then-contrast decode LLMs to enhance their factuality . they first induce hallucinations from the original model and penalize them . |
| Outcome: | The proposed strategy improves factuality of large language models across task formats, model sizes, and model families. |
MoDE: Effective Multi-task Parameter Efficient Fine-Tuning with a Mixture of Dyadic Experts (2025.findings-naacl)
Copied to clipboard
| Challenge: | Recent efforts have explored mixtures of LoRA modules for multi-task settings, but this study reveals redundancy in the down-projection matrix of these architectures. |
| Approach: | They propose a method to share down-projection matrix across tasks and employ atomic rank-one adapters coupled with routers that allow more sophisticated task-level specialization. |
| Outcome: | The proposed method outperforms state-of-the-art models on a SNI benchmark and provides a practical solution for deploying lightweight models. |
Unsupervised Sentence Representation Learning with Syntactically Aligned Negative Samples (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing approaches to sentence representation learning often encounter semantic inconsistencies and feature suppression. |
| Approach: | They propose a method for generating syntactically aligned negative (SAN) samples using a semantic importance-aware Masked Language Model (MLM) approach. |
| Outcome: | The proposed method produces negative samples with substantial textual overlap with the original sentences while conveying different meanings. |
Hierarchical Speculative Decoding with Dynamic Window (2025.findings-naacl)
Copied to clipboard
| Challenge: | Speculative decoding (SD) uses an efficient draft model to generate multiple tokens . previous methods depend on simple heuristics to select K or dynamically adjust the window size . |
| Approach: | They propose a framework that allows a draft model to generate multiple tokens . they propose HSDDW, which allows the draft model autonomously decide when to stop generating tokens. |
| Outcome: | The proposed framework outperforms existing state-of-the-art methods on four datasets. |
Q-FAKER: Query-free Hard Black-box Attack via Controlled Generation (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing adversarial approaches to verify the vulnerability of language models require numerous queries and the information on the target model. |
| Approach: | They propose a method that generates adversarial examples without accessing the target model and uses controlled generation techniques to generate adversarials. |
| Outcome: | The proposed method generates adversarial examples without accessing the target model and is highly transferable and high quality. |
PRDetect: Perturbation-Robust LLM-generated Text Detection Based on Syntax Tree (2025.findings-naacl)
Copied to clipboard
| Challenge: | Recent methods for detecting LLM-generated text have shown impressive performance, but in real-world scenarios, users often introduce perturbations to the text. |
| Approach: | They propose a method that detects syntactic trees that are minimally affected by perturbations and exhibit distinct differences between human-written and LLM-generated text. |
| Outcome: | The proposed method shows that it is significantly better against perturbations on the HC3 and GPT-3.5-mixed datasets and also has the shortest time expenditure. |
Enabling Natural Zero-Shot Prompting on Encoder Models via Statement-Tuning (2025.findings-naacl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) exhibit remarkable capabilities in zero-shot and few-shot settings, but they struggle with extending to few- shot and zero- shot settings due to their architectural design. |
| Approach: | They propose a technique that models discriminative tasks as a set of finite statements and trains an encoder model to discriminate between the potential statements to determine the label. |
| Outcome: | The proposed method achieves competitive performance compared to state-of-the-art LLMs with significantly fewer parameters. |
Faster Machine Translation Ensembling with Reinforcement Learning and Competitive Correction (2025.findings-naacl)
Copied to clipboard
| Challenge: | Recent approaches to ensembling neural machine translation models require inference across all candidate models, leading to significant computational overhead. |
| Approach: | They propose a reinforcement learning-based strategy that improves the CSB by selecting a small, fixed number of candidates and identifying optimal groups to pass to the fusion block for each input sentence. |
| Outcome: | The proposed approach improves the CSB by selecting a small, fixed number of candidates and identifying optimal groups to pass to the fusion block for each input sentence. |
Evaluating Numeracy of Language Models as a Natural Language Inference Task (2025.findings-naacl)
Copied to clipboard
| Challenge: | Recent advances in large language models (LLMs) have enhanced their capabilities to solve mathematical problems, but other aspects of numeracy remain underexplored. |
| Approach: | They propose to frame numeracy as a Natural Language Inference task to assess the models’ ability to understand both numbers and language contexts. |
| Outcome: | The proposed model outperforms smaller models in arithmetic tasks, indicating that mathematical reasoning cannot be generalized to other numeracy skills such as number comparison and normalization. |
Are Language Models Agnostic to Linguistically Grounded Perturbations? A Case Study of Indic Languages (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing studies do not focus on linguistically grounded attacks, but pre-trained models are susceptible to these perturbations. |
| Approach: | They propose to examine whether pre-trained language models are agnostic to linguistically grounded attacks . they find that PLMs are less susceptible to linguistic perturbations than non-linguistic ones . |
| Outcome: | The proposed model is agnostic to linguistically grounded attacks, but is less susceptible to linguist attacks than non-linguistic models. |
Do LLMs Have Distinct and Consistent Personality? TRAIT: Personality Testset designed for LLMs with Psychometrics (2025.findings-naacl)
Copied to clipboard
Seungbeen Lee, Seungwon Lim, Seungju Han, Giyeong Oh, Hyungjoo Chae, Jiwan Chung, Minju Kim, Beong-woo Kwak, Yeonsoo Lee, Dongha Lee, Jinyoung Yeo, Youngjae Yu
| Challenge: | Recent advances in Large Language Models (LLMs) have led to their adaptation as conversational agents. |
| Approach: | They propose a new benchmark that uses 8K multi-choice questions to assess the personality of Large Language Models. |
| Outcome: | The proposed personality test outperforms existing personality tests for LLMs in reliability and validity. |
Tell Me What You Know About Sexism: Expert-LLM Interaction Strategies and Co-Created Definitions for Zero-Shot Sexism Detection (2025.findings-naacl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) with chat interfaces are increasingly popular in various scientific fields, for a variety of tasks related to social science research questions. |
| Approach: | They propose to use large language models to combine human and machine expertise to improve their models' performance. |
| Outcome: | The proposed model performs better with co-created definitions than with expert-written definitions. |
The Role of Prosody in Spoken Question Answering (2025.findings-naacl)
Copied to clipboard
| Challenge: | lexical information is not available in most models, but prosody is important in understanding spoken language. |
| Approach: | They investigate the role of prosody in the process of answering a spoken question by isolating prosodic and lexical information from a natural speech dataset. |
| Outcome: | The proposed models can perform reasonably well on the SLUE-SQA-5 dataset, but when lexical information is available, models tend to predominantly rely on it. |
Target-Augmented Shared Fusion-based Multimodal Sarcasm Explanation Generation (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing systems that generate sarcastic explanations neglect the significance of the intended irony in generating them. |
| Approach: | They propose a multimodal sarcasm explanation that leverages the inter-modality relationships between an image and its caption to learn the intended irony for explanations. |
| Outcome: | The proposed model outperforms existing models on the sarcastic post-processing task and shows that it is more accurate than existing models. |
Seeds of Discourse: A Multilingual Corpus of Direct Quotations from African Media on Agricultural Biotechnologies (2025.findings-naacl)
Copied to clipboard
| Challenge: | a new study examines how media amplify messages around GM crops . agribusiness companies have placed ads for their products in newspapers . |
| Approach: | They present a multilingual corpora of 1,657 direct quotes from Africa-based news sources . they provide 665 instances annotated for Aspect-Based Sentiment Analysis . |
| Outcome: | The results of this study are available in English and French. |
Position Really Matters: Towards a Holistic Approach for Prompt Tuning (2025.findings-naacl)
Copied to clipboard
| Challenge: | Prompt tuning is effective in extracting knowledge from foundation models, but its effectiveness is uncertain. |
| Approach: | They propose a parametric prompt tuning strategy that dynamically determines different factors of prompts based on specific tasks or instances. |
| Outcome: | The proposed approach improves performance across a wide range of tasks including NLP, vision recognition, and vision-language tasks. |