Findings of the Association for Computational Linguistics: ACL 2025
Explicit vs. Implicit: Investigating Social Bias in Large Language Models through Self-Reflection (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods to quantify and quantify social biases in Large Language Models (LLMs) focus on explicit bias, with little attention to implicit bias. |
| Approach: | They propose a self-reflection-based evaluation framework that measures implicit bias and evaluates explicit bias by prompting LLMs to analyze their own generated content. |
| Outcome: | The proposed framework compares explicit and implicit biases in large language models . it demonstrates that explicit bias manifests as mild stereotypes, while implicit bias exhibits strong stereotypes. |
Beyond Perception: Evaluating Abstract Visual Reasoning through Multi-Stage Task (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing AVR benchmarks focus on single-step reasoning, emphasizing the end result but neglecting the multi-stage nature of reasoning process. |
| Approach: | They propose a multi-stage AVR benchmark based on RAVEN to assess reasoning across varying levels of complexity. |
| Outcome: | The proposed metric considers the correctness of intermediate steps in addition to the final outcomes. |
How Numerical Precision Affects Arithmetical Reasoning Capabilities of LLMs (2025.findings-acl)
Copied to clipboard
Guhao Feng, Kai Yang, Yuntian Gu, Xinyue Ai, Shengjie Luo, Jiacheng Sun, Di He, Zhenguo Li, Liwei Wang
| Challenge: | Despite the success of transformer-based large language models, understanding and enhancing their mathematical capabilities remains a significant challenge. |
| Approach: | They propose to use numerical precision as a key factor that influences LLMs' effectiveness in arithmetical tasks to determine their effectiveness. |
| Outcome: | The proposed models perform better in arithmetic tasks than transformer-based models with standard numerical precision. |
Diversifying the Expert Knowledge for Task-Agnostic Pruning in Sparse Mixture-of-Experts (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have outstanding performance by learning a large number of model parameters on large amounts of data. |
| Approach: | They propose a method of grouping and pruning similar experts to improve the model’s parameter efficiency by a range of natural language tasks. |
| Outcome: | The proposed method outperforms other model pruning methods on a range of natural language tasks. |
A Persona-Aware LLM-Enhanced Framework for Multi-Session Personalized Dialogue Generation (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing personalized dialogue models focus on dialogue history and personality information, reducing the responses’ consistency. |
| Approach: | They propose a Persona-Aware LLM-enAnCEd(PALACE) framework that generates responses consistent with dialogue history and personality information across multiple sessions to engage users’ interest in the dialogue. |
| Outcome: | The proposed framework outperforms the state-of-the-art methods in automatic and human evaluation metrics on the MSC and DuLeMon datasets. |
Exploring In-Image Machine Translation with Real-World Background (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing models for IIMT focus on simplified scenarios, which is far from reality and impractical for applications in the real world. |
| Approach: | They propose a model that separates the background and text-image from the source image and performs translation on the text- image directly. |
| Outcome: | The proposed model improves translation quality and visual effect in complex scenarios . it separates background and text-image from source image and performs translation on the text- image directly . |
BayesKD: Bayesian Knowledge Distillation for Compact LLMs in Constrained Fine-tuning Scenarios (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) have revolutionized various domains with their remarkable capabilities, but their massive parameter sizes pose significant challenges for fine-tuning and inference. |
| Approach: | They propose a Bayesian Knowledge Distillation framework for compact Large Language Models in resource-constrained fine-tuning scenarios that employs Logits Dual-Scaling, Knowledge Alignment Module, and Bayes Distillations Optimization. |
| Outcome: | The proposed framework outperforms baseline methods on various state-of-the-art LLMs, including LLaMA, Qwen2, Bloom, and Vicuna. |
GOLFer: Smaller LMs-Generated Documents Hallucination Filter & Combiner for Query Expansion in Information Retrieval (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs)-based query expansion for information retrieval necessitates larger, more advanced LLMs. |
| Approach: | They propose a method leveraging smaller open-source LMs for query expansion that augments queries with generated hypothetical documents with LLMs. |
| Outcome: | The proposed method outperforms existing methods on three web search and ten low-resource datasets and maintains competitive performance against larger LLMs. |
Exp4Fuse: A Rank Fusion Framework for Enhanced Sparse Retrieval using Large Language Model-based Query Expansion (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have shown potential in generating hypothetical documents for query expansion, but their efficacy is dependent on the quality of the generated documents. |
| Approach: | They propose a fusion ranking framework that uses zero-shot LLM-based query expansion to enhance sparse retrieval. |
| Outcome: | The proposed framework surpasses existing LLM-based query expansion methods on three MS MARCO-related datasets and seven low-resource datasets. |
Emo Pillars: Knowledge Distillation to Support Fine-Grained Context-Aware and Context-Less Emotion Classification (2025.findings-acl)
Copied to clipboard
| Challenge: | a recent study shows that sentiment analysis datasets lack context in which an opinion was expressed and are limited by a few emotion categories. |
| Approach: | They propose to ground an LLM-based model into a corpus of narratives to generate stories-character-centered utterances with unique contexts over 28 emotion classes. |
| Outcome: | The proposed model generates non-repetitive story-character-centered utterances with unique contexts over 28 emotion classes. |
Multi-Prompting Decoder Helps Better Language Understanding (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods to adapt Pre-trained Language Models to downstream tasks are limited by their inference APIs. |
| Approach: | They propose a multi-prompting decoding framework that query PLMs with multiple prompts . they propose to query Plms with optimal transport for hidden states and calibrated decoding for class scores . |
| Outcome: | The proposed framework achieves state-of-the-art results on multiple natural language understanding datasets under the few-shot setting. |
Visual Cues Enhance Predictive Turn-Taking for Two-Party Human Interaction (2025.findings-acl)
Copied to clipboard
| Challenge: | Predictive turn-taking models are based on speech, yet most rely on audio-only cues. |
| Approach: | They propose a multimodal PTTM which combines speech with visual cues including facial expression, head pose and gaze. |
| Outcome: | The proposed model outperforms the state-of-the-art audio-only turn-taking model (84% vs. 79% hold/shift prediction accuracy) it also outperformed the previous models which aggregated all holds and shifts, but grouped by duration of silence between turns. |
The Right Time Matters: Data Arrangement Affects Zero-Shot Generalization in Instruction Tuning (2025.findings-acl)
Copied to clipboard
Bingxiang He, Ning Ding, Cheng Qian, Jia Deng, Ganqu Cui, Lifan Yuan, Haiwen Hong, Huan-ang Gao, Longtao Huang, Hui Xue, Huimin Chen, Zhiyuan Liu, Maosong Sun
| Challenge: | Existing work on instruction tuning has focused on task level, without considering that tasks are artificially defined and, to LLMs, merely consist of tokens and representations. |
| Approach: | They propose a training data arrangement framework that allows for continual learning and loss reduction. |
| Outcome: | The proposed framework promotes continual learning and loss reduction on unseen tasks. |
MFinMeeting: A Multilingual, Multi-Sector, and Multi-Task Financial Meeting Understanding Evaluation Dataset (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing financial benchmarks rely on news articles, earnings reports, or announcements, making it challenging to capture the real-world dynamics of financial meetings. |
| Approach: | They propose a multilingual, multi-sector, and multi-task dataset called MFinMeeting that supports English, Chinese, and Japanese . |
| Outcome: | The proposed benchmark supports English, Chinese, and Japanese, enhancing comprehension of financial discussions in diverse linguistic contexts. |
ODDA: An OODA-Driven Diverse Data Augmentation Framework for Low-Resource Relation Extraction (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for low-resource relation extraction (LRE) lack diversity, leading to suboptimal performance. |
| Approach: | They propose to use large language models to augment relation extraction models by observing the RE model's behavior and replacing schema constraints with attribute constraints. |
| Outcome: | Experiments on three widely-used benchmarks show that the proposed method outperforms state-of-the-art methods while maintaining enhanced model stability. |
Detecting and Mitigating Challenges in Zero-Shot Video Summarization with Video LLMs (2025.findings-acl)
Copied to clipboard
Luca Cagliero, Lorenzo Vaiani, Eliana Pastor, Alkis Koudounas, Elena Baralis, Vittorio Mazzia, Sandro Pollastrini, Thomas Gueudre, Manuel Giollo, Daniele Amberti, Yue Wu
| Challenge: | Video Large Language Models (VLLMs) exhibit impressive zero-shot capabilities in video analysis, but their performance varies significantly depending on the LLM prompt, the characteristics of the video, and the properties of the training data and LLM architecture. |
| Approach: | They propose to use Chain-of-Thought prompting to inject knowledge extracted by external, lightweight models into video summarization benchmarks to evaluate their performance. |
| Outcome: | The proposed solutions improve summarization performance by injecting knowledge extracted by external, lightweight models. |
Entity Framing and Role Portrayal in the News (2025.findings-acl)
Copied to clipboard
Tarek Mahmoud, Zhuohan Xie, Dimitar Iliyanov Dimitrov, Nikolaos Nikolaidis, Purificação Silvano, Roman Yangarber, Shivam Sharma, Elisa Sartori, Nicolas Stefanovitch, Giovanni Da San Martino, Jakub Piskorski, Preslav Nakov
| Challenge: | a dataset of news articles containing 22 fine-grained characters is annotated for entity framing and role portrayal . the dataset includes 1,378 recent news articles in five languages focusing on the Ukraine-Russia War and climate change . |
| Approach: | They propose a multilingual and hierarchical corpus annotated for entity framing and role portrayal in news articles. |
| Outcome: | The proposed dataset includes 1,378 recent news articles in five languages focusing on the Ukraine-Russia War and climate change . the authors report evaluation results on state-of-the-art multilingual transformers and hierarchical zero-shot learning using LLMs at the level of a document, paragraph, and sentence . |
Derailer-Rerailer: Adaptive Verification for Efficient and Reliable Language Model Reasoning (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing prompting methods struggle with complex tasks and reasoning stability, limiting their practical deployment. |
| Approach: | They propose a framework that adaptively balances reasoning accuracy and computational efficiency by employing a lightweight Derailer mechanism to assess reasoning stability and selectively triggers an advanced Rerailer verification process only when necessary. |
| Outcome: | The proposed framework achieves significant accuracy improvements (8-11%) while maintaining 2-3 times better efficiency than existing verification methods. |
Leveraging Large Language Models for Conversational Multi-Doc Question Answering: The First Place of WSDM Cup 2024 (2025.findings-acl)
Copied to clipboard
| Challenge: | WSDM Cup 2024 presents a challenge for conversational multi-doc question answering using large language models . a hybrid training strategy is developed to make the most of in-domain unlabeled data . |
| Approach: | They propose a conversational multi-doc question answering challenge in WSDM Cup 2024 . they adapt LLMs to the task, then devise a hybrid training strategy to make the most of unlabeled data. |
| Outcome: | The proposed approach ranked 1st in the WSDM Cup 2024 challenge . it exploits the superior natural language understanding and generation capability of Large Language Models . |
TreeRAG: Unleashing the Power of Hierarchical Storage for Enhanced Knowledge Retrieval in Long Documents (2025.findings-acl)
Copied to clipboard
| Challenge: | Traditional RAG frameworks struggle to retrieve all relevant knowledge points . a new approach to retrieve long documents is proposed to improve performance in NLP . |
| Approach: | They propose a tree-based approach to document knowledge retrieval that preserves hierarchical structure . treeRAG is a key technique for enhancing the text generation capabilities of Large Language Models . |
| Outcome: | The proposed approach improves recall quality and precision compared to existing methods and better performance to question-answering tasks. |
Attention with Dependency Parsing Augmentation for Fine-Grained Attribution (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing fine-grained attribution methods rely on model-internal similarity metrics but lack a fine-grain representation of the data. |
| Approach: | They propose to use model-internal similarity metrics to validate RAG-generated content . they aggregate token-wise evidence through set union operations and integrate dependency parsing to enrich the semantic completeness of target spans. |
| Outcome: | The proposed method outperforms all prior works in the validation of RAG-generated content. |
ASTRO: Automatic Strategy Optimization For Non-Cooperative Dialogues (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for optimizing dialogues require substantial human effort for strategy optimization. |
| Approach: | They propose a fully automated solution that leverages large language models’ self-envolving capabilities to optimize dialogue strategies. |
| Outcome: | The proposed solution significantly improves on baseline models across non-cooperative dialogue tasks, highlighting the potential for autonomously developing such agents without human intervention. |
Defensive Prompt Patch: A Robust and Generalizable Defense of Large Language Models against Jailbreak Attacks (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent advances in large language models (LLMs) have showcased their ability to understand and generate text akin to human interaction. |
| Approach: | They propose a prompt-based defense mechanism specifically designed to protect LLMs against jailbreak attacks by introducing jailbreak prompts into malicious queries. |
| Outcome: | Empirical results show that the proposed defense outperforms existing defense strategies in balancing safety and utility while maintaining high utility. |
GUM-SAGE: A Novel Dataset and Approach for Graded Entity Salience Prediction (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for graded entity salience are subjective but lack consistency. |
| Approach: | They propose a method for graded entity salience that combines subjective judgments and summarization-based methods that define saliency as mention-worthiness in a summary. |
| Outcome: | The proposed approach outperforms existing methods and shows stronger correlation with human summaries and alignments. |
Verifying the Steps of Deductive Reasoning Chains (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models have been shown to improve the reasoning capabilities of the models. |
| Approach: | They propose to automate verification of individual reasoning steps in a logical deductive Chain-of-Thought. |
| Outcome: | The proposed method can detect unsound reasoning steps fairly well, but under-performs symbolic methods. |
Translate With Care: Addressing Gender Bias, Neutrality, and Reasoning in Large Language Model Translations (2025.findings-acl)
Copied to clipboard
| Challenge: | addressing gender bias and maintaining logical coherence in machine translation remains challenging, especially when translating between natural gender languages, like English, and genderless languages, such as Persian, Indonesian, and Finnish. |
| Approach: | They propose a dataset to assess translation systems' performance in six low- to mid-resource languages and a translation dataset to examine gender bias and logical coherence. |
| Outcome: | The Translate-with-Care dataset, comprising 3,950 challenging scenarios across six low- to mid-resource languages, reveals a universal struggle in translating genderless content, resulting in gender stereotyping and reasoning errors. |
Utilizing Semantic Textual Similarity for Clinical Survey Data Feature Selection (2025.findings-acl)
Copied to clipboard
| Challenge: | Survey data often contains many features that lead to high-dimensional inputs for machine learning models. |
| Approach: | They propose to use semantic textual similarity scores to score features in surveys . they compare STS scores to a statistically-based scoring approach . |
| Outcome: | The proposed method can be used in clinical surveys to score features . it can be applied to a clinical dataset and to NIH All of Us dataset . |
Distance between Relevant Information Pieces Causes Bias in Long-Context LLMs (2025.findings-acl)
Copied to clipboard
Runchu Tian, Yanghao Li, Yuepeng Fu, Siyang Deng, Qinyu Luo, Cheng Qian, Shuo Wang, Xin Cong, Zhong Zhang, Yesai Wu, Yankai Lin, Huadong Wang, Xiaojiang Liu
| Challenge: | Positional biases in large language models hinder their ability to process long inputs. |
| Approach: | They propose a benchmark to assess positional bias in large language models involving multiple pieces of relevant information. |
| Outcome: | The proposed benchmark assesses the performance of long-context language models by examining their models with different input lengths and tasks. |
Variable Layerwise Quantization: A Simple and Effective Approach to Quantize LLMs (2025.findings-acl)
Copied to clipboard
Razvan-Gabriel Dumitru, Vikas Yadav, Rishabh Maheshwary, Paul Ioan Clotan, Sathwik Tejaswi Madhusudhan, Mihai Surdeanu
| Challenge: | a meta quantization approach quantizes different layers of a large language model at different bit levels. |
| Approach: | They propose a meta quantization approach that quantizes different layers of a large language model at different bit levels. |
| Outcome: | The proposed method quantizes the most important layers to higher bit precision and less important layers at lower bits. |
Why Are Positional Encodings Nonessential for Deep Autoregressive Transformers? A Petroglyph Revisited (2025.findings-acl)
Copied to clipboard
| Challenge: | Autoregressive Transformer language models do not require explicit positional encodings (PEs) this is because a cascade of (permutation invariant) set processors collectively exhibit sequence-sensitive behavior in the autoregressively setting. |
| Approach: | They propose to explain why autoregressive Transformers require explicit positional encodings (PEs) this property has been known since early efforts adopting the Transformer for language modeling . |
| Outcome: | The proposed model can distinguish sequences with permuted tokens without the need for explicit PEs. |
CRPO: Confidence-Reward Driven Preference Optimization for Machine Translation (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) have shown great potential in natural language processing tasks, but their application to machine translation remains challenging due to pretraining on predominantly English-centric datasets. |
| Approach: | They propose a method that combines reward scores with model confidence to improve model selection for fine-tuning. |
| Outcome: | The proposed method outperforms existing methods in translation accuracy and data efficiency. |
Talking Point based Ideological Discourse Analysis in News Events (2025.findings-acl)
Copied to clipboard
Nishanth Sridhar Nakshatri, Nikhil Mehta, Siyi Liu, Sihao Chen, Daniel Hopkins, Dan Roth, Dan Goldwasser
| Challenge: | Existing models of ideological discourse analysis fail to capture the key elements that shape real-world narratives and lack the ability to integrate contextual information required for understanding abstract ideological views. |
| Approach: | They propose a framework motivated by the theory of ideological discourse analysis to analyze news articles related to real-world events. |
| Outcome: | The proposed framework can generate ideology-specific viewpoints (partisan perspectives) it can be used to generate event snapshots, a visual way of interpreting event discourse. |
FlashBack: Efficient Retrieval-Augmented Language Modeling for Fast Inference (2025.findings-acl)
Copied to clipboard
| Challenge: | Retrieval-Augmented Language Modeling (RALM) is a popular approach for large language models. |
| Approach: | They propose a modular RALM that integrates large language models with documents from an external corpus to improve inference efficiency. |
| Outcome: | The proposed method improves inference efficiency with appending context pattern while maintaining decent performance after fine-tuning by Low-Rank Adaption. |
CMQCIC-Bench: A Chinese Benchmark for Evaluating Large Language Models in Medical Quality Control Indicator Calculation (2025.findings-acl)
Copied to clipboard
Guangya Yu, Yanhao Li, Zongying Jiang, Yuxiong Jin, Li Dai, Yupian Lin, Ruihui Hou, Weiyan Zhang, Yongqi Fan, Qi Ye, Jingping Liu, Tong Ruan
| Challenge: | Medical quality control indicators are essential to assess the qualifications of healthcare institutions for medical services. |
| Approach: | They propose a Chinese electronic medical records-based dataset for MQCIC and propose CF-IR method that disentangles clinical fact verification and inferential rule reasoning actions. |
| Outcome: | The proposed method outperforms Chain-of-Thought methods on 20 representative LLMs, covering general and medical models. |
ConKE: Conceptualization-Augmented Knowledge Editing in Large Language Models for Commonsense Reasoning (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing knowledge editing methods face limited knowledge coverage in existing knowledge bases, infeasibility of annotating labels for an overabundance of commonsense knowledge, and strict knowledge formats. |
| Approach: | They propose a framework that integrates conceptualization and instantiation into the KE pipeline for LLMs to enhance their commonsense reasoning capabilities. |
| Outcome: | The proposed framework diagnoses implausible commonsense knowledge within an LLM and augments the source knowledge to be edited with conceptualization for stronger generalizability. |
Exploring Multi-Modal Data with Tool-Augmented LLM Agents for Precise Causal Discovery (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing statistical causal discovery methods rely on observational data and often overlook the semantic cues inherent in cause-and-effect relationships. |
| Approach: | They propose a multi-agent system powered by tool-augmented Large Language Models that can combine data from multiple modalities and integrate multi-modal data for knowledge-driven reasoning. |
| Outcome: | The proposed system has two agents: a Data Augmentation agent that retrieves and processes modality-augmented data, and a Causal Constraint agent that integrates multi-modal data for knowledge-driven reasoning. |
PARSQL: Enhancing Text-to-SQL through SQL Parsing and Reasoning (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models have made significant strides in text-to-SQL tasks, but small language models struggle to accurately interpret natural language questions due to resource limitations. |
| Approach: | They propose a SQL parser that extracts constraints from SQL to generate sub-SQLs . they use a rule-based and LLM-based method to generate step-by-step SQL explanations based on the results . |
| Outcome: | The proposed framework outperforms models with the same model size on BIRD and Spider benchmarks. |
Probing the Geometry of Truth: Consistency and Generalization of Truth Directions in LLMs Across Logical Transformations and Question Answering Tasks (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) are trained on vast corpora that contain substantial knowledge but their outputs often contain confidently stated inaccuracies. |
| Approach: | They propose to encode truthfulness as a distinct linear feature, termed the "truth direction", which can classify truthfulness reliably. |
| Outcome: | The proposed model can generalize to logical transformations, question-answering tasks, in-context learning, and external knowledge sources. |
Comparing Bad Apples to Good Oranges Aligning Large Language Models via Joint Preference Optimization (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent studies have shown that acquiring human preferences by comparing generations is not effective for large language models. |
| Approach: | They propose a preference optimization objective that elicits preferences jointly over the instruction-response pairs. |
| Outcome: | The proposed approach outperforms prior preference optimizations by 5.2% and 3.3% in summarization and open-ended dialogue datasets. |
TestAgent: An Adaptive and Intelligent Expert for Human Assessment (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing adaptive testing methods face several challenges due to mechanized nature of most algorithms and noisy response data. |
| Approach: | They propose to use large language models to enhance adaptive testing through interactive engagement to capture test-takers’ responses and anomalies. |
| Outcome: | The proposed agent achieves more accurate results with 20% fewer questions than state-of-the-art baselines and testers preferred it in speed, smoothness, and other dimensions. |
SPICA: Retrieving Scenarios for Pluralistic In-Context Alignment (2025.findings-acl)
Copied to clipboard
| Challenge: | In-context learning only considers similarity when drawing few-shot examples and not cross-group differences in values. |
| Approach: | They propose a framework that accounts for group-level differences during in-context example retrieval by using scenario banks, group-informed retrieval metrics, and in-constraint alignment prompts. |
| Outcome: | The proposed framework improves on an alignment task with groups seeing up to a +0.16 point improvement on a 5 point scale. |
First-Step Advantage: Importance of Starting Right in Multi-Step Math Reasoning (2025.findings-acl)
Copied to clipboard
| Challenge: | Language models can solve complex reasoning tasks better by learning to generate rationales for their predictions. |
| Approach: | They propose to use a larger model to guide smaller models to start . this allows them to generate rationales for their predictions when correct . |
| Outcome: | The proposed method improves performance on multistep reasoning datasets over multiple smaller models. |
Evaluating Instructively Generated Statement by Large Language Models for Directional Event Causality Identification (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing studies focus on identifying existence of causality between two event mentions, but the direction of causalities is crucial for understanding the causal relation. |
| Approach: | They propose to instruct a GLM to generate causality statements and identify directional event causality by evaluating the generated statements. |
| Outcome: | The proposed method significantly outperforms state-of-the-art methods even with fewer training data. |
CoinMath: Harnessing the Power of Coding Instruction for Math LLM (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have shown strong performance in solving mathematical problems, with code-based solutions proving particularly effective. |
| Approach: | They propose a learning strategy to enhance mathematical reasoning by diversifying the coding styles of code-based rationales. |
| Outcome: | The proposed learning strategy outperforms its baseline model, MAmmoTH, which uses code-based solutions. |
Profiling News Media for Factuality and Bias Using LLMs and the Fact-Checking Methodology of Human Experts (2025.findings-acl)
Copied to clipboard
| Challenge: | Important efforts to characterize news media outlets in terms of their political bias and factuality are labor-intensive and prone to human biases. |
| Approach: | They propose a method that emulates criteria used by professional fact-checkers to assess the factuality and political bias of an entire outlet. |
| Outcome: | The proposed method improves on baselines and with multiple LLMs. |
Structured Discourse Representation for Factual Consistency Verification (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods to verify factual consistency of text capture a performance gap compared with sentence-level entailment. |
| Approach: | They propose a method that combines structured discourse information extraction with a classifier, FDSpotter, for factual consistency verification. |
| Outcome: | The proposed method achieves competitive performance on two tasks: data-to-text generation and text summarisation. |
SHARP: Unlocking Interactive Hallucination via Stance Transfer in Role-Playing LLMs (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing studies on social interactions neglect hallucination while struggling with poor generalizability and implicit character fidelity judgments. |
| Approach: | They propose a generalizable and explicit paradigm for uncovering interactive patterns of Large Language Models across diverse worldviews by defining interactive hallucination through stance transfer and SHARP, a benchmark built by extracting relations from commonsense knowledge graphs. |
| Outcome: | The proposed paradigm is generalizable and explicit and demonstrates its effectiveness and stability. |
Understanding the Gap: an Analysis of Research Collaborations in NLP and Language Documentation (2025.findings-acl)
Copied to clipboard
| Challenge: | despite 20 years of NLP work, practical use of this work remains vanishingly scarce. |
| Approach: | They propose to use interviews and surveys to examine the lack of NLP adoption in LD . they find that linguists and language communities have little or no use of Nlp in their work . |
| Outcome: | a new study shows that linguists and language researchers are not using NLP in LD . the findings highlight the importance of misaligned professional incentives and LD software . |
PersonaBench: Evaluating AI Models on Understanding Personal Information through Accessing (Synthetic) Private User Data (2025.findings-acl)
Copied to clipboard
Juntao Tan, Liangwei Yang, Zuxin Liu, Zhiwei Liu, Rithesh R N, Tulika Manoj Awalgaonkar, Jianguo Zhang, Weiran Yao, Ming Zhu, Shirley Kokane, Silvio Savarese, Huan Wang, Caiming Xiong, Shelby Heinecke
| Challenge: | Existing research lacks direct access to such data, making benchmarking difficult due to privacy concerns. |
| Approach: | They propose a synthetic data pipeline that generates realistic user profiles and private documents and a benchmark to evaluate models' ability to understand personal information. |
| Outcome: | The proposed pipeline generates realistic user profiles and private documents, enabling PersonaBench, a benchmark for evaluating models’ ability to understand personal information. |
Leveraging Variation Theory in Counterfactual Data Augmentation for Optimized Active Learning (2025.findings-acl)
Copied to clipboard
| Challenge: | Active Learning (AL) allows users to provide focused annotations to integrate human preferences and domain knowledge into machine learning models. |
| Approach: | They propose a counterfactual data augmentation approach inspired by Variation Theory to generate targeted variations along key conceptual dimensions. |
| Outcome: | The proposed approach achieves significantly higher performance when there are fewer annotated data, showing it can address the cold start problem in Active Learning. |
ORBIT: Cost-Effective Dataset Curation for Large Language Model Domain Adaptation with an Astronomy Case Study (2025.findings-acl)
Copied to clipboard
| Challenge: | General-purpose models lack depth for expert-level tasks because of limited domain-specific information. |
| Approach: | They propose a method for curating domain-specific datasets from noisy web sources to improve model performance. |
| Outcome: | The proposed model outperforms the baseline model on the astronomy benchmark and on the AstroBench. |
Serial Position Effects of Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Serial position effects (SPE) are well-documented cognitive biases in human behavior. |
| Approach: | They propose to use binary choices instead of multiple choices where feasible . they also suggest limiting prompt length and placing crucial information at the beginning of prompts . |
| Outcome: | The proposed framework shows that the effects are widespread across LLMs and the proposed mitigation methods are effective. |
scRAG: Hybrid Retrieval-Augmented Generation for LLM-based Cross-Tissue Single-Cell Annotation (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) have demonstrated impressive potential in a wide range of fields, including biology, genomics and healthcare. |
| Approach: | They propose a framework that integrates advanced LLM-based RAG techniques into cross-tissue single-cell annotation. |
| Outcome: | The proposed framework outperforms baseline models, generalist models, domain-specific methods, and trained classifiers on a cross-tissue dataset. |
Can Large Language Models Address Open-Target Stance Detection? (2025.findings-acl)
Copied to clipboard
| Challenge: | Stance detection (SD) identifies a text’s position towards a target, typically labeled as favor, against, or none. |
| Approach: | They introduce Open-Target Stance Detection (OTSD) which aims to determine the position of a text towards a target, typically labeled as favor, against, or none. |
| Outcome: | The proposed model outperforms the only existing task, Target-Stance Extraction (TSE), which benefits from predefined targets. |
Improve Language Model and Brain Alignment via Associative Memory (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing studies have shown that associative memory is essential for language comprehension and comprehension. |
| Approach: | They propose to integrate associative memory into language models to improve alignment . they find alignment is improved in brain regions closely related to associativ memory processing . |
| Outcome: | The proposed model improves in brain regions closely related to associative memory processing. |
Towards Reliable Large Audio Language Model (2025.findings-acl)
Copied to clipboard
Ziyang Ma, Xiquan Li, Yakun Song, Wenxi Chen, Chenpeng Du, Jian Wu, Yuanzhe Chen, Zhuo Chen, Yuping Wang, Yuxuan Wang, Xie Chen
| Challenge: | Recent advances in large audio language models (LALMs) have demonstrated impressive results and promising prospects in universal understanding and reasoning across speech, music, and general sound. |
| Approach: | They propose to use training-free and training-based methods to enhance LALM reliability to different extents. |
| Outcome: | The proposed methods improve the reliability of large audio language models to different extents. |
Large Vocabulary Size Improves Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing studies have investigated the properties of internal layers in large language models, but no studies have defined the vocabulary size. |
| Approach: | They propose a method to use a new vocabulary instead of the pre-defined one in a continual training scenario. |
| Outcome: | The proposed method outperforms the model with the pre-defined vocabulary in a continual training scenario. |
MUSE: A Multimodal Conversational Recommendation Dataset with Scenario-Grounded User Profiles (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing research focuses solely on text, leaving a gap with practical applications. |
| Approach: | They propose to synthesize a multimodal conversational recommendation dataset using multimodal large language models to automatically synthesized data from 7,000 conversations in the Clothing domain. |
| Outcome: | The proposed dataset contains 83,148 utterances from 7,000 conversations centered around the Clothing domain. |
Machine Translation Models are Zero-Shot Detectors of Translation Direction (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing approaches to detect the translation direction of parallel text are lacking in the machine translation community. |
| Approach: | They propose an unsupervised approach to detection of translation direction of parallel texts . they use a simple hypothesis that p(translation|original)>p(original|translation) they confirm the approach is effective for high-resource language pairs . |
| Outcome: | The proposed approach achieves document-level accuracies of 82–96% for NMT-produced translations and 60–81% for human translations, based on the model used. |
Do Robot Snakes Dream like Electric Sheep? Investigating the Effects of Architectural Inductive Biases on Hallucination (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) have a tendency to hallucinate false or misleading information, limiting their reliability. |
| Approach: | They examine how architecture-based inductive biases affect the propensity to hallucinate . they find that the models are more reliable and more reliable than traditional models . |
| Outcome: | The proposed models can be used to train and train large language models that are factual or able to explain themselves through their knowledge. |
GenTool: Enhancing Tool Generalization in Language Models through Zero-to-One and Weak-to-Strong Simulation (2025.findings-acl)
Copied to clipboard
Jie He, Jennifer Neville, Mengting Wan, Longqi Yang, Hui Liu, Xiaofeng Xu, Xia Song, Jeff Z. Pan, Pei Zhou
| Challenge: | Large Language Models (LLMs) can expand their capabilities by integrating external tools. |
| Approach: | They propose a training framework that prepares LLMs for diverse generalization challenges in tool utilization. |
| Outcome: | The proposed framework improves the tool-usage capabilities of LLMs by up to 8B parameters, surpassing GPT-4o. |
SWE-Fixer: Training Open-Source LLMs for Effective and Efficient GitHub Issue Resolution (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models excel in code generation benchmarks, but these benchmarks focus on single-file scenarios with constrained context scope. |
| Approach: | They propose an open-source framework to effectively resolve GitHub issues using a code file retrieval module and a model-based code editing module. |
| Outcome: | The proposed approach achieves state-of-the-art performance on two GitHub benchmarks. |
GlyphPattern: An Abstract Pattern Recognition for Vision-Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing benchmarks for abstract pattern recognition are easier because they do not involve a natural language description of the pattern. |
| Approach: | They present a dataset that pairs human-written descriptions of visual patterns with three visual presentation styles. |
| Outcome: | The proposed benchmark pairs human-written and human-verified patterns with three visual presentation styles. |
FitCF: A Framework for Automatic Feature Importance-guided Counterfactual Example Generation (2025.findings-acl)
Copied to clipboard
Qianli Wang, Nils Feldhus, Simon Ostermann, Luis Felipe Villa-Arenas, Sebastian Möller, Vera Schmitt
| Challenge: | Existing frameworks for counterfactual examples are lacking for many tasks. |
| Approach: | They propose a faithful approach for leveraging important words from feature attribution methods to generate counterfactual examples in a zero-shot setting. |
| Outcome: | The proposed framework outperforms state-of-the-art frameworks on many tasks. |
From Misleading Queries to Accurate Answers: A Three-Stage Fine-Tuning Method for LLMs (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods focus on correcting the output but overlook the ability of LLMs to detect and correct misleading content in the input itself. |
| Approach: | They propose a three-stage fine-tuning method that improves LLMs' ability to detect and correct misleading information in input queries. |
| Outcome: | The proposed method improves accuracy and factuality of LLM responses while also reducing hallucinations. |
Separate the Wheat from the Chaff: A Post-Hoc Approach to Safety Re-Alignment for Fine-Tuned Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models achieve effective safety alignment at the time of release, but fine-tuning often compromises safety mechanisms. |
| Approach: | They propose a method that performs safety realignment for large language models . they identify unsafe delta parameters from the fine-tuned models and recalibrate the retained parameters . |
| Outcome: | The proposed method improves safety performance on safety benchmarks and jailbreak attacks while maintaining their performance on downstream tasks. |
Nuclear Deployed!: Analyzing Catastrophic Risks in Decision-making of Autonomous LLM Agents (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) are evolving into autonomous decision-makers, raising concerns about catastrophic risks in high-stakes domains, particularly in Chemical, Biological, Radiological and Nuclear (CBRN) . |
| Approach: | They propose a framework that is carefully constructed to effectively and naturally expose catastrophic risks in high-stakes domains such as CBRN. |
| Outcome: | The proposed framework exposes LLM agents to catastrophic behaviors and deception without being deliberately induced. |
MoRE: A Mixture of Low-Rank Experts for Adaptive Multi-Task Learning (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent advances in Large Language Models (LLMs) have revolutionized various domains, offering unprecedented performance across numerous tasks. |
| Approach: | They propose a new Mixture of Low-Rank Experts (MoRE) for multi-task PEFT to improve performance of LLMs with fewer parameters. |
| Outcome: | The proposed method improves performance over multiple tasks and no additional inference cost. |
Lunar Twins: We Choose to Go to the Moon with Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Lunar Twins is the first LLM designed specifically for lunar exploration . Lunar GenData is a multi-agent collaborative workflow for generating lunar instructions . |
| Approach: | They propose a framework that combines both large and small LLMs and Lunar GenData, which integrates real data from Chang'e lunar missions. |
| Outcome: | Experimental validation shows that the proposed framework enhances domain expertise and reveals indications of embodied intelligence potential. |
SPHERE: An Evaluation Card for Human-AI Systems (2025.findings-acl)
Copied to clipboard
Dora Zhao, Qianou Ma, Xinran Zhao, Chenglei Si, Chenyang Yang, Ryan Louie, Ehud Reiter, Diyi Yang, Tongshuang Wu
| Challenge: | Existing evaluation methods and standards for human-AI systems are unclear, especially for large language models. |
| Approach: | They propose an evaluation card SPHERE which provides a template for evaluation protocols . they outline current evaluation practices and areas for improvement . |
| Outcome: | The evaluation card provides a template for designing evaluation protocols . it outlines current evaluation practices and areas for improvement . |
Data-Centric Improvements for Enhancing Multi-Modal Understanding in Spoken Conversation Modeling (2025.findings-acl)
Copied to clipboard
| Challenge: | Conversational assistants are increasingly popular across diverse real-world applications . speech data constitute high-dimensional signals that are difficult to model even for frontier models . |
| Approach: | They propose a data-centric customization approach for enhancing multimodal understanding in conversational speech modeling. |
| Outcome: | The proposed model achieves state-of-the-art on the Spoken-SQuAD benchmark using 10% of training data with open-weight models. |
Question-Aware Knowledge Graph Prompting for Enhancing Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for large language models require costly fine-tuning or retrieve noisy KG information. |
| Approach: | They propose to generate KG-based input embedding prefixes as soft prompts but fail to account for question relevance, resulting in noisy prompts. |
| Outcome: | The proposed model outperforms state-of-the-art methods across multiple datasets. |
UQ-Merge: Uncertainty Guided Multimodal Large Language Model Merging (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing models merging methods often lead to suboptimal performance due to harmful models . et al., 2018; 59: 59-64. |
| Approach: | They propose an uncertainty-guided MLLM merging algorithm that integrates models into a single MLML. |
| Outcome: | The proposed algorithm improves on held-in and held-out vision-language benchmarks. |
AQuAECHR: Attributed Question Answering for European Court of Human Rights (2025.findings-acl)
Copied to clipboard
| Challenge: | LLMs are widely used for information seeking, but their generated responses often suffer from hallucinations, hindering their widespread adoption in high stakes domains such as law. |
| Approach: | They propose to attribute legal question answering to an actual source to improve factuality and verifiability of the answer. |
| Outcome: | The proposed framework improves the factuality and verifiability of legal question answering by combining a dataset from ECHR case law guides with an LLM-based filtering pipeline. |
Leveraging Unit Language Guidance to Advance Speech Modeling in Textless Speech-to-Speech Translation (2025.findings-acl)
Copied to clipboard
Yuhao Zhang, Xiangnan Ma, Kaiqi Kou, Peizhuo Liu, Weiqiao Shan, Benyou Wang, Tong Xiao, Yuxin Huang, Zhengtao Yu, JingBo Zhu
| Challenge: | Existing textless speech-to-speech translation models have two main challenges: 1) learning cross-modal features and 2) learning alignment of difference languages in long sequences. |
| Approach: | They propose a unit language to overcome two main modeling challenges . they propose task prompt modeling to utilize the unit language in guiding the modeling process. |
| Outcome: | The proposed language improves over a strong baseline and achieves comparable performance to models trained with text. |
Ponder & Press: Advancing Visual GUI Agent towards General Computer Control (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing multimodal large language models (MLLMs) lack visual inputs to ground objects, limiting flexibility across diverse software environments and platforms. |
| Approach: | They propose a divide-and-conquer framework for general computer control that uses only visual inputs to create a purely human-like interaction paradigm. |
| Outcome: | The proposed framework outperforms existing models by +22.5% on the ScreenSpot GUI grounding benchmark. |
LogicGame: Benchmarking Rule-Based Reasoning Abilities of Large Language Models (2025.findings-acl)
Copied to clipboard
Jiayi Gui, Yiming Liu, Jiale Cheng, Xiaotao Gu, Xiao Liu, Hongning Wang, Yuxiao Dong, Jie Tang, Minlie Huang
| Challenge: | Large Language Models (LLMs) have demonstrated notable capabilities across various tasks, showcasing complex problem-solving abilities. |
| Approach: | They propose a benchmark to evaluate the rule-based logical reasoning capabilities of Large Language Models (LLMs) they create simulated scenarios in which models execute or plan operations to achieve specific outcomes. |
| Outcome: | The proposed benchmark evaluates the performance of large language models on a variety of scenarios with varying difficulty levels. |
LLM-Based Multi-Agent Systems are Scalable Graph Generative Models (2025.findings-acl)
Copied to clipboard
Jiarui Ji, Runlin Lei, Jialing Bi, Zhewei Wei, Xu Chen, Yankai Lin, Xuchen Pan, Yaliang Li, Bolin Ding
| Challenge: | Social graphs are mathematical structures stem from pairwise interactions between entities through nodes and edges. |
| Approach: | They propose a framework for dynamic, text-attributed social graph generation that simulates the temporal node and edge generation processes for zero-shot social graphs. |
| Outcome: | The proposed framework improves macroscopic graph structure metrics by 11% . the proposed model can generate graphs with up to 100,000 nodes or 10 million edges . |
AD-LLM: Benchmarking Large Language Models for Anomaly Detection (2025.findings-acl)
Copied to clipboard
Tiankai Yang, Yi Nian, Li Li, Ruiyao Xu, Yuangang Li, Jiaqi Li, Zhuo Xiao, Xiyang Hu, Ryan A. Rossi, Kaize Ding, Xia Hu, Yue Zhao
| Challenge: | Anomaly detection (AD) is an important machine learning task with many real-world uses, including fraud detection, medical diagnosis, and industrial monitoring. |
| Approach: | They propose a benchmark that evaluates how large language models (LLMs) can help with NLP anomaly detection. |
| Outcome: | The proposed model can perform zero-shot detection without tasks-specific training, data augmentation and model selection, and it can suggest unsupervised AD models. |
RTADev: Intention Aligned Multi-Agent Framework for Software Development (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) are efficient assistants to humans in software development tasks, but they can cause errors during the development process. |
| Approach: | They propose an intention aligned multi-agent framework that ensures that all agents work based on a consensus. |
| Outcome: | The proposed framework reduces errors and improves the quality of generated software code. |
TACO-RL: Task Aware Prompt Compression Optimization with Reinforcement Learning (2025.findings-acl)
Copied to clipboard
Shivam Shandilya, Menglin Xia, Supriyo Ghosh, Huiqiang Jiang, Jue Zhang, Qianhui Wu, Victor Rühle, Saravan Rajmohan
| Challenge: | Existing prompt compression techniques rely on sub-optimal metrics such as information entropy or model it as a task-agnostic token classification problem that fails to capture task-specific information. |
| Approach: | They propose a task-aware prompt compression method that leverages existing Transformer encoders and a lightweight REINFORCE algorithm to ensure low latency requirements. |
| Outcome: | The proposed method improves task performance by 8% - 189% on three diverse and challenging tasks over state-of-the-art techniques while satisfying the same compression rate and latency requirements. |
A Character-Centric Creative Story Generation via Imagination (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing narrative generation models lack diversity and character depth, but they are inadequate for human creativity. |
| Approach: | They propose a novel story generation framework called CCI that leverages images to create stories that are diverse and creative in their themes and richer in content. |
| Outcome: | The proposed framework significantly improves various aspects of the stories’ creativity. |
Proverbs Run in Pairs: Evaluating Proverb Translation Capability of Large Language Model (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent research has demonstrated that large language models (LLMs) can translate cultural elements in languages such as idioms and proverbs. |
| Approach: | They propose to use large language models to translate culturally rooted proverbs in conversation and between languages with similar cultural backgrounds to compare their results. |
| Outcome: | The proposed models can achieve good translation between languages with similar cultural backgrounds and outperform NMT models in proverb translation. |
Towards Efficient LLM Grounding for Embodied Multi-Agent Collaboration (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for grounding large language models suffer from inefficient querying . Existing approaches that rely on physical verification or self-reflection suffer from excessive querying. |
| Approach: | They propose a framework that introduces Reinforced Advantage feedback for efficient self-refinement of plans. |
| Outcome: | The proposed framework surpasses baselines in success rate and significantly decreases interaction steps of agents and query rounds of LLMs. |
UAQFact: Evaluating Factual Knowledge Utilization of LLMs on Unanswerable Questions (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing datasets to assess LLMs' performance on unanswerable questions lack factual knowledge support. |
| Approach: | They propose a bilingual unanswerable question dataset with auxiliary factual knowledge created from a Knowledge Graph and two new tasks to measure LLMs' ability to utilize internal and external factual information. |
| Outcome: | The proposed datasets show that LLMs do not consistently perform well even when they have factual knowledge stored. |
Exploring Knowledge Filtering for Retrieval-Augmented Discriminative Tasks (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent studies have focused on generative tasks, while its potential in discriminative tasks remains largely unexplored. |
| Approach: | They propose a framework that incorporates knowledge filtering and prediction fusion mechanisms to improve model performance. |
| Outcome: | The proposed framework improves model performance on discriminative tasks by filtering out harmful knowledge and integrating it into the input context. |
Group then Scale: Dynamic Mixture-of-Experts Multilingual Language Model (2025.findings-acl)
Copied to clipboard
| Challenge: | Experimental results show that language group specialization on experts improves multilingual performance. |
| Approach: | They propose to dynamically group and scale up parameters of multilingual Large Language Models while boosting positive transfer among similar languages. |
| Outcome: | The proposed method reduces negative transfer between languages and boosts performance on 18 to 128 languages. |
Beyond Verbal Cues: Emotional Contagion Graph Network for Causal Emotion Entailment (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent studies have focused on identifying the causes of emotions by understanding verbal contextual utterances, but this study often lacks recognizing the underlying emotional stimuli present in these utterrances. |
| Approach: | They propose an Emotional Contagion Graph Network that simulates the impact of non-verbal emotional cues on the counterpart’s emotions. |
| Outcome: | The proposed model is compared with state-of-the-art models on a benchmark dataset and the results are encouraging. |
Critic-CoT: Boosting the Reasoning Abilities of Large Language Model via Chain-of-Thought Critic (2025.findings-acl)
Copied to clipboard
Xin Zheng, Jie Lou, Boxi Cao, Xueru Wen, Yuqiu Ji, Hongyu Lin, Yaojie Lu, Xianpei Han, Debing Zhang, Le Sun
| Challenge: | Existing approaches to improve the reasoning performance of large language models rely on intuitive instance-level feedback, which limits the reasoning capabilities. |
| Approach: | They propose a framework that pushes LLMs toward System-2-like critic capability by using a step-wise CoT reasoning paradigm and automatic construction of weak-supervision data without human annotation. |
| Outcome: | The proposed model significantly improves task-solving performance by filtering out invalid solutions or iterative refinement. |
Systematic Generalization in Language Models Scales with Information Entropy (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing benchmarks for assessing compositional behavior are unclear on how to measure the difficulty of a systematic generalization problem. |
| Approach: | They propose a framework for measuring entropy in a sequence-to-sequence task and a method for measuring it. |
| Outcome: | The proposed framework scales with the entropy of the distribution of component parts in the training data. |
The Inverse Scaling Effect of Pre-Trained Language Model Surprisal Is Not Due to Data Leakage (2025.findings-acl)
Copied to clipboard
| Challenge: | Language models (LMs) have been shown to flexibly capture many linguistic regularities from raw text, but the source stimuli of reading time datasets are often naturalistic text that are available online. |
| Approach: | They propose to replicate the negative relationship between language model size and the fit of surprisal to reading times using models trained on ‘leakage-free’ data that overlaps only minimally with the reading time corpora. |
| Outcome: | The proposed models show that language models trained on 'leakage-free' data are not driven by data leakage. |
Logical Consistency is Vital: Neural-Symbolic Information Retrieval for Negative-Constraint Queries (2025.findings-acl)
Copied to clipboard
Ganlin Xu, Zhoujia Zhang, Wangyi Mei, Jiaqing Liang, Weijia Lu, Xiaodong Zhang, Zhifei Yang, Xiaofeng Ma, Yanghua Xiao, Deqing Yang
| Challenge: | Current dense retrieval methods compute similarities between dense vectors but overlook the real query intents. |
| Approach: | They propose a neuro-symbolic information retrieval method that leverages first-order logic to optimize the embeddings of naive natural language by considering the logical consistency between queries and documents. |
| Outcome: | The proposed method outperforms existing methods on negative-constraint queries under zero-shot and low-resource retrieval tasks. |
‘No’ Matters: Out-of-Distribution Detection in Multimodality Multi-Turn Interactive Dialogue Download PDF (2025.findings-acl)
Copied to clipboard
| Challenge: | Out-of-distribution (OOD) detection is essential for multimodal learning systems . a novel scoring framework is proposed to efficiently detect OOD in multi-round long dialogues . |
| Approach: | They propose a scoring framework that integrates visual language models with a score framework that detects OOD in two key scenarios. |
| Outcome: | The proposed framework detects OOD in two key scenarios: mismatches between dialogue and image input pair and previously unseen labels. |
Event Pattern-Instance Graph: A Multi-Round Role Representation Learning Strategy for Document-Level Event Argument Extraction (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing role-based span selection strategies ignore interrelations between events . authors propose a multi-round role representation learning strategy for document-level event argument extraction . |
| Approach: | They propose a pattern-instance graph to capture role semantics embedded in various associations . they also propose re-inventing the role representations learned from previous analyzed documents . |
| Outcome: | The proposed model captures role semantics embedded in various associations . iteratively updates representations of role nodes and edges to enrich their semantic information . the model improves prediction performance in subsequent rounds of span selection . |
EXECUTE: A Multilingual Benchmark for LLM Token Understanding (2025.findings-acl)
Copied to clipboard
| Challenge: | EXECUTE is an expandable X(Cross)-Lingual Extension of CUTE that can be expanded to any language. |
| Approach: | They extend the CUTE benchmark to more languages with diverse scripts and writing systems, introducing EXECUTE. |
| Outcome: | The extended framework allows expansion to any language. |
Explainable Hallucination through Natural Language Inference Mapping (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) often generate hallucinated content, making it crucial to identify and quantify inconsistencies in their outputs. |
| Approach: | They propose a framework that maps entailment and contradiction relations between inputs and outputs using a natural language inference model. |
| Outcome: | The proposed framework outperforms state-of-the-art methods by five percentage points while providing clear, interpretable explanations. |
HopRAG: Multi-Hop Reasoning for Logic-Aware Retrieval-Augmented Generation (2025.findings-acl)
Copied to clipboard
| Challenge: | Traditional retrieval systems focus on lexical or semantic similarity rather than logical relevance. |
| Approach: | They propose a new RAG framework that augments retrieval with logical reasoning . hopRAG uses a retrieve-reason-prune mechanism to explore multi-hop neighbors . |
| Outcome: | The proposed framework outperforms conventional retrieval systems and state-of-the-art benchmarks on multi-hop QA tasks. |
Double Entendre: Robust Audio-Based AI-Generated Lyrics Detection via Multi-View Fusion (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for detecting AI-generated music are weak and vulnerable to audio perturbations. |
| Approach: | They propose a multimodal late-fusion pipeline that combines automatically transcribed sung lyrics and speech features capturing lyrics related information within the audio. |
| Outcome: | The proposed method outperforms existing detectors while being more robust to audio perturbations. |
Don’t Miss the Forest for the Trees: Attentional Vision Calibration for Large Vision Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Vision Language Models suffer from hallucinations, attributing incorrect or misleading features to images. |
| Approach: | They propose a test-time approach that recalibrates the influence of blind tokens . they identify blind token by analyzing layer-wise attention distributions over image tokens. |
| Outcome: | The proposed approach reduces hallucinations in large vision language models . it uses a contrastive decoding strategy to balance the influence of blind tokens . |
SATA: A Paradigm for LLM Jailbreak via Simple Assistive Task Linkage (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods to design sophisticated instructions for the LLM to follow, or rely on multiple iterations, could hinder the performance and efficiency of jailbreaks. |
| Approach: | They propose a simple assistive task linkage paradigm which masks harmful keywords within malicious queries and uses a masked language model task to encode the semantics of the mangled keywords. |
| Outcome: | The proposed paradigm can effectively circumvent LLM safeguards and elicit harmful responses. |
Chain-Talker: Chain Understanding and Rendering for Empathetic Conversational Speech Synthesis (2025.findings-acl)
Copied to clipboard
| Challenge: | Current generative CSS models face interpretability limitations due to insufficient emotional perception and redundant discrete speech coding. |
| Approach: | They propose a framework that aligns synthesized speech with the emotional context of user-agent interactions to achieve empathy. |
| Outcome: | The proposed framework produces more expressive speech than existing methods on three datasets. |
Parameter-Efficient Fine-Tuning via Circular Convolution (2025.findings-acl)
Copied to clipboard
| Challenge: | Low-Rank Adaptation (LoRA) has gained popularity for fine-tuning large foundation models, but its intrinsic low-rank characteristic may limit its performance. |
| Approach: | They propose a low-rank adaptive method that uses low-ranked matrices to represent weight changes. |
| Outcome: | The proposed method reduces trainable parameters and mitigates heavy memory consumption associated with full delta matrices by sequentially multiplying mathbf A and mathbb B with the activation. |
Alleviating Hallucinations in Large Language Models via Truthfulness-driven Rank-adaptive LoRA (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods to improve truthfulness are training-free without modifying the LLM itself. |
| Approach: | They propose a rank-adaptive LoRA method to improve LLM truthfulness that allocates ranks according to truthfulness correlations of LLM modules. |
| Outcome: | The proposed method outperforms state-of-the-art methods on the LLM family and makes the performance of 7B LLMs exceed GPT-4. |
ScEdit: Script-based Assessment of Knowledge Editing (2025.findings-acl)
Copied to clipboard
Xinye Li, Zunwen Zheng, Qian Zhang, Dekai Zhuang, Jiabao Kang, Liyan Xu, Qingbin Liu, Xi Chen, Zhiying Tu, Dianhui Chu, Dianbo Sui
| Challenge: | Knowledge Editing (KE) has gained increasing attention, yet current evaluation frameworks do not integrate KE into real-world application scenarios. |
| Approach: | They propose a script-based benchmark which encompasses both counterfactual and temporal edits and integrates token-level and text-level evaluation methods. |
| Outcome: | The proposed method combines token-level and text-level evaluation methods with a new fact-based evaluation framework. |
SafeRoute: Adaptive Model Selection for Efficient and Accurate Safety Guardrails in Large Language Models (2025.findings-acl)
Copied to clipboard
Seanie Lee, Dong Bok Lee, Dominik Wagner, Minki Kang, Haebin Seong, Tobias Bocklet, Juho Lee, Sung Ju Hwang
| Challenge: | Deploying large language models (LLMs) requires robust safety guard models to detect and block harmful user prompts. |
| Approach: | They propose a binary router that selectively applies a larger safety guard model to the data that the router considers hard. |
| Outcome: | The proposed method outperforms baselines on multiple benchmark datasets on hard and hard examples. |
Moderation Matters: Measuring Conversational Moderation Impact in English as a Second Language Group Discussion (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing tools for ESL assessment focus on writing skills and lack in support for dynamic spoken interactions. |
| Approach: | They propose an approach that integrates automatic ESL dialogue assessment and a framework that categorizes moderation strategies to assess conversational engagement and moderation effectiveness. |
| Outcome: | The proposed approach integrates automatic ESL dialogue assessment and categorizes moderation strategies. |
Measuring Bias and Agreement in Large Language Model Presupposition Judgments (2025.findings-acl)
Copied to clipboard
| Challenge: | Identifying linguistic bias in text requires the identification of explicit statements and presuppositions . large language models can be used to detect subtle forms of bias with no clear lexical signals . |
| Approach: | They propose to prompt large language models to evaluate presuppositions across texts . they find that LLMs may inadvertently reflect societal biases when identifying presuposed content . |
| Outcome: | The proposed model can be used to detect linguistic biases in text, but its accuracy is unclear . linguistic factors associated with human-model alignment suggest biase influenced by gender and ideology. |
Harnessing PDF Data for Improving Japanese Large Multimodal Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Multimodal Models (LMMs) have demonstrated strong performance in English, but their effectiveness in Japanese remains limited due to the lack of high-quality training data. |
| Approach: | They propose a pipeline that leverages pretrained models to extract image-text pairs from PDFs . they use layout analysis, OCR, and vision-language pairing to enrich the training data . |
| Outcome: | The proposed pipeline extracts image-text pairs from Japanese PDFs, eliminating manual annotations. |
EnerGIZAr: Leveraging GIZA++ for Effective Tokenizer Initialization (2025.findings-acl)
Copied to clipboard
| Challenge: | Continual pre-training has long been considered the default strategy for adapting models to non-English languages, but struggles with initializing new embeddings, especially for non-Latin scripts. |
| Approach: | They propose a method that leverages statistical word alignment techniques to improve continual pre-training by leveraging word alignment matrix between source and target tokens. |
| Outcome: | The proposed method outperforms existing methods on key NLP tasks including POS tagging, Sentiment Analysis, NLI, and NER in Hindi, Basque, Arabic and Korean. |
AMEX: Android Multi-annotation Expo Dataset for Mobile GUI Agents (2025.findings-acl)
Copied to clipboard
Yuxiang Chai, Siyuan Huang, Yazhe Niu, Han Xiao, Liang Liu, Guozhi Wang, Dingyu Zhang, Shuai Ren, Hongsheng Li
| Challenge: | a new dataset is being developed to improve the capabilities of mobile GUI-control agents. |
| Approach: | They propose a dataset designed for generalist mobile GUI-control agents . they use screenshots from popular mobile applications to create a detailed GUI-annotated dataset . |
| Outcome: | The Android Multi-annotation EXpo (AMEX) is a large-scale dataset for generalist mobile GUI-control agents . it includes screenshots from popular mobile applications, which are annotated at multiple levels . |
Drop Dropout on Single Epoch Language Model Pretraining (2025.findings-acl)
Copied to clipboard
| Challenge: | Initial dropout was seen as a breakthrough regularization technique that reduced overfitting, yet single-epoch pretraining tasks common to modern LLMs yield minimal overfit. |
| Approach: | They propose to use dropout during single-epoch pretraining to reduce overfitting in language modeling, morpho-syntax, question answering, and MNLI to improve performance. |
| Outcome: | The results show that dropout is not used in large LLMs and improves performance in language modeling, morpho-syntax, question answering, and MNLI. |
Robust and Minimally Invasive Watermarking for EaaS (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing watermarking methods use a target embedding to create watermarks, but this method results in each embeddable having the same component, making it difficult to remove the watermark. |
| Approach: | They propose to use embedding watermarks to protect EaaS from model extraction attacks . eaas is vulnerable to model extraction, highlighting the need for copyright protection . |
| Outcome: | The proposed method can watermark embeddings against model extraction attacks without sacrificing the quality of the embeddables. |
Task-Informed Anti-Curriculum by Masking Improves Downstream Performance on Text (2025.findings-acl)
Copied to clipboard
| Challenge: | Masked language modeling is widely adopted, but the process of selecting tokens for masking is random and the percentage of masked tokens is typically fixed for the entire training process. |
| Approach: | They propose to adjust the masking ratio based on a task-informed anti-curriculum learning scheme to mask useful and harmful tokens. |
| Outcome: | The proposed approach improves the ability of the model to focus on key task-relevant features, contributing to statistically significant performance gains across tasks. |
CARMO: Dynamic Criteria Generation for Context Aware Reward Modelling (2025.findings-acl)
Copied to clipboard
Taneesh Gupta, Shivam Shandilya, Xuchao Zhang, Rahul Madhavan, Supriyo Ghosh, Chetan Bansal, Huaxiu Yao, Saravan Rajmohan
| Challenge: | Reward modeling in large language models is susceptible to reward hacking . flawed reward signals often lead to outputs that optimize for spurious correlates . |
| Approach: | They propose a new approach that generates dynamic, context-relevant criteria to ground the reward model prior to producing reward scores. |
| Outcome: | The proposed approach generates dynamic, context-relevant criteria to ground the model prior to producing reward scores. |
SLAM-Omni: Timbre-Controllable Voice Interaction System with Single-Stage Training (2025.findings-acl)
Copied to clipboard
Wenxi Chen, Ziyang Ma, Ruiqi Yan, Yuzhe Liang, Xiquan Li, Ruiyang Xu, Zhikang Niu, Yanqiao Zhu, Yifan Yang, Zhanxun Liu, Kai Yu, Yuxuan Hu, Jinyu Li, Yan Lu, Shujie Liu, Xie Chen
| Challenge: | a new spoken dialogue system with single-stage training is demonstrating its low latency and high quality . SLAM-Omni achieves zero-shot timbre control by modeling spoken language with semantic tokens . |
| Approach: | They propose a timbre-controllable, end-to-end voice interaction system with single-stage training. |
| Outcome: | The proposed system outperforms previous models on 4 GPUs with limited data. |
C2LEVA: Toward Comprehensive and Contamination-Free Language Model Evaluation (2025.findings-acl)
Copied to clipboard
Yanyang Li, Wong Tin Long, Cheung To Hung, Jianqiao Zhao, Duo Zheng, Liu Ka Wai, Michael R. Lyu, Liwei Wang
| Challenge: | Recent advances in large language models (LLMs) have shown significant promise, yet their evaluation raises concerns regarding data contamination due to the lack of access to proprietary training data. |
| Approach: | They propose a bilingual benchmark that offers a holistic evaluation and systematic contamination prevention. |
| Outcome: | The proposed evaluations of 15 open-source and proprietary models show that they are reliable and free of data contamination. |
Texts or Images? A Fine-grained Analysis on the Effectiveness of Input Representations and Models for Table Question Answering (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing studies on table question answering use either serialized text or images . lack of controlled setups limits fine-grained distinctions between these approaches. |
| Approach: | They propose a method selecting table representations dynamically and compare their performance to previous studies. |
| Outcome: | The proposed method improves performance 10% compared to using both representations indiscriminately. |
Adaptive-VP: A Framework for LLM-Based Virtual Patients that Adapts to Trainees’ Dialogue to Facilitate Nurse Communication Training (2025.findings-acl)
Copied to clipboard
Keyeun Lee, Seolhee Lee, Esther Hehsun Kim, Yena Ko, Jinsu Eun, Dahee Kim, Hyewon Cho, Haiyi Zhu, Robert E. Kraut, Eunyoung E. Suh, Eun-mee Kim, Hajin Lim
| Challenge: | standardized patient (SP) simulations are costly and inflexible. |
| Approach: | They propose a framework that leverages large language models to dynamically adapt VP behavior based on trainee input. |
| Outcome: | The proposed framework reflects real-world communication skills and produces more natural and realistic interactions than existing methods. |
Enhancing Multimodal Unified Representations for Cross Modal Generalization (2025.findings-acl)
Copied to clipboard
Hai Huang, Yan Xia, Shengpeng Ji, Shulei Wang, Hanting Wang, Minghui Fang, Jieming Zhu, Zhenhua Dong, Sashuai Zhou, Zhou Zhao
| Challenge: | Existing studies on discrete unified representations overlook important distinctions between different dimensions of features. |
| Approach: | They propose to use a codebook to optimize unified representations from pretraining and fine- and coarse-grained disentangling to optimize the representations. |
| Outcome: | The proposed methods improve the interpretability of multimodal unified representations . they use training-free optimization of codebook and fine and coarse cross-modal disentangling . |
Domain Regeneration: How well do LLMs match syntactic properties of text domains? (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent improvements in large language models have improved their ability to approximate distributions . authors find that LLMs can suffer from model collapse due to domain considerations based on pretraining . |
| Approach: | They use open source LLMs to regenerate permissively licensed English text from Wikipedia and news text. |
| Outcome: | The proposed model can faithfully match the human-generated distributions in a semantically-controlled setting. |
Structural Deep Encoding for Table Question Answering (2025.findings-acl)
Copied to clipboard
| Challenge: | Tabular data is a common data format, but many models flatten the structure of a table into a sequence of tokens, resulting in computational costs and over-fitting issues. |
| Approach: | They propose to use special tokens to mark rows and columns, structured embeddings, and sparse attention patterns to preserve structural information of tabular data. |
| Outcome: | The proposed models enhance computational efficiency and preserve structural integrity, leading to better overall performance. |
MPL: Multiple Programming Languages with Large Language Models for Information Extraction (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing research focuses on Python for code-style simulation, overlooking the potential of other widely-used PLs during the supervised fine-tuning phase. |
| Approach: | They propose a framework that incorporates programming languages into IE tasks . they introduce function-prompt with virtual running to simulate code-style inputs . |
| Outcome: | The proposed framework exploits the potential of different programming languages during the supervised fine-tuning phase. |
Self-Critique Guided Iterative Reasoning for Multi-hop Question Answering (2025.findings-acl)
Copied to clipboard
Zheng Chu, Huiming Fan, Jingchang Chen, Qianyu Wang, Mingda Yang, Jiafeng Liang, Zhongjie Wang, Hao Li, Guo Tang, Ming Liu, Bing Qin
| Challenge: | Large language models (LLMs) have demonstrated remarkable reasoning capabilities, but they still face challenges in knowledge-intensive multi-hop reasoning. |
| Approach: | They propose a method that uses self-critique feedback to guide iterative reasoning by enabling iteration and self-evaluation of its intermediate reasoning steps. |
| Outcome: | The proposed method surpasses the previous SOTA by 8.6% on three multi-hop reasoning datasets. |
Anchored Answers: Unravelling Positional Bias in GPT-2’s Multiple-Choice Questions (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) exhibit a positional bias, particularly an even worse “anchored bias” in the GPT-2 family, where they consistently favour the first choice ‘A’ in MCQs. |
| Approach: | They propose to use the “logit lens” method to trace and modify the internal modules within GPT-2 models responsible for this bias. |
| Outcome: | The proposed approach mitigates the positional bias and improves the accuracy of the GPT-2 model across multiple datasets. |
Failing Forward: Improving Generative Error Correction for ASR with Synthetic Data and Retrieval Augmentation (2025.findings-acl)
Copied to clipboard
Sreyan Ghosh, Mohammad Sadegh Rasooli, Michael Levit, Peidong Wang, Jian Xue, Dinesh Manocha, Jinyu Li
| Challenge: | Generative Error Correction (GEC) is a powerful post-processing method to boost the performance of Automatic Speech Recognition systems. |
| Approach: | They propose a method to augment GEC models with retrieved entities to improve accuracy in out-of-domain and out-od scenarios. |
| Outcome: | The proposed method outperforms baseline models on multiple datasets and settings. |
LTRAG: Enhancing Autoformalization and Self-refinement for Logical Reasoning with Thought-Guided RAG (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) have shown promise in natural language reasoning, especially with techniques like chain-of-thought prompting. |
| Approach: | They propose a framework to enhance autoformalization and self-refinement for logical reasoning with Retrieval-Augmented Generation (RAG) by building knowledge bases of thought-guided examples. |
| Outcome: | The proposed framework outperforms Logic-LM and LINC on FOLIO and AR-LSAT, and achieves an accuracy gain of 13% over Logic LM and the proposed methods on GPT-4 and AR LSAT. |
Eta-WavLM: Efficient Speaker Identity Removal in Self-Supervised Speech Representations Using a Simple Linear Equation (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for learning meaningful representations from unannotated data are resource-intensive and degrade other speech components. |
| Approach: | They propose a method that decomposes SSL representations into speaker-specific components and generates speaker disentangled representations. |
| Outcome: | The proposed method achieves speaker independence and improves on state-of-the-art methods. |
MathCoder-VL: Bridging Vision and Code for Enhanced Multimodal Mathematical Reasoning (2025.findings-acl)
Copied to clipboard
Ke Wang, Junting Pan, Linda Wei, Aojun Zhou, Weikang Shi, Zimu Lu, Han Xiao, Yunqiao Yang, Houxing Ren, Mingjie Zhan, Hongsheng Li
| Challenge: | Large Language Models (LMMs) struggle with simple tasks such as geometry, e.g., arithmetic, and reasoning. |
| Approach: | They propose to leverage code as supervision for cross-modal alignment . they propose to use FigCodifier and ImgCode-8.6M to synthesize novel mathematical figures . |
| Outcome: | The proposed model surpasses GPT-4o and Claude 3.5 Sonnet in the geometry problem-solving subset of MathVista, achieving improvements of 8.9% and 9.2%. |
MlingConf: A Comprehensive Study of Multilingual Confidence Estimation on Large Language Models (2025.findings-acl)
Copied to clipboard
Boyang Xue, Hongru Wang, Rui Wang, Sheng Wang, Zezhong Wang, Yiming Du, Bin Liang, Wenxuan Zhang, Kam-Fai Wong
| Challenge: | Existing studies on LLM confidence estimations in languages other than English have been limited to English. |
| Approach: | They propose to use question-related language to prompt LLMs to assess their confidence in large language models. |
| Outcome: | The proposed model improves on question-related language prompts for LS tasks, while English exhibits notable linguistic dominance in confidence estimations. |
COMPKE: Complex Question Answering under Knowledge Editing (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing benchmarks for knowledge editing do not accurately evaluate how well models apply knowledge in real-life situations. |
| Approach: | They propose a benchmark to evaluate how well updated models apply new knowledge in real-life situations. |
| Outcome: | The proposed method achieves 39.47 accuracy on GPT-4o-mini but drops significantly to 3.83 on Qwen2.5-3B. |
RaaS: Reasoning-Aware Attention Sparsity for Efficient LLM Reasoning (2025.findings-acl)
Copied to clipboard
Junhao Hu, Wenrui Huang, Weidong Wang, Zhenwen Li, Tiancheng Hu, Zhixia Liu, Xusheng Chen, Tao Xie, Yizhou Shan
| Challenge: | Large Language Models (LLMs) have demonstrated strong capabilities across various domains, but their large-scale deployment faces a major obstacle: the high computational cost of long-sequence inference. |
| Approach: | They propose an algorithm that retains key-value vectors until they are no longer needed to solve reasoning tasks. |
| Outcome: | The proposed algorithm achieves high accuracy with O(L) time but O(N) memory complexities. |
One-for-All Pruning: A Universal Model for Customized Compression of Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing pruning methods for large language models (LLMs) focus on achieving high compression rates while maintaining model performance. |
| Approach: | They propose a Univeral Model for Customized Compression (UniCuCo) which introduces a StratNet that learns to map arbitrary requests to their optimal pruning strategy. |
| Outcome: | The proposed model is 28 times faster than baselines in processing 64 requests, while maintaining comparable accuracy to baselines. |
CLaMP 3: Universal Music Information Retrieval Across Unaligned Modalities and Unseen Languages (2025.findings-acl)
Copied to clipboard
Shangda Wu, Guo Zhancheng, Ruibin Yuan, Junyan Jiang, SeungHeon Doh, Gus Xia, Juhan Nam, Xiaobing Li, Feng Yu, Maosong Sun
| Challenge: | Music information retrieval (MIR) is a field that aims at developing computational tools for processing, organizing, and accessing music data. |
| Approach: | They propose a framework that aligns music modalities with multilingual text in a shared representation space. |
| Outcome: | Experiments show CLaMP 3 performs state-of-the-art on multiple MIR tasks . it surpasses baselines and shows excellent generalization in multimodal and multilingual contexts . |
PFDial: A Structured Dialogue Instruction Fine-tuning Method Based on UML Flowcharts (2025.findings-acl)
Copied to clipboard
Ming Zhang, Yuhui Wang, Yujiong Shen, Tingyi Yang, Changhao Jiang, Yilong Wu, Shihan Dou, Qinhao Chen, Zhiheng Xi, Zhihao Zhang, Yi Dong, Zhen Wang, Zhihui Fei, Mingyang Wan, Tao Liang, Guojun Ma, Qi Zhang, Tao Gui, Xuanjing Huang
| Challenge: | Large Language Models (LLMs) have shown remarkable progress in dialogue and reasoning, but they struggle to solve strictly constrained dialogue tasks. |
| Approach: | They construct a dataset that contains 12,705 high-quality Chinese dialogue instructions from 440 flowcharts containing 5,055 process nodes. |
| Outcome: | The proposed model outperforms GPT-4o models on backward transitions and outperformed GPT-42 models on the same dataset. |
Listening to Patients: Detecting and Mitigating Patient Misreport in Medical Dialogue System (2025.findings-acl)
Copied to clipboard
| Challenge: | Medical Dialogue Systems (MDSs) aim to provide automated healthcare support through natural language interactions between patients and system agents. |
| Approach: | They propose a framework that detects misreports and mitigates them by generating controlled clarifying questions. |
| Outcome: | The proposed framework can detect misreports and mitigate them through generating controlled clarifying questions. |
Do Language Models Understand the Cognitive Tasks Given to Them? Investigations with the N-Back Paradigm (2025.findings-acl)
Copied to clipboard
| Challenge: | Cognitive tasks originally developed for humans are now increasingly used to study language models. |
| Approach: | They propose to challenge the best-performing model with progressively harder versions of the task and experiment with alternative prompting strategies before analyzing model attentions. |
| Outcome: | The proposed model underperforms on 2-back and 3-back tasks on GPT 3.5 and has a working 1 Code available at https://github.com/hxiaoyang/lm-nback. |
Graph-guided Cross-composition Feature Disentanglement for Compositional Zero-shot Learning (2025.findings-acl)
Copied to clipboard
Yuxia Geng, Runkai Zhu, Jiaoyan Chen, Jintai Chen, Xiang Chen, Zhuo Chen, Shuofei Qiao, Yuxiang Wang, Xiaoliang Xu, Sheng-Jun Huang
| Challenge: | Disentanglement of visual features of primitives (i.e., attributes and objects) has shown exceptional results in Compositional Zero-shot Learning (CZSL). |
| Approach: | They propose a solution that takes multiple compositions as inputs and constrains disentangled primitive features to be general across compositions. |
| Outcome: | The proposed architecture significantly improves performance on three popular CZSL benchmarks and has been verified by solid ablation studies. |
Training Long-Context LLMs Efficiently via Chunk-wise Optimization (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent advances in long-context large language models have demonstrated superior retrieval quality compared to retrievalaugmented generation (RAG) approaches. |
| Approach: | They propose a memory-efficient training paradigm that partitions lengthy inputs into manageable chunks. |
| Outcome: | The proposed model expands maximum sequence length from 1K to 16K tokens on a single RTX 3090 GPU, while SpaCO achieves accelerated training speed. |
Revisiting LoRA through the Lens of Parameter Redundancy: Spectral Encoding Helps (2025.findings-acl)
Copied to clipboard
| Challenge: | Low-Rank Adaptation (LoRA) has emerged as a prominent technique for fine-tuning large foundation models. |
| Approach: | They propose a low-rank Adaptation technique that harnesses the expressiveness of spectral bases to re-parameterize LoRA from a sparse spectral subspace. |
| Outcome: | The proposed technique achieves greater efficiency with fewer parameters than baselines on various downstream tasks, including commonsense reasoning, math reasoning, and code generation. |
CODEMENV: Benchmarking Large Language Models on Code Migration (2025.findings-acl)
Copied to clipboard
Keyuan Cheng, Xudong Shen, Yihao Yang, TengyueWang TengyueWang, Yang Cao, Muhammad Asif Ali, Hanbin Wang, Lijie Hu, Di Wang
| Challenge: | Large language models (LLMs) have demonstrated remarkable proficiency in handling a wide range of tasks within the software engineering domain, but their ability to perform code migration—adapting code to different environments—remains underexplored. |
| Approach: | They propose a benchmark to evaluate large language models’ performance in handling code migration tasks. |
| Outcome: | The proposed benchmark comprises 922 data points across 19 Python and Java packages and offers three tasks to systematically evaluate code migration: identifying version-incompatible functions, determining function changes, and adapting code to target environments. |
A Case Study of Cross-Lingual Zero-Shot Generalization for Classical Languages in LLMs (2025.findings-acl)
Copied to clipboard
V.S.D.S.Mahesh Akavarapu, Hrishikesh Terdalkar, Pramit Bhattacharyya, Shubhangi Agarwal, Dr. Vishakha Deulgaonkar, Chaitali Dangarikar, Pralay Manna, Arnab Bhattacharya
| Challenge: | Large Language Models (LLMs) have demonstrated remarkable generalization capabilities across diverse tasks and languages. |
| Approach: | They focus on named entity recognition and machine translation into English to examine factors affecting cross-lingual zero-shot generalization. |
| Outcome: | The proposed models perform better than fine-tuned baselines on out-of-domain data, but smaller models struggle with niche or abstract entity types. |
BrainECHO: Semantic Brain Signal Decoding through Vector-Quantized Spectrogram Reconstruction for Whisper-Enhanced Text Generation (2025.findings-acl)
Copied to clipboard
| Challenge: | Current EEG/MEG-to-text decoding systems rely on teacher-forcing methods . pre-trained large language models are over-dominant in decoding text from brain activity . |
| Approach: | They propose a framework that employs decoupled representation learning to achieve state-of-the-art performance on EEG and MEG datasets. |
| Outcome: | The proposed framework achieves state-of-the-art performance on EEG and MEG datasets. |
Progressive LoRA for Multimodal Continual Instruction Tuning (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing approaches to MCIT address Catastrophic Forgetting and Knowledge Transfer (KT) but using a fixed number of shared LoRA blocks across tasks can lead to knowledge interference. |
| Approach: | They propose a framework that uses a fixed number of shared LoRA blocks to reduce knowledge interference. |
| Outcome: | The proposed framework outperforms existing approaches on the latest MCIT benchmark. |
ARC ‘Challenge’ Is Not That Challenging (2025.findings-acl)
Copied to clipboard
| Challenge: | ARC Challenge appears to be more difficult than ARC Easy for modern LLMs due to an evaluation setup that prevents direct comparison of answer choices rather than inherent complexity. |
| Approach: | They propose a setup where multiple choice problems are evaluated and the one with the highest likelihood is compared against the gold standard to determine accuracy. |
| Outcome: | The proposed evaluation setup is more difficult than ARC Easy for modern LLMs because it prevents direct comparison of answer choices rather than inherent complexity. |
Cross-Lingual Transfer of Debiasing and Detoxification in Multilingual LLMs: An Extensive Investigation (2025.findings-acl)
Copied to clipboard
| Challenge: | Prior work has shown that finetuning on specialized datasets can mitigate this behavior, and doing so in English can transfer to other languages. |
| Approach: | They propose to fine tune generative large language models to provide safe responses to harmful user input and to use direct preference optimization to mitigate toxicity. |
| Outcome: | The proposed models show that finetuning on specialized datasets reduces biases but also produces fluent and diverse text in non-English languages. |
Tracr-Injection: Distilling Algorithms into Pre-trained Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent efforts to characterize symbolic abilities of the transformer architecture have shown that the tasks that can be implemented in RASP are uncommon to learn from natural unsupervised data. |
| Approach: | They propose a programming language, called RASP, which can be directly compiled into transformer weights to implement these algorithms. |
| Outcome: | The proposed method improves out-of-distribution performance compared to baselines, indicating that indeed a more symbolic mechanism is taking place in the inner workings of the model. |
Model Performance-Guided Evaluation Data Selection for Effective Prompt Optimization (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing prompt engineering methods rely on randomly selected evaluation subsets, leading to suboptimal prompts. |
| Approach: | They propose an iterative evaluation data selection approach for effective prompt optimization using real time model performance. |
| Outcome: | The proposed approach improves effectiveness by 1.6% to 3.1% and stability by 50% to 55.5% on two datasets BIG-bench and LIAR and two models GPT-3.5 and GPT-4o-mini. |
Revisiting Weak-to-Strong Generalization in Theory and Practice: Reverse KL vs. Forward KL (2025.findings-acl)
Copied to clipboard
| Challenge: | Weak-to-strong generalization is a promising approach to guide stronger systems, but its effectiveness is constrained by the inherent imperfections of weak model supervision. |
| Approach: | They propose a theoretically grounded approach that replaces forward KL divergence with reverse KL, which prioritizes high-confidence predictions. |
| Outcome: | The proposed approach replaces forward KL divergence with reverse KL, reducing the influence of unreliable weak supervision. |
Stories that (are) Move(d by) Markets: A Causal Exploration of Market Shocks and Semantic Shifts across Different Partisan Groups (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing attempts to model the relationship between the real world and written or spoken text have focused on more interpretable and simplistic text representations. |
| Approach: | They propose to link shifts in semantic embedding space to real-world market shocks and partisanship to shape predictions of market fluctuations. |
| Outcome: | The proposed model demonstrates that partisanship can influence the predictive power of text for market fluctuations and shape reactions to those same shocks. |
NetSafe: Exploring the Topological Safety of Multi-agent System (2025.findings-acl)
Copied to clipboard
Miao Yu, Shilong Wang, Guibin Zhang, Junyuan Mao, Chenlong Yin, Qijiong Liu, Kun Wang, Qingsong Wen, Yang Wang
| Challenge: | Large language models (LLMs) have fueled significant progress in intelligent Multi-agent Systems (MAS), with expanding academic and industrial applications. |
| Approach: | They propose a framework that unifies diverse MAS workflows via iterative RelCom interactions to enable generalized analysis. |
| Outcome: | The proposed framework unifies diverse MAS workflows via iterative RelCom interactions to enable generalized analysis. |
Reasoning is All You Need for Video Generalization: A Counterfactual Benchmark with Sub-question Evaluation (2025.findings-acl)
Copied to clipboard
Qiji Zhou, YiFan Gong, Guangsheng Bao, Hongjie Qiu, Jinqiang Li, Xiangrong Zhu, Huajian Zhang, Yue Zhang
| Challenge: | Existing multimodal benchmarks often overlook counterfactual reasoning, which is crucial for robust video understanding. |
| Approach: | They propose a multidimensional multimodal benchmark that systematically evaluates MLLMs across the abstract-concrete and perception-cognition dimensions. |
| Outcome: | The proposed model decomposes complex queries into structured sub-questions, enabling fine-grained reasoning analysis. |
Initializing and Retrofitting Key-Value Adaptors for Traceable Model Editing (2025.findings-acl)
Copied to clipboard
| Challenge: | Language models (LMs) are becoming imperative tools for consulting in realworld scenarios. |
| Approach: | They propose a model editing method that initializes and retrofits key-value pairs into MLP blocks to construct a new mapping of a piece of knowledge without damaging irrelevant knowledge. |
| Outcome: | The proposed method outperforms baseline methods on a series of GPT series models on edit success and generalization without influencing specificity. |
Know the Unknown: An Uncertainty-Sensitive Method for LLM Instruction Tuning (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) have impressive capabilities but face significant challenges from hallucinations, which arise from insufficient knowledge or context. |
| Approach: | They propose a novel two-stage approach for contextual question answering that enhances LLMs’ ability to recognise their knowledge boundaries while the second reinforces instruction adherence through carefully designed causal prompts. |
| Outcome: | The proposed approach significantly reduces incorrect answers in contextual QA and improves models’ faithfulness to parametric knowledge, mitigating hallucinations in general QA tasks. |
Position-Aware Depth Decay Decoding (D3): Boosting Large Language Model Inference Efficiency (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent dynamic computation methods show that not all components are required for inference, enabling a training-free pipeline. |
| Approach: | They propose a token-position aware layer skipping framework to save 1.5x times operations efficiently while maintaining performance. |
| Outcome: | The proposed algorithm achieves 1.5x speedup on large language models with no retraining and with comparable performance on the GSM8K and BBH benchmarks. |
Explaining Puzzle Solutions in Natural Language: An Exploratory Study on 6x6 Sudoku (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) are effective in human-AI collaborative decision-making, but they lack the ability to provide trustworthy, gradual, and tailored explanations. |
| Approach: | They evaluate the performance of five Large Language Models in solving and explaining Sudoku puzzles. |
| Outcome: | The proposed model can solve and explain complex Sudoku puzzles in a controlled environment. |
Stress-testing Machine Generated Text Detection: Shifting Language Models Writing Style to Fool Detectors (2025.findings-acl)
Copied to clipboard
Andrea Pedrotti, Michele Papucci, Cristiano Ciaccio, Alessio Miaschi, Giovanni Puccetti, Felice Dell’Orletta, Andrea Esuli
| Challenge: | Recent advances in Generative AI and Large Language Models (LLMs) have enabled the creation of highly realistic synthetic content, raising concerns about the potential for malicious use, such as misinformation and manipulation. |
| Approach: | They evaluate the resilience of state-of-the-art MGT detectors to linguistically informed adversarial attacks by using Direct Preference Optimization to shift the MGT style toward human-written text. |
| Outcome: | The proposed pipeline fine-tunes language models to shift the MGT style toward human-written text (HWT) it obtains generations more challenging to detect by current models, and shows that detectors can be easily fooled with relatively few examples, resulting in a significant drop in detecting performances. |
InfiniSST: Simultaneous Translation of Unbounded Speech with Large Language Model (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing models for simultaneous speech translation assume pre-segmented speech, limiting their real-world applicability. |
| Approach: | They propose a multi-turn dialogue task that can translate unbounded streaming speech . they construct translation trajectories and robust segments from MuST-C with multi-latency augmentation during training and develop a cache management strategy to facilitate efficient inference. |
| Outcome: | The proposed approach reduces computation-aware latency by 0.5 to 1 second while maintaining the same translation quality compared to baselines. |
VSCBench: Bridging the Gap in Vision-Language Model Safety Calibration (2025.findings-acl)
Copied to clipboard
Jiahui Geng, Qing Li, Zongxiong Chen, Yuxia Wang, Derui Zhu, Zhuohan Xie, Chenyang Lyu, Xiuying Chen, Preslav Nakov, Fakhri Karray
| Challenge: | Existing safety calibration methods focus on model undersafety, where the model responds to hazardous queries, while neglecting oversafetiness, where models refuse to answer safe queries. |
| Approach: | They propose safety calibration which addresses both undersafety and oversafetiness by comparing model responses to a novel dataset of 3,600 image-text pairs. |
| Outcome: | The proposed methods have been used to evaluate safety calibration across image-centric and text-centric scenarios. |
To Code or not to Code? Adaptive Tool Integration for Math Language Models via Expectation-Maximization (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing tools that integrate chain-of-thought reasoning and code execution lack metacognitive awareness to integrate tools. |
| Approach: | They propose a framework that synergizes structured exploration with off-policy RL optimization to create a cycle between metacognitive tool-use decisions and evolving capabilities. |
| Outcome: | The proposed framework improves over 11% on MATH500 and 9.4% on AIME without o1-like CoT. |
GOODLIAR: A Reinforcement Learning-Based Deceptive Agent for Disrupting LLM Beliefs on Foundational Principles (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent advances indicate that LLMs exhibit increasingly complex reasoning abilities . |
| Approach: | They propose a reinforcement learning framework that generates deceptive contexts to rewrite an LLM’s core axiomatic beliefs. |
| Outcome: | The proposed framework induces persistent belief shifts rather than one-off policy breaches. |
How Does Response Length Affect Long-Form Factuality (2025.findings-acl)
Copied to clipboard
| Challenge: | Despite growing attention to LLM factuality, the effect of response length on factual accuracy remains underexplored. |
| Approach: | They propose an automatic and bi-level long-form factuality evaluation framework which achieves high agreement with human annotations while being cost-effective. |
| Outcome: | The proposed framework achieves high agreement with human annotations while being cost-effective. |
Scaling LLMs’ Social Reasoning: Sprinkle Cognitive “Aha Moment” into Fundamental Long-thought Logical Capabilities (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing studies have examined how large language models’ social reasoning capabilities evolve during model size scaling or reasoning tokens scaling. |
| Approach: | They propose to optimize evaluation of Large Language Models from both data and model perspectives and to analyze their reasoning trajectories to identify notable cognitive "Aha Moments" |
| Outcome: | The proposed model outperforms the o1-preview model by 19.0 points in the evaluation of large language models. |
SimGRAG: Leveraging Similar Subgraphs for Knowledge Graphs Driven Retrieval-Augmented Generation (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent advances in large language models (LLMs) have shown impressive versatility across various tasks. |
| Approach: | They propose a retrieval-augmented generation method that integrates LLMs with external knowledge sources to produce grounded outputs. |
| Outcome: | The proposed method outperforms state-of-the-art KG-driven methods in question answering and fact verification. |
RuleEdit: Towards Rule-Level Knowledge Generalization to Mitigate Over-Editing in Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing knowledge editing methods focus on instance-level editing, which is prone to knowledge degradation and general ability deterioration due to redundant instance-specific modifications. |
| Approach: | They propose a rule-level editing method that generalizes rule-derived knowledge to update rule-based instances. |
| Outcome: | The proposed method improves portability and performance over baselines for LLaMA-2-7B on RULEmix. |
Eliciting In-context Retrieval and Reasoning for Long-context Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing benchmarks like LOFT often overestimate LCLM performance by providing overly simplified contexts. |
| Approach: | They propose to use retrieval-attention-probing to filter and de-noise long contexts during decoding and joint retrieval head training alongside the generation head to improve LCLM performance. |
| Outcome: | The proposed approach outperforms RAG and GPT-4-Turbo on most tasks despite being a much smaller model. |
GeAR: Generation Augmented Retrieval (2025.findings-acl)
Copied to clipboard
Haoyu Liu, Shaohan Huang, Jianfeng Liu, Yuefeng Zhan, Hao Sun, Weiwei Deng, Feng Sun, Furu Wei, Qi Zhang
| Challenge: | Document retrieval techniques are used to compute semantic similarity between a query and documents, but the scalar similarity fails to reflect enough information, hindering the interpretation of retrieval results. |
| Approach: | They propose a method which improves the global document-query similarity through contrastive learning and integrates well-designed fusion and decoding modules. |
| Outcome: | The proposed method improves the global document-query similarity through contrastive learning and integrates well-designed fusion and decoding modules. |
A Unified Taxonomy-Guided Instruction Tuning Framework for Entity Set Expansion and Taxonomy Expansion (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing studies view entity set expansion, taxonomy expansion, and seed-guided taxonomies as three separate tasks. |
| Approach: | They propose a taxonomy-guided instruction tuning framework to teach a large language model to generate siblings and parents for query entities. |
| Outcome: | The proposed framework outperforms baselines on multiple benchmark datasets. |
Zero-Shot Conversational Stance Detection: Dataset and Approaches (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing stance detection datasets are limited to a limited set of specific targets . current models are limited in their ability to detect large numbers of unseen targets based on a large number of unidentified targets. |
| Approach: | They propose a speaker interaction and target-aware prototypical contrastive learning model that can detect public opinion towards specific targets using social media data. |
| Outcome: | The proposed model achieves state-of-the-art in zero-shot conversational stance detection with only an F1-macro score of 43.81%. |
LongFaith: Enhancing Long-Context Reasoning in LLMs with Faithful Synthetic Data (2025.findings-acl)
Copied to clipboard
| Challenge: | Long-context processing ability has emerged as a significant challenge for large language models. |
| Approach: | They propose a pipeline for synthesizing faithful long-context reasoning instruction datasets . they integrate ground truth and citation-based reasoning prompts integrating them . |
| Outcome: | The proposed pipeline eliminates distractions and improves reasoning chains. |
SYNTHVERIFY: Enhancing Zero-Shot Claim Verification through Step-by-Step Synthetic Data Generation (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for claim verification are inefficient or rely on external documents. |
| Approach: | They propose a step-by-step prompting-based synthetic data generation framework to enhance zero-shot claim verification. |
| Outcome: | The proposed framework bridges LLMs’ knowledge gaps in specialized domains without access to external corpora or sacrificing generalizability. |
Domaino1s: Guiding LLM Reasoning for Explainable Answers in High-Stakes Domains (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing Large Language Models (LLMs) generate brief answers without reasoning processes and explanations. |
| Approach: | They propose supervised fine-tuning and tree search to enhance LLMs’ reasoning capabilities on domain tasks. |
| Outcome: | The proposed model improves on stock investment recommendation and legal reasoning QA tasks. |
Dynamic Prefix as Instructor for Incremental Named Entity Recognition: A Unified Seq2Seq Generation Framework (2025.findings-acl)
Copied to clipboard
| Challenge: | Named Entity Recognition (NER) is a fundamental problem in information extraction. |
| Approach: | They propose a parameter-efficient method for Incremental Named Entity Recognition (INER) task aimed at updating a model to extract entities from an expanding set of entity type candidates by employing a dynamic prefix as a task instructor to guide the generative model. |
| Outcome: | Empirical results show that the proposed method preserves task-invariant knowledge while adapting to new entities with minimal parameter updates. |
Who Taught You That? Tracing Teachers in Model Distillation (2025.findings-acl)
Copied to clipboard
| Challenge: | Xu et al., 2006, show that model distillation can imbue efficient small language models with task-specific capabilities competitive with expensive teacher LLMs. |
| Approach: | They propose to distill outputs from a large teacher model to a small student model . they propose to use part-of-speech templates as higher-order linguistic features capable of capturing distinctive signals from teacher models that persist in distilled student outputs. |
| Outcome: | The proposed model distillation technique can imbue efficient small language models with task-specific capabilities competitive with (expensive) teacher LLMs. |
D-GEN: Automatic Distractor Generation and Evaluation for Reliable Assessment of Generative Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for generating generative models with open-ended generation rely on predefined distractors and are costly and time-consuming. |
| Approach: | They propose a ranking alignment and entropy analysis to evaluate distractors' quality. |
| Outcome: | The proposed model preserves ranking consistency and matches the entropy distribution of ground-truth distractors. |
HammerBench: Fine-Grained Function-Calling Evaluation in Real Mobile Assistant Scenarios (2025.findings-acl)
Copied to clipboard
Jun Wang, Jiamu Zhou, Xihuai Wang, Xiaoyun Mo, Haoyu Zhang, Qiqiang Lin, Jincheng Jincheng, Muning Wen, Weinan Zhang, Qiuying Peng, Jun Wang
| Challenge: | Evaluating the performance of LLMs in multi-turn interactions presents significant challenges due to the complexity and variability of user behavior. |
| Approach: | They propose a benchmark framework for assessing LLMs’ function-calling capabilities in multi-turn dialogues. |
| Outcome: | The proposed framework is based on a dataset derived from popular mobile apps and anonymized user logs. |
Beyond In-Context Learning: Aligning Long-form Generation of Large Language Models via Task-Inherent Attribute Guidelines (2025.findings-acl)
Copied to clipboard
Do Xuan Long, Duong Ngoc Yen, Do Xuan Trong, Anh Tuan Luu, Kenji Kawaguchi, Shafiq Joty, Min-Yen Kan, Nancy F. Chen
| Challenge: | In-context learning is an important but not fully understood ability of pre-trained large language models. |
| Approach: | They propose a tool that generates two streams of guidelines capturing task language and format distributions and prompts them to define them by prompting. |
| Outcome: | The proposed model improves both strong open- and closed-source LLMs by over 5% in both zero- and few-shot settings. |
GRAMMAR-LLM: Grammar-Constrained Natural Language Generation (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing approaches to fine-tuning and prompting are insufficient to ensure compliance with predefined taxonomies, syntactic structures, or domain-specific rules. |
| Approach: | They propose a framework that integrates formal grammatical constraints into the decoding process to enforce syntactic correctness in linear time while maintaining expressiveness in grammar rule definition. |
| Outcome: | The proposed framework enforces syntactic correctness in linear time while maintaining expressiveness in grammar rule definition. |
MANBench: Is Your Multimodal Model Smarter than Human? (2025.findings-acl)
Copied to clipboard
| Challenge: | Multimodal Large Language Models (MLLMs) have been gaining popularity in multimodal tasks . a bilingual benchmark is available for MLLM users to evaluate their multimodal capabilities . |
| Approach: | They propose a bilingual multimodal ability norms benchmark that measures multimodality across nine tasks. |
| Outcome: | The proposed benchmark compared human performance against state-of-the-art MLLMs. |
BanStereoSet: A Dataset to Measure Stereotypical Social Biases in LLMs for Bangla (2025.findings-acl)
Copied to clipboard
| Challenge: | ***BanStereoSet*** is a dataset designed to evaluate stereotypical social biases in multilingual LLMs for the Bangla language. |
| Approach: | They propose to localize the content from StereoSet, IndiBias, and kamruzzaman-etal's datasets to capture biases prevalent within the Bangla language. |
| Outcome: | The proposed dataset consists of 1,194 sentences spanning 9 categories of bias: race, profession, gender, ageism, beauty, beauty in profession, region, caste, and religion. |
mOSCAR: A Large-scale Multilingual and Multimodal Document-level Corpus (2025.findings-acl)
Copied to clipboard
Matthieu Futeral, Armel Randy Zebaze, Pedro Ortiz Suarez, Julien Abadji, Rémi Lacroix, Cordelia Schmid, Rachel Bawden, Benoît Sagot
| Challenge: | Existing studies show that multimodal large language models can learn from text-image data. |
| Approach: | They propose to train multimodal large language models on large amounts of text-image data . they also show a boost in few-shot learning performance across various multilingual tasks . |
| Outcome: | The proposed dataset is not public and is only in English . it is the first large-scale multilingual and multimodal document corpus crawled from the web. |
NorEval: A Norwegian Language Understanding and Generation Evaluation Benchmark (2025.findings-acl)
Copied to clipboard
Vladislav Mikhailov, Tita Enstad, David Samuel, Hans Christian Farsethås, Andrey Kutuzov, Erik Velldal, Lilja Øvrelid
| Challenge: | NorEval is a new evaluation suite for large-scale standardized benchmarking of Norwegian generative language models (LMs). |
| Approach: | They propose a new evaluation suite for large-scale standardized benchmarking of Norwegian generative language models (LMs) NorEval consists of 24 high-quality human-created datasets, of which five are created from scratch. |
| Outcome: | The evaluation framework and materials are publicly available. |
Massively Multilingual Instruction-Following Information Extraction (2025.findings-acl)
Copied to clipboard
| Challenge: | Past literature on information extraction (IE) has focused on a few high-resource languages, hindering their applications on multilingual corpora. |
| Approach: | They propose a collection of data that unifies and standardizes instruction-following multilingual IE and introduce a structure-aware metric that captures partially matched spans. |
| Outcome: | The proposed framework standardizes and unifies 215 manually annotated datasets, covering 96 typologically diverse languages from 18 language families. |
DALR: Dual-level Alignment Learning for Multimodal Sentence Representation Learning (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing multimodal sentence representation learning methods focus on aligning images and text at a coarse level, resulting in cross-modal misalignment bias and intra-modal semantic divergence. |
| Approach: | They propose a dual-level alignment learning framework for multimodal sentence representation learning that promotes cross-modal and intra-modal alignment. |
| Outcome: | The proposed framework outperforms state-of-the-art methods on semantic textual similarity and transfer tasks on semantic similarity, ranking distillation and global intra-modal alignment learning. |
Large Language Models in Bioinformatics: A Survey (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) are revolutionizing bioinformatics, enabling advanced analysis of DNA, RNA, proteins, and single-cell data. |
| Approach: | They examine the evolution of Large Language Models (LLMs) in bioinformatics and precision medicine by focusing on genomic sequence modeling, RNA structure prediction, protein function inference, and single-cell transcriptomics. |
| Outcome: | The proposed models are capable of predicting RNA structure and function and predicting single-cell transcriptomics. |
ChartEdit: How Far Are MLLMs From Automating Chart Analysis? Evaluating MLLMs’ Capability via Chart Editing (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing evaluations of multimodal large language models rely on limited case studies . however, they lack the ability to generate accurate edits according to the instructions . |
| Approach: | They propose a benchmark for chart editing that includes 1,405 edit instructions applied to 233 real-world charts. |
| Outcome: | The proposed benchmark includes 1,405 diverse editing instructions applied to 233 real-world charts. |
Unraveling and Mitigating Safety Alignment Degradation of Vision-Language Models (2025.findings-acl)
Copied to clipboard
Qin Liu, Chao Shang, Ling Liu, Nikolaos Pappas, Jie Ma, Neha Anna John, Srikanth Doss, Lluis Marquez, Miguel Ballesteros, Yassine Benajiba
| Challenge: | LLaVA-7B demonstrated a decline in safety alignment ability on multi-modal inputs compared to its LLM backbone. |
| Approach: | They propose a method to recover alignment ability from LLM backbone while preserving functional capabilities of VLMs. |
| Outcome: | The proposed framework recovers alignment ability that is inherent in the LLM backbone with minimal impact on fluency and linguistic capabilities of pre-trained VLMs. |
Turbocharging Web Automation: The Impact of Compressed History States (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing web automation approaches ignore the importance of history states to accomplish tasks. |
| Approach: | They propose a web history compressor approach to turbocharge web automation using history states by concatenating history states with other inputs. |
| Outcome: | The proposed approach achieves 1.2-5.4% accuracy improvements over baseline methods on Mind2Web and WebLINX datasets. |
Making RALM Robust to Irrelevant Contexts via Layer Knowledge Guided Attention (2025.findings-acl)
Copied to clipboard
Weijie Shi, Hao Chen, Jiaming Li, Yao Zhao, Yazhong Zhang, Qijin Chen, Jipeng Zhang, Ruiyuan Zhang, Jia Zhu, Jiajie Xu, Xiaofang Zhou
| Challenge: | Large language models (LLMs) face factual hallucination and knowledge obsolescence when tackling knowledge-intensive tasks. |
| Approach: | They propose a layer-knowledge guided attention method which harnesses the layer-wise knowledge of large language models to optimize per-layer attention on useful passages. |
| Outcome: | The proposed method outperforms existing methods on RALM benchmarks. |
Rewrite to Jailbreak: Discover Learnable and Transferable Implicit Harmfulness Instruction (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing jailbreak methods create a forced instruction-following scenario, or search adversarial prompts with prefix or suffix tokens to achieve a specific representation manually or automatically. |
| Approach: | They propose a method that rewrites the original instruction to achieve a jailbreak . they propose rewriting the original instructions to improve the attack strategy . |
| Outcome: | The proposed method is more efficient and easier to identify since no additional features are introduced. |
SignAlignLM: Integrating Multimodal Sign Language Processing into Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Deaf and Hard-of-Hearing (DHH) users increasingly utilize Large Language Models (LLMs), yet face significant challenges due to these models’ limited understanding of sign language grammar, multimodal sign inputs, and Deafic cultural contexts. |
| Approach: | They propose to use sign language support in LLMs to integrate sign linguistic rules and conventions into prompting and fine-tuning strategies to address the needs of DHH users. |
| Outcome: | The proposed model can be generalized interfaces for both spoken and signed languages if trained with a multitasking paradigm. |
NegVQA: Can Vision Language Models Understand Negation? (2025.findings-acl)
Copied to clipboard
| Challenge: | NegVQA is a visual question answering (VQA) benchmark consisting of 7,379 two-choice questions covering diverse negation scenarios and image-question distributions. |
| Approach: | They propose a visual question answering benchmark consisting of 7,379 two-choice questions covering diverse negation scenarios and image-question distributions. |
| Outcome: | The proposed model fails to correctly interpret negation, leading to critical errors in interactive AI systems. |
Natural Language Reasoning in Large Language Models: Analysis and Evaluation (2025.findings-acl)
Copied to clipboard
| Challenge: | Argumentative reasoning presents unique challenges due to its reliance on context, implicit assumptions, and value judgments. |
| Approach: | They propose a large-scale evaluation of LLMs' unconstrained natural language reasoning capabilities . they formalise a new strategy designed to evaluate argumentative reasoning in LLM . |
| Outcome: | The proposed model performs better on a range of reasoning tasks than other models. |
SWE-Dev: Building Software Engineering Agents with Training and Inference Scaling (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) have advanced rapidly from conversational problem solving to addressing real-world tasks involving tool use, such as software engineering (SWE). |
| Approach: | They propose to build an LLM-based software engineering agent that synthesizes test cases and scales up agent trajectories to build training data. |
| Outcome: | The proposed model outperforms state-of-the-art models on the SWE-bench-Verified benchmark. |
The Two Paradigms of LLM Detection: Authorship Attribution vs Authorship Verification (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for detecting texts generated by large language models are disputed . authors argue that there are limitations in the current technology . |
| Approach: | They propose to make LLM detectors robust against domain shifts and build benchmarks . they argue that the limitations lie elsewhere, and open the realm of authorship analysis technology . |
| Outcome: | The proposed method systematically analyzes the benchmarks and validates it using state-of-the-art detectors. |
Unveiling Confirmation Bias in Chain-of-Thought Reasoning (2025.findings-acl)
Copied to clipboard
| Challenge: | Chain-of-thought (CoT) prompting has been widely adopted to enhance the reasoning capabilities of large language models (LLMs). |
| Approach: | They propose to examine how internal beliefs affect reasoning generation and reasoning-guided answer prediction in CoT by decomposing CoT into a two-stage process. |
| Outcome: | The proposed model beliefs affect reasoning generation and reasoning-guided answer prediction in CoT, and the results provide strong evidence of confirmation bias in LLMs. |
GRNFormer: A Biologically-Guided Framework for Integrating Gene Regulatory Networks into RNA Foundation Models (2025.findings-acl)
Copied to clipboard
Mufan Qiu, Xinyu Hu, Fengwei Zhan, Sukwon Yun, Jie Peng, Ruichen Zhang, Bhavya Kailkhura, Jiekun Yang, Tianlong Chen
| Challenge: | Foundation models for single-cell RNA sequencing ignore biological prior knowledge encoded in gene regulatory relationships and fail to leverage multi-omics signals. |
| Approach: | They propose a framework that integrates multi-scale gene regulatory networks into RNA foundation model training. |
| Outcome: | The proposed framework improves on state-of-the-art models on three downstream tasks . it integrates multi-scale gene regulatory networks (GRNs) from multi-omics data into training . |
RemoteRAG: A Privacy-Preserving LLM Cloud RAG Service (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) have a tendency to generate factually incorrect or purely fictional responses, a phenomenon known as hallucination. |
| Approach: | They propose to use remote RAG to protect user query from privacy leakage . they introduce (n,)-DistanceDP to characterize privacy leakages of user query . |
| Outcome: | The proposed solution can resist embedding inversion attacks while achieving no loss in retrieval under various settings. |
“My life is miserable, have to sign 500 autographs everyday”: Exposing Humblebragging, the Brags in Disguise (2025.findings-acl)
Copied to clipboard
| Challenge: | Humblebragging is a phenomenon in which individuals present self-promotional statements under the guise of modesty or complaints. |
| Approach: | They propose a task of automatically detecting humblebragging in text and propose '4-tuple definition' they also propose machine learning, deep learning, and large language models to perform the task . |
| Outcome: | The proposed model achieves an F1-score of 0.88 and is non-trivial even for humans. |
SCITAT: A Question Answering Benchmark for Scientific Tables and Text Covering Diverse Reasoning Types (2025.findings-acl)
Copied to clipboard
Xuanliang Zhang, Dingzirui Wang, Baoxin Wang, Longxu Dou, Xinyuan Lu, Keyan Xu, Dayong Wu, Qingfu Zhu
| Challenge: | Existing scientific question answering datasets lack diverse reasoning types and neglect relevance between tables and text. |
| Approach: | They propose a scientific question answering benchmark for scientific tables and text with diverse reasoning types (SCITAT) to address these challenges, they propose QA benchmark which incorporates tables and texts to ensure that the questions encompass both tables and textes. |
| Outcome: | The proposed benchmark improves by 4.1% over baselines on SCITAT. |
TokenShapley: Token Level Context Attribution with Shapley Value (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) have strong capabilities in in-context learning, but verifying the correctness of their generated responses remains a challenge. |
| Approach: | They propose a token-level attribution method that combines Shapley value-based data attribution with KNN-based retrieval techniques to improve attribution accuracy. |
| Outcome: | TokenShapley outperforms state-of-the-art methods on four benchmarks . it achieves an 11–23% improvement in accuracy on the benchmarks. |
Entropy-based Exploration Conduction for Multi-step Reasoning (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods to automatically decide the depth of exploration of the reasoning procedure lead to high cost and a lack of flexibility. |
| Approach: | They propose a method that dynamically adjusts the exploration depth during multi-step reasoning by monitoring LLM’s output entropy and variance entropic. |
| Outcome: | The proposed method captures the uncertainty of the current step and the fluctuation of uncertainty across consecutive reasoning steps and then selects whether to deepen, expand, or stop exploration according to the probability. |
Taxonomizing Representational Harms using Speech Act Theory (2025.findings-acl)
Copied to clipboard
Emily Corvi, Hannah Washington, Stefanie Reed, Chad Atalla, Alexandra Chouldechova, P. Alex Dow, Jean Garcia-Gathright, Nicholas J Pangakis, Emily Sheng, Dan Vann, Matthew Vogel, Hanna Wallach
| Challenge: | a theoretical framework defines representational harms as perlocutionary effects of illocutional acts . the framework provides a granular taxonomy of ils that cause representational damages . |
| Approach: | They propose a framework that defines representational harms as perlocutionary effects of system behaviors . they propose illocutional acts that cause representational damage and a taxonomy that supports measurement instruments . |
| Outcome: | The proposed framework defines representational harms as perlocutionary effects of illocutionaries . it can support the development of valid measurement instruments, the authors show . |
Turning Conversations into Workflows: A Framework to Extract and Evaluate Dialog Workflows for Service AI Agents (2025.findings-acl)
Copied to clipboard
Prafulla Kumar Choubey, Xiangyu Peng, Shilpa Bhagavath, Caiming Xiong, Shiva Kumar Pentyala, Chien-Sheng Wu
| Challenge: | Existing workflow extraction methods for service agents are time-consuming and outdated, causing inconsistent and inconsistent results. |
| Approach: | They propose a framework for extracting and evaluating dialog workflows from historical interactions. |
| Outcome: | The proposed framework improves workflow extraction by 12.16% over baseline. |
Statistical inference on black-box generative models in the data kernel perspective space (2025.findings-acl)
Copied to clipboard
| Challenge: | Generative models are capable of producing human-expert level content across a variety of topics and domains. |
| Approach: | They extend recent results on representations of black-box generative models to model-level statistical inference tasks. |
| Outcome: | The proposed models are effective for multiple inference tasks and meet or surpass human-level standards on benchmarks across a range of tasks. |
Do Large Language Models Perform Latent Multi-Hop Reasoning without Exploiting Shortcuts? (2025.findings-acl)
Copied to clipboard
| Challenge: | Latent multi-hop reasoning is a problem in Large Language Models that can develop shortcuts by encountering the head entity and answer entity in training sequences. |
| Approach: | They propose desiderata for shortcut-free evaluation of latent multi-hop reasoning ability . they exclude test queries where head and answer entities might have co-appeared . |
| Outcome: | The proposed model can latently recall and compose single-hop facts without shortcuts, but only for certain types of queries. |
AceMath: Advancing Frontier Math Reasoning with Post-Training and Reward Modeling (2025.findings-acl)
Copied to clipboard
| Challenge: | In this paper, we introduce a suite of math models that excel in solving complex math problems. |
| Approach: | They propose a supervised fine-tuning process that achieves competitive performance across general domains, followed by targeted fine- tuning for the math domain using a carefully curated set of prompts and synthetically generated responses. |
| Outcome: | The proposed model outperforms Qwen2.5-Math-72B-Instruct, GPT-4o and Claude-3.5 Sonnet in the math domain. |
WXImpactBench: A Disruptive Weather Impact Understanding Benchmark for Evaluating Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Climate change adaptation requires the understanding of disruptive weather impacts on society. |
| Approach: | They propose a large language model to evaluate the capacity of LLMs on disruptive weather impacts by using a four-stage construction pipeline. |
| Outcome: | The proposed model is based on a four-stage well-crafted construction pipeline and requires two evaluation tasks, multi-label classification and ranking-based question answering. |
MeMoTune: A Measure and Moment-Driven Fine-Tuning Framework for Quantized Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods combine quantization with parameter-efficient fine-tuning but fail to meet practical performance requirements. |
| Approach: | They propose a measure and moment approach to optimize objective function for superior fine-tuning results by scaling the update process through a gradient. |
| Outcome: | The proposed framework outperforms state-of-the-art methods on tasks like text generation, summarization, and understanding. |
MALAMUTE: A Multilingual, Highly-granular, Template-free, Education-based Probing Dataset (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing cloze-style benchmarks for language models lack specific, granular areas of knowledge and often rely on templates that can bias models. |
| Approach: | They propose a multilingual, template-free, and highly granular probing dataset comprising expert-written, peer-reviewed probes from 71 university-level textbooks across three languages. |
| Outcome: | The proposed dataset covers eight domains, each with up to 14 subdomains, further broken down into concepts and concept-based prompts. |
Sentimental Image Generation for Aspect-based Sentiment Analysis (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent work on textual Aspect-Based Sentiment Analysis (ABSA) has demonstrated promising performance, but limited semantics derived from raw data. |
| Approach: | They propose a method that provides visual semantics to reinforce textual ABSA by adding additional augmentations to the input data. |
| Outcome: | The proposed method can provide visual semantics to reinforce the textual extraction. |
Long-form Hallucination Detection with Self-elicitation (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for hallucination detection tend to decompose text into isolated statements, unable to understand contextual semantics. |
| Approach: | They propose a framework to leverage self-generated thoughts derived from prior statements as catalysts to elicit the expression of intrinsic knowledge and understand contextual semantics. |
| Outcome: | The proposed framework enables self-elicitation to elicit expressions of knowledge and understand semantics. |
ComparisonQA: Evaluating Factuality Robustness of LLMs Through Knowledge Frequency Control and Uncertainty (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing studies on LLMs' factual knowledge are unreliable since the questions can vary not only in entity frequency but also in difficulty themselves. |
| Approach: | They propose a benchmark to study the role of knowledge frequency in the performance of large language models (LLMs) it aims to avoid possible semantic shortcuts which is a serious problem of current QA study. |
| Outcome: | The proposed method avoids possible semantic shortcuts and improves on existing proofs. |
One-Dimensional Object Detection for Streaming Text Segmentation of Meeting Dialogue (2025.findings-acl)
Copied to clipboard
Rui He, Zhongqing Wang, Minjie Qiang, Hongling Wang, Yifan.zhang Yifan.zhang, Hua Xu, Shuai Fan, Guodong Zhou
| Challenge: | Current text segmentation models exhibit numerous limitations, such as imbalances in labels that affect the stability of model training and discrepancies between the model’s training tasks (sentence classification) and the actual text segmenting. |
| Approach: | They implement a sliding window-based segmentation method and employ two different levels of sliding window based balanced label strategies to stabilize the training process of the streaming segmentation model. |
| Outcome: | The proposed method is robust, controllable, and achieves state-of-the-art performance. |
CodeTaxo: Enhancing Taxonomy Expansion with Limited Examples via Code Language Prompts (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing taxonomies are mainly constructed by experts or through crowd-sourcing, making the process time-consuming, labor-intensive, and restricted in coverage. |
| Approach: | They propose a method that leverages large language models to capture taxonomic structure . existing taxonomies are mainly constructed by experts or through crowd-sourcing . |
| Outcome: | Experiments on five real-world domains show that CodeTaxo outperforms state-of-the-art methods. |
Predicate-Conditional Conformalized Answer Sets for Knowledge Graph Embeddings (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods provide probabilistic guarantees over a reference set of queries and answers, but they fail to identify when the answers to a query are uncertain. |
| Approach: | They propose a method that approximates predicate-conditional coverage guarantees while maintaining compact prediction sets. |
| Outcome: | The proposed method provides predicate-conditional coverage guarantees while maintaining compact prediction sets. |
Autonomous Data Selection with Zero-shot Generative Classifiers for Mathematical Texts (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods that require human annotations or training a dedicated data filter to curate high-quality mathematical texts are based on autonomous data selection. |
| Approach: | They propose a method that leverages base language models as zero-shot "generative classifiers" they use a model's logits to determine whether a given passage is mathematically informative and educational . |
| Outcome: | The proposed method significantly boosts downstream performance on math benchmarks while using far fewer tokens than previous methods. |
Learning from Committee: Reasoning Distillation from a Mixture of Teachers with Peer-Review (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) have proven to be highly effective in addressing a wide range of complex tasks. |
| Approach: | They propose a method that asks teachers to identify and explain student’s mistakes and then asks them to provide customized instruction learning data. |
| Outcome: | The proposed method reduces the chance of teachers guessing incorrectly with flawed rationales, improving instructional data quality. |
Investigating Prosodic Signatures via Speech Pre-Trained Models for Audio Deepfake Source Attribution (2025.findings-acl)
Copied to clipboard
| Challenge: | x-vector (speaker recognition PTM) achieves the highest performance in prosodic tasks . despite its low parameter, x vector captures unique prosodic characteristics of the sources . |
| Approach: | They propose to use SOTA speech pre-trained models to capture prosodic sig-natures of generative sources for audio deepfake source attribution. |
| Outcome: | The proposed model captures prosodic sig-natures of generative sources better than other models on ASVSpoof and CFAD. |
Multilingual Retrieval Augmented Generation for Culturally-Sensitive Tasks: A Benchmark for Cross-lingual Robustness (2025.findings-acl)
Copied to clipboard
Bryan Li, Fiona Luo, Samar Haider, Adwait Agashe, Siyu Li, Runqi Liu, Miranda Muqing Miao, Shriya Ramakrishnan, Yuan Yuan, Chris Callison-Burch
| Challenge: | Existing approaches to retrieval-augmented generated (RAG) can be useful in multilingual settings, but they also introduce biases in the retrieved documents. |
| Approach: | They propose a dataset of territorial disputes paired with retrieved Wikipedia documents in 49 languages to evaluate cross-lingual robustness. |
| Outcome: | The proposed paradigm helps mitigate hallucinations of large language models (LLMs). |
Bridging Relevance and Reasoning: Rationale Distillation in Retrieval-Augmented Generation (2025.findings-acl)
Copied to clipboard
Pengyue Jia, Derong Xu, Xiaopeng Li, Zhaocheng Du, Xiangyang Li, Yichao Wang, Yuhao Wang, Qidong Liu, Maolin Wang, Huifeng Guo, Ruiming Tang, Xiangyu Zhao
| Challenge: | Existing approaches to rerank and align documents based on reasoning capabilities of large language models (LLMs) . prior work shows that LLMs have exceptional reasoning and text generation capabilities . |
| Approach: | They propose a rationale extraction method that leverages reasoning capabilities of large language models to extract the rationales necessary for answering a query. |
| Outcome: | The proposed method is compared with baseline methods on two tasks across three datasets. |
Scaling Laws for Multilingual Language Models (2025.findings-acl)
Copied to clipboard
Yifei He, Alon Benhaim, Barun Patra, Praneetha Vaddamanu, Sanchit Ahuja, Parul Chopra, Vishrav Chaudhary, Han Zhao, Xia Song
| Challenge: | Existing scaling laws for language models are limited to a limited number of languages, but they can be applied to arbitrary number of different languages. |
| Approach: | They propose a scaling law for general-purpose decoder-only language models trained on multilingual data that shifts focus from individual languages to language families. |
| Outcome: | The proposed scaling law can be applied to models trained on multilingual data . it can be used to predict performance across multiple languages and models . |
Corpus Poisoning via Approximate Greedy Gradient Descent (2025.findings-acl)
Copied to clipboard
| Challenge: | Dense retrievers have been shown to be vulnerable to corpus poisoning attacks . a malicious user injects a small fraction of adversarial passages into the retrieval corpus to trick the system into returning these passages among the top-ranked results for a broad set of user queries. |
| Approach: | They propose an attack on dense retrieval systems based on the widely used HotFlip method for efficiently generating adversarial passages. |
| Outcome: | The proposed method achieves a high success rate on several datasets and using several retrievers, and can generalize to unseen queries and new domains. |
Taxonomy-Driven Knowledge Graph Construction for Domain-Specific Scientific Applications (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for constructing domain-specific knowledge graphs neglect curated taxonomies and LLMs fail to extract KGs in specialized domains. |
| Approach: | They propose a taxonomy-driven framework for constructing domain-specific knowledge graphs . they use structured taxonomies, Large Language Models and Retrieval-Augmented Generation . |
| Outcome: | The proposed framework can be adapted for other specialized domains. |
Wanda++: Pruning Large Language Models via Regional Gradients (2025.findings-acl)
Copied to clipboard
Yifan Yang, Kai Zhen, Bhavana Ganesh, Aram Galstyan, Goeric Huybrechts, Markus Müller, Jonas M. Kübler, Rupak Vignesh Swaminathan, Athanasios Mouchtaris, Sravan Babu Bodapati, Nathan Susanj, Zheng Zhang, Jack FitzGerald, Abhishek Kumar
| Challenge: | Existing pruning methods suffer from accuracy degradation without full-model sparsity-aware fine-tuning. |
| Approach: | They propose a pruning framework that uses decoder-block-level regional gradients to improve pruning accuracy. |
| Outcome: | The proposed pruning framework outperforms the state-of-the-art pruning frameworks by utilizing decoder-block-level regional gradients. |
MATCHED: Multimodal Authorship-Attribution To Combat Human Trafficking in Escort-Advertisement Data (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for human trafficking detection ignore the multimodal nature of online ads . sex trafficking is a pervasive crime exploiting individuals of all ages and genders . |
| Approach: | They propose to use multimodal authorship attributes to identify suspicious ads that combine text and images to improve vendor identification and verification tasks. |
| Outcome: | The proposed model outperforms existing methods for vendor identification and verification tasks using text-only, vision-only and multimodal training objectives. |
Fraud-R1 : A Multi-Round Benchmark for Assessing the Robustness of LLM Against Augmented Fraud and Phishing Inducements (2025.findings-acl)
Copied to clipboard
Shu Yang, Shenzhe Zhu, Zeyu Wu, Keyu Wang, Junchi Yao, Junchao Wu, Lijie Hu, Mengdi Li, Derek F. Wong, Di Wang
| Challenge: | Existing fraud detection benchmarks focus on single-turn classification tasks, failing to capture dynamic nature of real-world fraud attempts. |
| Approach: | They propose a bilingual benchmark to assess LLMs' ability to resist fraud and phishing attacks across five key fraud categories: Fraudulent Services, Impersonation, Phishing Scams, Fake Job Postings, and Online Relationships. |
| Outcome: | The proposed model improves in role-play settings and in e-commerce and recommendation systems. |
Mitigating Paraphrase Attacks on Machine-Text Detection via Paraphrase Inversion (2025.findings-acl)
Copied to clipboard
| Challenge: | Paraphrases applied to machine-generated texts can degrade performance of machine-text detectors. |
| Approach: | They propose an approach which frames the problem as translation from paraphrased text back to the original text. |
| Outcome: | The proposed approach yields an average improvement of +22% AUROC across seven detectors and three different domains. |
SANSKRITI: A Comprehensive Benchmark for Evaluating Language Models’ Knowledge of Indian Culture (2025.findings-acl)
Copied to clipboard
| Challenge: | Language models excel in syntactic and semantic analysis, while small language models struggle in region-specific contexts. |
| Approach: | They evaluate SANSKRITI on leading Large Language Models, Indic Language Model, and Small Language Model (SLM) it covers 16 key attributes of Indian culture including rituals and ceremonies, history, tourism, cuisine, dance and music, costume, language, art, festivals, religion, medicine, transport, sports, nightlife and personalities. |
| Outcome: | The SANSKRITI dataset covers 16 attributes of Indian culture . it reveals that many models struggle in region-specific contexts . |
System Prompt Hijacking via Permutation Triggers in LLM Supply Chains (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) are rapidly transforming the landscape of artificial intelligence due to the substantial resources required for training. |
| Approach: | They propose a post-deployment attack that bypasses system prompts to compromise models . they introduce Precise Activation Guarding and Unit Deviation Sampling to protect against attack . |
| Outcome: | The proposed attack bypasses system prompts, enabling unrestricted model outputs and safety violations. |
Frequency matters: Modeling irregular morphological patterns in Spanish with Transformers (2025.findings-acl)
Copied to clipboard
| Challenge: | A common generation task in morphology is morphological inflection, where a target form has to be generated from its corresponding lemma and feature tag. |
| Approach: | They propose to solve the Paradigm Cell Filling Problem (PCFP) by using encoder-decoder transformers to generate inflected verbs in Spanish. |
| Outcome: | The proposed model performs better on L-shaped verbs than regular verbs, but no consistent recency effects are observed. |
From Heart to Words: Generating Empathetic Responses via Integrated Figurative Language and Semantic Context Signals (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing research on empathy generation focuses on understanding the emotions of the speaker rather than on how the responder conveys empathy. |
| Approach: | They propose to use figurative language and causal semantic context to facilitate targeted empathy generation in a mental health support domain. |
| Outcome: | The proposed approach achieves 7.6% improvement in BLEU, 36.7% reduction in Perplexity, and 7.6% increase in lexical diversity. |
There’s No Such Thing as Simple Reasoning for LLMs (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing work has focused on relatively complex “many-hop” reasoning problems. |
| Approach: | They analyse the performance of fine-tuned LLMs on simple reasoning problems . they find the models remain highly brittle, being susceptible to seemingly innocent perturbations . |
| Outcome: | The proposed models fail on simple reasoning problems, but are highly brittle . they are susceptible to seemingly innocent perturbations, such as adding duplicates to the set of premises and shuffling the order in which the premises are presented. |
CLIX: Cross-Lingual Explanations of Idiomatic Expressions (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing definition generation systems are difficult to use in second language learning due to the presence of unfamiliar words and grammar. |
| Approach: | They propose to use cross-lingual explanations of idiomatic expressions to support vocabulary expansion for language learners. |
| Outcome: | The proposed system is able to explain idiomatic expressions in non-standard language. |
Beyond Semantic Entropy: Boosting LLM Uncertainty Quantification with Pairwise Semantic Similarity (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) generate long one-sentence responses that are less effective because they overlook two crucial factors: intra-cluster similarity and inter-c cluster similarity. |
| Approach: | They propose a method that generalizes semantic entropy and uses token probabilities to quantify uncertainty in large language models. |
| Outcome: | The proposed method can be extended to white-box settings by incorporating token probabilities. |
R3Mem: Bridging Memory Retention and Retrieval via Reversible Compression (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing memory solutions that store information via parameters struggle with reliable retrieval. |
| Approach: | They propose a memory network that optimizes both information Retention and Retrieval through Reversible context compression. |
| Outcome: | The proposed memory network outperforms conventional memory modules in long-horizon interaction tasks like conversational agents and achieves state-of-the-art performance in language modeling and retrieval-augmented generation tasks. |
Vision Language Model Helps Private Information De-Identification in Vision Data (2025.findings-acl)
Copied to clipboard
| Challenge: | Visual Language Models (VLMs) have gained popularity due to their ability to solve imagerelated tasks. |
| Approach: | They propose a framework to enhance privacy awareness of visual language models . they use a specialized instruction-tuning dataset and a tailored training methodology . |
| Outcome: | The proposed framework outperforms existing approaches in handling private information. |
Unveiling Privacy Risks in Multi-modal Large Language Models: Task-specific Vulnerabilities and Mitigation Challenges (2025.findings-acl)
Copied to clipboard
| Challenge: | Privacy risks in text-only Large Language Models are well-documented, especially their tendency to memorize and leak sensitive information. |
| Approach: | They propose a dataset to assess privacy risks across multi-modal tasks and scenarios . they demonstrate how models leak sensitive data across various tasks . |
| Outcome: | The proposed model can leak sensitive data embedded in images or stored in memory, exposing privacy risks. |
DeFine: Decision-Making with Analogical Reasoning over Factor Profiles (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models are ideal for decision-making, but they can be difficult to process when they are verbose and include repetition, hedging, and vagueness. |
| Approach: | They propose a framework that constructs probabilistic factor profiles from complex scenarios and integrates them with analogical reasoning to guide LLMs in making decisions in new situations. |
| Outcome: | The proposed framework separates the tasks of quantifying uncertainty and incorporating it into LLM decision-making. |
SMART: Self-Aware Agent for Tool Overuse Mitigation (2025.findings-acl)
Copied to clipboard
Cheng Qian, Emre Can Acikgoz, Hongru Wang, Xiusi Chen, Avirup Sil, Dilek Hakkani-Tür, Gokhan Tur, Heng Ji
| Challenge: | Current Large Language Models (LLMs) lack self-awareness to balance reasoning and tool use, increasing computational overhead. |
| Approach: | They propose a paradigm that enhances an agent’s self-awareness to optimize task handling and reduce tool overuse. |
| Outcome: | The proposed model reduces tool use by 24% while improving performance by over 37%. |
Continued Pretraining and Interpretability-Based Evaluation for Low-Resource Languages: A Galician Case Study (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent advances in large language models have led to remarkable improvements in language understanding and text generation. |
| Approach: | They propose a framework to evaluate large language models for underrepresented languages . they examine CPT strategies for languages with limited representation in multilingual models . |
| Outcome: | The proposed evaluation framework is based on the case of Galician language . it assesses trade-offs between linguistic enrichment and task-solving capabilities . |
TC-Bench: Benchmarking Temporal Compositionality in Conditional Video Generation (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing video generation models struggle to interpret compositional changes and synthesize components across different time steps. |
| Approach: | They propose a temporal compositionality benchmark that uses text prompts and ground truth videos to evaluate compositional changes in video. |
| Outcome: | The proposed benchmark can be used for text-to-video and image-to video generation. |
DAM: Dynamic Attention Mask for Long-Context Large Language Model Inference Acceleration (2025.findings-acl)
Copied to clipboard
| Challenge: | Long-context understanding is crucial for many NLP applications, but transformers struggle with efficiency due to quadratic complexity of self-attention. |
| Approach: | They propose a dynamic sparse attention mechanism that assigns adaptive masks at the attention-map level, preserving heterogeneous attention patterns. |
| Outcome: | The proposed method achieves high alignment with full-attention models while reducing memory and compute overhead. |
Arbiters of Ambivalence: Challenges of using LLMs in No-Consensus tasks (2025.findings-acl)
Copied to clipboard
| Challenge: | LLMs are increasingly being used to replace humans in "aligning" LLM training . studies question this trend, but have found they can be more effective in ambivalent scenarios where humans disagree . |
| Approach: | They develop a “no-consensus” benchmark by curating examples that encompass a variety of a priori ambivalent scenarios. |
| Outcome: | The proposed benchmarks show that LLMs can provide nuanced assessments when generating open-ended answers, but tend to take a stance on no-consensus topics when employed as judges or debaters. |
Beyond Text: Characterizing Domain Expert Needs in Document Research (2025.findings-acl)
Copied to clipboard
| Challenge: | Document research is a key part of almost all knowledge work, but are text-based NLP systems able to model these tasks as experts conceptualize and perform them? |
| Approach: | They interview 16 domain experts to understand their processes of document research . they find that processes are idiosyncratic, iterative, and rely heavily on social context . |
| Outcome: | The findings show that document research processes are idiosyncratic, iterative, and rely heavily on the social context of a document in addition to its content. |
Efficient but Vulnerable: Benchmarking and Defending LLM Batch Prompting Attack (2025.findings-acl)
Copied to clipboard
| Challenge: | a recent study reveals a significant security vulnerability in batch prompting . malicious users can inject attack instructions into a batch, leading to unwanted interference . |
| Approach: | They construct a batch prompting benchmark and test it against other LLMs to find out if batch prompts are vulnerable. |
| Outcome: | The proposed approach achieves 95% accuracy in detecting attacks. |
MM-R3: On (In-)Consistency of Vision-Language Models (VLMs) (2025.findings-acl)
Copied to clipboard
| Challenge: | a flurry of research has been conducted on the performance of state-of-the-art (SoTA) Vision Language Models (VLMs) on a variety of tasks. |
| Approach: | They propose a benchmarking tool to analyze performance of SoTA Vision Language Models (VLMs) on three tasks: Question Rephrasing, Image Restyling, and Context Reasoning. |
| Outcome: | The proposed model achieves absolute improvements of 5.7% and 12.5% on widely used VLMs such as BLIP-2 and LLaVa 1.5M in terms of consistency over their existing counterparts. |
Investigating Context Faithfulness in Large Language Models: The Roles of Memory Strength and Evidence Style (2025.findings-acl)
Copied to clipboard
| Challenge: | Retrieval-augmented generation improves Large Language Models (LLMs) by integrating external information into the response generation process. |
| Approach: | They investigate the impact of memory strength and evidence presentation on LLMs’ receptiveness to external evidence by measuring the divergence in LLM responses to different paraphrases of the same question. |
| Outcome: | The proposed method improves Large Language Models (LLMs) by integrating external information into the response generation process. |
Shadow-Activated Backdoor Attacks on Multimodal Large Language Models (2025.findings-acl)
Copied to clipboard
Ziyi Yin, Muchao Ye, Yuanpu Cao, Jiaqi Wang, Aofei Chang, Han Liu, Jinghui Chen, Ting Wang, Fenglong Ma
| Challenge: | Existing backdoor attacks on Multimodal Large Language Models are less applicable to open-ended conversations with users. |
| Approach: | They propose a shadow-activated backdoor attack scenario where attackers inject malicious content into the responses of MLLMs when the responses explicitly relate to the shadowed object. |
| Outcome: | The proposed framework achieves the desired behaviors by constructing a poisoned dataset and implementing an attention-regularized tuning strategy. |
Why Vision Language Models Struggle with Visual Arithmetic? Towards Enhanced Chart and Geometry Understanding (2025.findings-acl)
Copied to clipboard
| Challenge: | Vision Language Models struggle with visual arithmetic, seemingly simple tasks like object counting or length comparison, which are essential for relevant complex tasks like chart understanding and geometric reasoning. |
| Approach: | They propose a novel post-training strategy inspired by Piaget’s theory of cognitive development that trains VLMs to recognize invariant properties under visual transformations. |
| Outcome: | The proposed approach outperforms supervised fine-tuning methods while requiring 60% less training data. |
K-order Ranking Preference Optimization for Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing list-wise methods focus on optimizing list ranking consistency for LLMs to improve ranking abilities. |
| Approach: | They propose to extend the Plackett-Luce model to accommodate top-K ranking by extending the DPO’s Plact-Lucer model to dynamically determine appropriate K for different samples. |
| Outcome: | The proposed model can be extended to accommodate top-K ranking and improve training efficiency. |
Spectral Insights into Data-Oblivious Critical Layers in Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent studies have identified critical layers linked to specific functions or behaviors, limiting their use to post-hoc settings. |
| Approach: | They propose a data-oblivious approach to identify intrinsic critical layers in pre-fine-tuned LLMs by analyzing representation dynamics via Centered Kernel Alignment. |
| Outcome: | The proposed approach identifies critical layers in pre-fine-tuned models . layers with significant shifts in representation space are also those most affected during fine-tuning . |
SynFix: Dependency-Aware Program Repair via RelationGraph Analysis (2025.findings-acl)
Copied to clipboard
Xunzhu Tang, Jiechao Gao, Jin Xu, Tiezhu Sun, Yewei Song, Saad Ezzini, Wendkûuni C. Ouédraogo, Jacques Klein, Tegawendé F. Bissyandé
| Challenge: | Existing methods for resolving repository-level debugging are limited by their interdependencies. |
| Approach: | They propose a RelationGraph-based approach that integrates large language models with structural search and synchronization techniques for coordinated program repair across codebases. |
| Outcome: | SynFix resolves 52.33% of issues in SWE-bench-lite, 55.8% in Swe-bech-verified and 29.86% in S WE-beach-full. |
EXIT: Context-Aware Extractive Compression for Enhancing Retrieval-Augmented Generation (2025.findings-acl)
Copied to clipboard
| Challenge: | Current retrieval-augmented generation systems struggle when retrieval models fail to rank the most relevant documents . existing extractive methods reduce latency but rely on independent, non-adaptive sentence selection . |
| Approach: | They introduce an extractive context compression framework that enhances retrieval-augmented generation in question answering. |
| Outcome: | EXIT surpasses existing compression methods and uncompressed baselines in QA accuracy . the framework reduces inference time and token count while preserving contextual dependencies . |
Re-TASK: Revisiting LLM Tasks from Capability, Skill, and Knowledge Perspectives (2025.findings-acl)
Copied to clipboard
Zhihu Wang, Shiwan Zhao, Yu Wang, Heyuan Huang, Sitao Xie, Yubo Zhang, Jiaxin Shi, Zhixing Wang, Hongyan Li, Junchi Yan
| Challenge: | Existing approaches to solving complex tasks with large language models (LLMs) fail to decompose tasks accurately or execute subtasks effectively. |
| Approach: | They propose a Chain-of-Learning (CoL) paradigm that highlights task dependencies on specific capability items, further broken down into their constituent knowledge and skill components. |
| Outcome: | The proposed model improves Yi-1.5-9B and Llama3-Chinese-8B for legal tasks by 45.00% and 24.50% on different domains. |
Unlearning Backdoor Attacks for LLMs with Weak-to-Strong Knowledge Distillation (2025.findings-acl)
Copied to clipboard
| Challenge: | Parameter-efficient fine-tuning (PEFT) can bridge the gap between large language models and downstream tasks, but is vulnerable to malicious attacks. |
| Approach: | They propose a weak-to-strong unlearning algorithm based on feature alignment knowledge distillation to defend against backdoor attacks . they first train a small-scale language model through full-parameter fine-tuning to serve as the clean teacher model and then guide the large-scale poisoned student model in unlearning the backdoor. |
| Outcome: | The proposed method can unlearn backdoor features without compromising model performance. |
Packing Analysis: Packing Is More Appropriate for Large Models or Datasets in Supervised Fine-tuning (2025.findings-acl)
Copied to clipboard
| Challenge: | Packing is an optimization technique that optimizes training time and resources by combining different training sequences to fit the model’s maximum input length. |
| Approach: | They perform extensive comparisons between packing and padding methods, covering datasets ranging from 69K to 1.2M and models from 8B to 70B. |
| Outcome: | The proposed method has been shown to improve training efficiency while maintaining performance. |
Better Red Teaming via Searching with Large Language Model (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for evaluating large language models face challenges in managing semantic intricacies and optimizing the efficiency of the search process. |
| Approach: | They propose a framework that reconceptualizes test case generation as a strategic planning problem, leveraging Monte Carlo Tree Search. |
| Outcome: | Experiments on a range of LLM architectures show that the proposed framework achieves state-of-the-art attack success rates without sacrificing computational efficiency. |
AdaV: Adaptive Text-visual Redirection for Vision-Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Vision-language models often generate excessive visual tokens, leading to poor performance . a novel training-free visual token pruning method is proposed to improve performance despite the computational cost associated with VLMs. |
| Approach: | They propose a training-free visual token pruning method that reduces biased token pruning . they plan to open-source the code upon publication . |
| Outcome: | The proposed method reduces biased token pruning and enhances model robustness with limited visual token budget. |
MegaAgent: A Large-Scale Autonomous LLM-based Multi-Agent System Without Predefined SOPs (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing multi-agent systems lack agent coordination and rely on predefined procedures . existing systems lack adaptive task coordination when task is big and complex . |
| Approach: | They propose a large-scale autonomous LLM-based multi-agent system that generates agents based on task complexity and enables dynamic task decomposition, parallel execution, efficient communication and comprehensive system monitoring. |
| Outcome: | The proposed system outperforms existing systems in task completion efficiency and scalability. |
Persona-judge: Personalized Alignment of Large Language Models via Token-level Self-judgment (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for aligning language models with human preferences rely on reward signals and additional annotated data, limiting their scalability and adaptability to diverse human values. |
| Approach: | They propose a discriminative paradigm that leverages the intrinsic preference judgment capabilities of the model to align language models with human preferences. |
| Outcome: | The proposed model is scalable and efficient, paving the way for more adaptive personalized alignment. |
A Self-Distillation Recipe for Neural Machine Translation (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for Neural Machine Translation (NMT) have been proven effective in improving the performance of computer vision tasks without pre-training a teacher. |
| Approach: | They propose a rank-order augmented Pearson correlation loss and an iterative distillation method to prevent the discrepancy of predictions between the student and a stronger teacher from disturbing the training. |
| Outcome: | The proposed method can lead to significant improvements over the strong Transformer baseline on low/middle/high-resource tasks, obtaining comparable or better performance with fewer layers. |
BlockPruner: Fine-grained Pruning for Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) have significant computational and memory costs associated with training and inference. |
| Approach: | They propose a training-free structured pruning approach that targets redundancies in MHA and MLP blocks. |
| Outcome: | The proposed pruning approach achieves more granular and effective pruning compared to state-of-the-art pruning methods. |
Evaluating Implicit Bias in Large Language Models by Attacking From a Psychometric Perspective (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing studies have shown that large language models (LLMs) can elicit implicit biases that hurt certain demographics without explicit harmful words. |
| Approach: | They propose three attack approaches to elicit agreements to biased viewpoints from LLMs from a psychometric perspective and built two benchmarks to compare them. |
| Outcome: | The proposed methods elicit agreements to biased viewpoints more effectively than baselines. |
LongCite: Enabling LLMs to Generate Fine-grained Citations in Long-Context QA (2025.findings-acl)
Copied to clipboard
Jiajie Zhang, Yushi Bai, Xin Lv, Wanjun Gu, Danqing Liu, Minhao Zou, Shulin Cao, Lei Hou, Yuxiao Dong, Ling Feng, Juanzi Li
| Challenge: | Current long-context large language models lack citations to support their responses, making verification difficult due to potential hallucinations. |
| Approach: | They propose to use off-the-shelf LLMs to automatically construct long-context QA instances with precise sentence-level citations and leverage this pipeline to construct a large-scale SFT dataset for LQAC. |
| Outcome: | The proposed pipeline can generate responses with fine-grained citations on the fly, surpassing existing models including GPT-4o. |
An Empirical Study of Group Conformity in Multi-Agent Systems (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent advances in Large Language Models (LLMs) have enabled multi-agent systems that simulate real-world interactions with near-human reasoning. |
| Approach: | They analyze how LLM agents shape public opinion through debates on five contentious topics by simulating over 2,500 debates. |
| Outcome: | The proposed models show that LLM agents adopt specific stances over time and align with numerically dominant groups or more intelligent agents, exerting a greater influence. |
Combining the Best of Both Worlds: A Method for Hybrid NMT and LLM Translation (2025.findings-acl)
Copied to clipboard
Zhanglin Wu, Daimeng Wei, Xiaoyu Chen, Hengchao Shang, Jiaxin Guo, Zongyao Li, Yuanchang Luo, Jinlong Yang, Zhiqiang Rao, Hao Yang
| Challenge: | Large language models have advantages over neural machine translation systems, but they suffer from high computational costs and significant latency. |
| Approach: | They propose a scheduling policy that optimizes translation result while ensuring fast speed and as little LLM usage as possible. |
| Outcome: | The proposed model achieves optimal translation performance with less LLM usage on multilingual test sets. |
ASPO: Adaptive Sentence-Level Preference Optimization for Fine-Grained Multimodal Reasoning (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent advances have extended DPO to multimodal scenarios, achieving strong performance. |
| Approach: | They propose to use a sentence-level preference optimization technique to optimize individual sentences for more precise preference optimization without additional models or parameters. |
| Outcome: | Experiments show that Adaptive Sentence-level Preference Optimization significantly improves the alignment of multimodal models. |
NovelCR: A Large-Scale Bilingual Dataset Tailored for Long-Span Coreference Resolution (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing coreference resolution datasets are either small in scale or restrict coreference to a limited text span. |
| Approach: | They present a large-scale bilingual benchmark for long-span coreference resolution . they find that NovelCR is notably rich in long-spanning coreference pairs . |
| Outcome: | The proposed benchmark is rich in long-span coreference pairs and notably low baselines. |
Dynamic Attention-Guided Context Decoding for Mitigating Context Faithfulness Hallucinations in Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods, such as a n-terminal coding, do not provide accurate data for large language models. |
| Approach: | They propose a lightweight framework that leverages attention distributions and uncertainty signals in a single-pass decoding. |
| Outcome: | Experiments on open-book QA datasets show that DAGCD improves faithfulness and robustness while preserving computational efficiency. |
Exploring the Choice Behavior of Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) are increasingly being adopted across various domains where they help to make choices. |
| Approach: | They construct a virtual QA platform that includes three different experimental conditions, with four models from GPT and Llama series participating in repeated experiments. |
| Outcome: | The proposed model includes three experimental conditions and four models from GPT and Llama series. |
On-Policy Self-Alignment with Fine-grained Knowledge Feedback for Hallucination Mitigation (2025.findings-acl)
Copied to clipboard
Xueru Wen, Jie Lou, Xinyu Lu, Yuqiu Ji, Xinyan Guan, Yaojie Lu, Hongyu Lin, Ben He, Xianpei Han, Debing Zhang, Le Sun
| Challenge: | Large language models exhibit behavior that deviates from the boundaries of their knowledge during response generation. |
| Approach: | They propose a framework that allows large language models to explore their knowledge boundaries and self-correct generation behavior through fine-grained feedback signals. |
| Outcome: | The proposed framework enables LLMs to explore their knowledge boundaries and self-correct generation behavior through fine-grained feedback signals. |
From Phrases to Subgraphs: Fine-Grained Semantic Parsing for Knowledge Graph Question Answering (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing approaches to knowledge graph question answering (KGQA) face semantic misalignment and reasoning noise. |
| Approach: | They propose a fine-grained semantic parsing framework for KGQA that maps natural language queries to executable logical forms. |
| Outcome: | The proposed framework achieves 18.5% performance improvement over the SOTA on a multi-hop CWQ dataset. |
StableToolBench-MirrorAPI: Modeling Tool Environments as Mirrors of 7,000+ Real-World APIs (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing tool environments face challenges in balancing stability, scale, and realism, especially for benchmarking purposes. |
| Approach: | They propose a framework that trains specialized LLMs to accurately simulate real API responses by supervised fine-tuning and chain-of-thought reasoning. |
| Outcome: | The proposed framework achieves superior accuracy and stability compared to state-of-the-art methods on the newly constructed MirrorAPI-Bench and its integration into StableToolBench. |
ClaimPKG: Enhancing Claim Verification via Pseudo-Subgraph Generation with Lightweight Specialized LLM (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing verification methods rely on unstructured text corpora to break down claims . despite strong reasoning abilities, modern LLMs struggle with modular pipelines . |
| Approach: | They propose a framework that integrates knowledge graphs with LLM reasoning . they propose KGs provide structured, semantically rich representations . |
| Outcome: | The proposed framework outperforms baselines on the FactKG dataset by 9%-12% accuracy points across multiple categories. |
TriEmbed: Bridge the Gap between Text and Token Indices with Embedding Reparameterization (2025.findings-acl)
Copied to clipboard
| Challenge: | a current paradigm of language modeling discards linguistic relations between tokens during tokenization, creating a fundamental gap . empirical results show that TriEmbed provides more linguistically informative token embeddings . |
| Approach: | They propose a reparameterization method that incorporates morphological relationships . they propose to organize the vocabulary into a Trie structure to reparametrize embeddings . |
| Outcome: | Empirical results show that TriEmbed outperforms existing token embeddings while offering more linguistically informative token embeds. |
Chain of Methodologies: Scaling Test Time Computation without Training (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing prompts for complex reasoning tasks are limited to specific tasks with few-shot examples due to constraints like context length and information extraction accuracy. |
| Approach: | They propose a method to build structured reasoning processes by injecting human insights into LLMs' training data. |
| Outcome: | The proposed framework outperforms baselines in the analysis of large language models. |
A Survey on Personalized Alignment—The Missing Piece for Large Language Models in Real-World Applications (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have demonstrated remarkable capabilities, yet their transition to real-world applications reveals a critical limitation: the inability to adapt to individual preferences while maintaining alignment with universal human values. |
| Approach: | They propose a framework that enables LLMs to adapt their behavior within ethical boundaries based on individual preferences. |
| Outcome: | The proposed framework analyzes implementation approaches and evaluates their effectiveness across various scenarios. |
SuLoRA: Subspace Low-Rank Adaptation for Parameter-Efficient Fine-Tuning (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for fine-tuning large language models (LLMs) introduce parameter interference, leading to a gap in generalization performance for specific tasks compared to full fine-uning. |
| Approach: | They propose a parameter-separated low-rank adapter to account for task differences by decomposing LoRA’s parameter matrix into multiple independent subspaces and assigning them differentially to distinct tasks. |
| Outcome: | The proposed method outperforms LoRA in trainable parameter efficiency and overall model performance on various NLP tasks. |
MIRe: Enhancing Multimodal Queries Representation via Fusion-Free Modality Interaction for Multimodal Retrieval (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods focus on textual queries that include visual information, but lack the ability to address multimodal queries that encompass both textual and visual information. |
| Approach: | They propose a retrieval framework that achieves modality interaction without fusing textual features during the alignment. |
| Outcome: | The proposed method achieves modality interaction without fusing textual features during the alignment. |
Correcting on Graph: Faithful Semantic Parsing over Knowledge Graphs with Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Complex multi-hop questions require comprehensive retrieval and reasoning. |
| Approach: | They propose a semantic parsing framework to establish faithful logical queries that connect LLMs and knowledge graphs. |
| Outcome: | The proposed framework outperforms state-of-the-art KGQA methods on knowledge-intensive questions. |
COPR: Continual Human Preference Learning via Optimal Policy Regularization (2025.findings-acl)
Copied to clipboard
Han Zhang, Lin Gui, Yu Lei, Yuanzhao Zhai, Yehong Zhang, Zhuo Zhang, Yulan He, Hui Wang, Yue Yu, Kam-Fai Wong, Bin Liang, Ruifeng Xu
| Challenge: | Reinforcement Learning from Human Feedback (RLHF) is effective for aligning Large Language Models with human preferences, but its complex process limits its ability to continually learn human feedback. |
| Approach: | They propose a non-RL offline method to convert historical optimal policies into optimization constraints when continually learning new preferences. |
| Outcome: | The proposed method outperforms strong CL baselines in terms of reward-based evaluations and human assessment. |
Robust Preference Optimization via Dynamic Target Margins (2025.findings-acl)
Copied to clipboard
| Challenge: | Direct Preference Optimization (DPO) is an efficient method for ensuring safety and reliability in practical applications. |
| Approach: | They propose a dynamic target margin preference optimization algorithm that adjusts reward margins at the pairwise level. |
| Outcome: | The proposed method achieves an average 4.4% improvement over baselines, setting new benchmarks for state-of-the-art performance. |
AdaReTaKe: Adaptive Redundancy Reduction to Perceive Longer for Video-language Understanding (2025.findings-acl)
Copied to clipboard
| Challenge: | Multimodal Large Language Models (MLLMs) are limited by context length when processing long videos. |
| Approach: | They propose a training-free method that flexibly reduces redundancy by allocating compression ratios among time and model layers with theoretical guarantees. |
| Outcome: | Experiments on videoMME, MLVU, LongVideoBench, and LVBench show that AdaRETAKE outperforms existing methods by 2.3% and 2.8% for 7B and 72B models. |
Rethinking Stateful Tool Use in Multi-Turn Dialogues: Benchmarks and Challenges (2025.findings-acl)
Copied to clipboard
Hongru Wang, Wenyu Huang, Yufei Wang, Yuanhao Xi, Jianqiao Lu, Huan Zhang, Nan Hu, Zeming Liu, Jeff Z. Pan, Kam-Fai Wong
| Challenge: | Existing benchmarks that assess Language Models (LMs) as Language Agents (LAs) for tool use focus on stateless, single-turn interactions or partial evaluations, overlooking the inherent stateful nature of interactions in multi-turn applications. |
| Approach: | They propose a multi-turn dialogue dataset with stateful tool interactions considering the whole life cycle of tool use across six key tasks in three stages . they also build VirtualMobile – an embodied virtual mobile evaluation environment to simulate API calls and assess the robustness of the created APIs. |
| Outcome: | The proposed dataset evaluates 13 open- and closed-source LLMs and provides detailed analysis at each stage. |
Open-Set Living Need Prediction with Large Language Models (2025.findings-acl)
Copied to clipboard
Xiaochong Lan, Jie Feng, Yizhou Sun, Chen Gao, Jiahuan Lei, Xinleishi Xinleishi, Hengliang Luo, Yong Li
| Challenge: | Existing approaches to living need prediction treat it as a closed-set classification problem, severely limiting their ability to capture diversity and complexity of living needs. |
| Approach: | They propose a system leveraging large language models for unrestricted need prediction that leverages Maslow's hierarchy of needs to align predictions with human living needs. |
| Outcome: | The proposed system outperforms closed-set approaches on need-based life service recall by an average of 19.37% on real-world datasets. |
Improve Rule Retrieval and Reasoning with Self-Induction and Relevance ReEstimate (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing rule retrieval methods suffer from low accuracy due to semantic gap between instantiated facts and abstract representations of rules. |
| Approach: | They propose a method that induces inferential rules that might offer benefits for reasoning by abstracting the underlying knowledge and logical structure in queries. |
| Outcome: | The proposed method improves retrieval effectiveness and accuracy across settings. |
Beyond Words: Integrating Theory of Mind into Conversational Agents for Human-Like Belief, Desire, and Intention Alignment (2025.findings-acl)
Copied to clipboard
| Challenge: | Empirical evaluations of LLaMA-3 models demonstrate that ToM-informed alignment improves response quality, achieving win rates of 63% and 67%, respectively. |
| Approach: | They investigate whether open-source LLaMA models can represent and retain ToM-related constructs and whether they can be used to generate more aligned responses. |
| Outcome: | The proposed models can represent and retain ToM-related constructs and improve response quality. |
Multimodal Causal Reasoning Benchmark: Challenging Multimodal Large Language Models to Discern Causal Links Across Modalities (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing MLLMs lack robustness in multimodal causal reasoning compared to their performance in textual settings. |
| Approach: | They propose a novel multimodal chain-of-thought (CoT) reasoning benchmark that leverages siamese images and text pairs to challenge MLLMs. |
| Outcome: | The proposed benchmark leverages siamese images and text pairs to challenge MLLMs. |
Context-Aware Hierarchical Merging for Long Document Summarization (2025.findings-acl)
Copied to clipboard
| Challenge: | Hierarchical merging is a technique used to summarize very long texts . it can amplify LLM hallucinations, increasing the risk of factual inaccuracies . |
| Approach: | They propose to enrich hierarchical merging with context from the source document to reduce the risk of factual inaccuracies. |
| Outcome: | The proposed methods outperform zero-shot and hierarchical merging baselines on legal and narrative datasets. |
VCD: A Dataset for Visual Commonsense Discovery in Images (2025.findings-acl)
Copied to clipboard
| Challenge: | Visual commonsense data sets lack visual grounded representations of commonsensense . existing knowledge bases lack visual-based knowledge tied to actual visual scenes . |
| Approach: | They present a large-scale visual commonsense dataset with over 100,000 images and 14 million object-commonsense pairs that integrates both Seen (directly observable) and Unseen (inferrable) commonsens. |
| Outcome: | The proposed model integrates Seen (directly observable) and Unseen (inferrable) commonsense across Property, Action, and Space aspects. |
Self-Reasoning Language Models: Unfold Hidden Reasoning Chains with Few Reasoning Catalyst (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent studies have demonstrated that inference-time scaling increases performance of Large Language Models (LLMs) in various reasoning tasks such as mathematics and complex question answering by increasing the length of Chain-of-Thought (CoT). |
| Approach: | They propose a model which synthesizes longer CoT data and iteratively improves performance through self-training by incorporating a few demonstration examples. |
| Outcome: | The proposed model achieves an average improvement of more than +2.5 points across five reasoning tasks: MMLU, GSM8K, ARC-C, HellaSwag, and BBH on two backbone models. |
HyperCRS: Hypergraph-Aware Multi-Grained Preference Learning to Burst Filter Bubbles in Conversational Recommendation System (2025.findings-acl)
Copied to clipboard
Yongsen Zheng, Mingjie Qian, Guohua Wang, Yang Liu, Ziliang Chen, Mingzhi Mao, Liang Lin, Kwok-Yan Lam
| Challenge: | Existing methods to analyze filter bubbles in the static recommendation environment are unable to burst them during user interactions. |
| Approach: | They propose a paradigm to learn multi-grained user preferences during dynamic user-system interactions via natural language conversations to burst filter bubbles. |
| Outcome: | The proposed paradigm achieves state-of-the-art performance and the superior of bursting filter bubbles in the conversational recommendation system. |
Is LLM an Overconfident Judge? Unveiling the Capabilities of LLMs in Detecting Offensive Language with Annotation Disagreement (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing studies focus on evaluating large language models' ability to handle disagreement cases. |
| Approach: | They evaluate the performance of large language models in detecting offensive language at varying levels of agreement. |
| Outcome: | The proposed model improves detection accuracy and model alignment with human judgment by using disagreement samples in training. |
Language Repository for Long Video Understanding (2025.findings-acl)
Copied to clipboard
| Challenge: | Language-based learning models (LLMs) support long context-lengths but their effectiveness in handling long-term information gradually declines with input length. |
| Approach: | They propose a Language Repository (LangRepo) that maintains concise and structured information as an interpretable representation. |
| Outcome: | The proposed framework is evaluated on zero-shot visual question-answering benchmarks. |
Investigating Language Preference of Multilingual RAG Systems (2025.findings-acl)
Copied to clipboard
| Challenge: | Empirical results show that DKM-RAG mitigates language preference in generation and enhances performance across diverse linguistic settings. |
| Approach: | They propose a framework that integrates translated passages with internal knowledge to overcome these issues. |
| Outcome: | The proposed framework mitigates language preference in generation and enhances performance across diverse linguistic settings. |
FGDGNN: Fine-Grained Dynamic Graph Neural Network for Rumor Detection on Social Media (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for detecting rumors on social media focus on coarse-grained temporal information and ignore fine-grain temporal dynamics. |
| Approach: | They propose a fine-grained dynamic graph neural network model which incorporates fine-grain temporal information into a unified framework for rumor detection. |
| Outcome: | The proposed model improves on three public real-world datasets. |
Self-Tuning: Instructing LLMs to Effectively Acquire New Knowledge through Self-Teaching (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing approaches to keeping large language models current involve continued pre-training on new documents. |
| Approach: | They propose a learning framework that augments documents with knowledge-intensive tasks created in a self-supervised manner, focusing on memorization, comprehension, and self-reflection. |
| Outcome: | The proposed learning framework improves an LLM’s ability to acquire new knowledge from unseen raw documents through self-teaching. |
QueryAttack: Jailbreaking Aligned Large Language Models Using Structured Non-natural Query Language (2025.findings-acl)
Copied to clipboard
Qingsong Zou, Jingyu Xiao, Qing Li, Zhi Yan, Yuhang Wang, Li Xu, Wenxuan Wang, Kuofeng Gao, Ruoyu Li, Yong Jiang
| Challenge: | Existing methods to bypass security defenses of large language models (LLMs) are not effective, but QueryAttack can be jailbroken. |
| Approach: | They propose a framework to examine generalizability of safety alignment by translating malicious queries into structured non-natural query languages. |
| Outcome: | The proposed framework can achieve high attack success rates and jailbreak various defense methods on mainstream LLMs. |
Memory or Reasoning? Explore How LLMs Compute Mixed Arithmetic Expressions (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) can solve complex multi-step math reasoning problems, but their internal implementation is limited. |
| Approach: | They propose to use a "C**ausal **E**ffect **D**riven **F**ine-tuning method" to improve LLMs' reasoning ability. |
| Outcome: | The proposed method improves the model's reasoning ability by enhancing key components that are used to execute mixed arithmetic calculations. |
PersonaX: A Recommendation Agent-Oriented User Modeling Framework for Long Behavior Sequence (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for user profile modeling extract only partial segments from full historical behavior sequence, resulting in incomplete modeling and suboptimal profiling. |
| Approach: | They propose an agent-agnostic LLM-UM framework to augment downstream recommendation agents . it segments complete historical behaviors into clustered groups and performs offline multi-persona profiling . |
| Outcome: | The proposed framework improves agent performance and inference efficiency by 31% and 10% using 30–50% of behavioral data. |
Judge as A Judge: Improving the Evaluation of Retrieval-Augmented Generation through the Judge-Consistency of Large Language Models (2025.findings-acl)
Copied to clipboard
Shuliang Liu, Xinze Li, Zhenghao Liu, Yukun Yan, Cheng Yang, Zheni Zeng, Zhiyuan Liu, Maosong Sun, Ge Yu
| Challenge: | Existing evaluation metrics cannot fairly evaluate the outputs of RAG models during training and evaluation. |
| Approach: | They propose a method which prompts LLMs to generate different judgments based on various combinations of judgment dimensions and utilizes the judge-consistency to evaluate these judgments. |
| Outcome: | The proposed method generates more accurate evaluations for RAG models across different RAG model and datasets. |
Rationales Are Not Silver Bullets: Measuring the Impact of Rationales on Model Performance and Reliability (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing studies have shown that training language models with rationales augmentation is beneficial, but this view does not hold consistently. |
| Approach: | They conduct comprehensive investigations to thoroughly inspect the impact of rationales on model performance and a novel perspective of model reliability. |
| Outcome: | The proposed method outperforms untrained models in several areas and provides informative regulations on the broad utilization of rationales. |
CA-GAR: Context-Aware Alignment of LLM Generation for Document Retrieval (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent techniques such as Generation-Augmented Retrieval (GAR) and Generative Document Retrieleval (GDR) leverage LLMs to enhance retrieval performance but face key challenges: GAR’s generated content may not always align with the target document corpus, while GDR limits the generative capacity of LLM. |
| Approach: | They propose a Context-Aware Generation-Augmented Retrieval approach which integrates corpus information into their generation process. |
| Outcome: | Experimental results show that CA-GAR outperforms existing methods on seven tasks and four non-English languages. |
AgentCourt: Simulating Court with Adversarial Evolvable Lawyer Agents (2025.findings-acl)
Copied to clipboard
Guhong Chen, Liyang Fan, Zihan Gong, Nan Xie, Zixuan Li, Ziqiang Liu, Chengming Li, Qiang Qu, Hamid Alinejad-Rokny, Shiwen Ni, Min Yang
| Challenge: | Existing legal language models struggle with dynamic courtroom interactions, resulting in overfitting to standardized legal tasks. |
| Approach: | They propose a new adversarial evolutionary approach for agents that performs dynamic knowledge learning and evolution through structured adversarials in a simulated courtroom program. |
| Outcome: | The proposed approach outperforms existing LLM-based models in three critical dimensions: cognitive agility, professional knowledge, and logical rigor. |
MLDebugging: Towards Benchmarking Code Debugging Across Multi-Library Scenarios (2025.findings-acl)
Copied to clipboard
JinYang Huang, Xiachong Feng, Qiguang Chen, Hanjie Zhao, Zihui Cheng, Jiesong Bai, Jingxuan Zhou, Min Li, Libo Qin
| Challenge: | MLDebugging is a benchmark designed to assess debugging challenges within multi-library Python code. |
| Approach: | They propose to introduce a benchmark to assess debugging challenges within multi-library Python code using 126 Python libraries. |
| Outcome: | The proposed benchmark covers 126 Python libraries and a wide range of multi-library code issues. |
An Empirical Study of LLM-as-a-Judge for LLM Evaluation: Fine-tuned Judge Model is not a General Substitute for GPT-4 (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent studies have fine-tuned judge models based on open-source LLMs to evaluate the quality of other LLM. |
| Approach: | They propose to use open-source LLMs to evaluate Large Language Models (LLMs) their empirical results show that the models underperform GPT-4 in several dimensions . |
| Outcome: | The proposed models outperform GPT-4 on several dimensions including generalizability, fairness and adaptability. |
Expectation Confirmation Preference Optimization for Multi-Turn Conversational Recommendation Agent (2025.findings-acl)
Copied to clipboard
Xueyang Feng, Jingsen Zhang, Jiakai Tang, Wei Li, Guohao Cai, Xu Chen, Quanyu Dai, Yue Zhu, Zhenhua Dong
| Challenge: | Recent advances in Large Language Models (LLMs) have propelled the development of Conversational Recommendation Agents (CRAs). |
| Approach: | They propose a multi-turn preference optimization paradigm that leverages Expectation Confirmation Theory to explicitly model the evolution of user satisfaction throughout multi-turned dialogues. |
| Outcome: | The proposed paradigm eliminates the significant sampling overhead of existing MTPO methods while ensuring the optimization process drives meaningful improvements. |
ProMedTS: A Self-Supervised, Prompt-Guided Multimodal Approach for Integrating Medical Text and Time Series (2025.findings-acl)
Copied to clipboard
Shuai Niu, Jing Ma, Hongzhan Lin, Liang Bai, Zhihua Wang, V. W., Richard Yi Da Xu, Guo Li, Xian Yang
| Challenge: | Large language models excel at processing unstructured data, but integrating time series data with text remains a challenge. |
| Approach: | They propose a self-supervised multimodal framework that uses prompt-guided learning to unify heterogeneous data types. |
| Outcome: | The proposed framework outperforms state-of-the-art approaches on disease diagnosis tasks using real-world datasets. |
CipherBank: Exploring the Boundary of LLM Reasoning Capabilities through Cryptography Challenge (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) have demonstrated remarkable capabilities, but their capabilities in cryptographic decryption tasks remain underexplored. |
| Approach: | They propose a benchmark to evaluate the reasoning capabilities of large language models in cryptographic decryption tasks. |
| Outcome: | The proposed benchmark examines the reasoning capabilities of large language models in cryptographic decryption tasks. |
Which Retain Set Matters for LLM Unlearning? A Case Study on Entity Unlearning (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) are prone to retaining unauthorized or sensitive information from their training data, which raises privacy concerns. |
| Approach: | They propose to use a group of queries that share similar syntactic structures with the data targeted for removal to investigate the effects of unlearning on various subsets of the retain set. |
| Outcome: | The proposed method reduces the retention set, the portion of training data that is not targeted for removal, and improves model performance across subsets. |
Tell Me What You Don’t Know: Enhancing Refusal Capabilities of Role-Playing Agents via Representation Space Analysis and Editing (2025.findings-acl)
Copied to clipboard
Wenhao Liu, Siyu An, Junru Lu, Muling Wu, Tianlong Li, Xiaohua Wang, Changze Lv, Xiaoqing Zheng, Di Yin, Xing Sun, Xuanjing Huang
| Challenge: | Role-playing Agents (RPAs) struggle to recognize and respond to hard queries that conflict with their role-play knowledge. |
| Approach: | They propose a lightweight representation editing approach that conveniently shifts conflicting requests to the rejection region, thereby enhancing the model’s refusal accuracy. |
| Outcome: | The proposed model improves RPAs’ refusal ability of conflicting requests while maintaining their general role-playing capabilities. |
LR²Bench: Evaluating Long-chain Reflective Reasoning Capabilities of Large Language Models via Constraint Satisfaction Problems (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent advances in o1-like models have significantly enhanced the reasoning abilities of Large Language Models (LLMs). |
| Approach: | They propose a benchmark to evaluate the Long-chain Reflective Reasoning capabilities of Large Language Models. |
| Outcome: | The proposed benchmark evaluates the Long-chain Reflective Reasoning capabilities of Large Language Models (LLMs) it consists of 850 samples across six Constraint Satisfaction Problems (CSPs) |
McBE: A Multi-task Chinese Bias Evaluation Benchmark for Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing datasets on bias evaluation for large language models focus on English and North American culture and are limited to one task. |
| Approach: | They propose to evaluate Chinese language models' biases from multiple perspectives using a multi-task Chinese Bias Evaluation Benchmark. |
| Outcome: | The proposed model covers 12, 82 subcategories and 5 evaluation tasks covering a wide range of categories and content diversity. |
MARK: Multi-agent Collaboration with Ranking Guidance for Text-attributed Graph Clustering (2025.findings-acl)
Copied to clipboard
Yiwei Fu, Yuxing Zhang, Chunchun Chen, JianwenMa JianwenMa, Quan Yuan, Rong-Cheng Tu, Xinli Huang, Wei Ye, Xiao Luo, Minghua Deng
| Challenge: | Existing approaches to cluster graphs with GNNs are limited due to label scarcity. |
| Approach: | They propose to leverage large language models to enhance text-attributed graph clustering by using three LLMs as ranking-based supervision signals. |
| Outcome: | The proposed approach generates reliable guidance using collaboration of three LLM-based agents as ranking-based supervision signals. |
Can Language Models Capture Human Writing Preferences for Domain-Specific Text Summarization? (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent studies employ large language models as auxiliary tools for humancentered NLP. |
| Approach: | They construct a model to capture human writing preferences by fine-tuning pre-trained models with data and designing prompts to optimize the output of large language models. |
| Outcome: | The proposed model captures human writing preferences through the dimensions of length, content depth, tone & style, and summary format. |
Mitigate Position Bias in LLMs via Scaling a Single Hidden States Channel (2025.findings-acl)
Copied to clipboard
Yijiong Yu, Huiqiang Jiang, Xufang Luo, Qianhui Wu, Chin-Yew Lin, Dongsheng Li, Yuqing Yang, Yongfeng Huang, Lili Qiu
| Challenge: | Long-context language models exhibit position bias, also known as "lost in the middle" research shows that even long-contemporary LLMs fail to utilize all context information effectively . |
| Approach: | They propose a method to mitigate position bias by scaling positional hidden states . they propose to use a channel of hidden states to modify positional Hidden states a LCLM's positional bias . |
| Outcome: | The proposed method can improve performance by 15.2% in a "lost in the middle" benchmark. |
Self-attention-based Graph-of-Thought for Math Problem Solving (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for generating reasoning paths in a chain structure are inefficient and non-human-like. |
| Approach: | They propose a decoding method for a chain-based LLM that constructs a thought graph simultaneously as an LLM inference and generates reasoning steps with a graph-structured self-attention mechanism. |
| Outcome: | The proposed method improves reasoning accuracy without huge computational over-expensive LLMs and avoids performance degradation issues when the LLM is too small to comprehend complex prompts. |
BAR: A Backward Reasoning based Agent for Complex Minecraft Tasks (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing studies focus on forward reasoning based planning, but this paradigm doesn't work well for complex tasks. |
| Approach: | They propose to decompose a task into easily executed steps by planning and use a backward reasoning based agent to make the planning starting from the terminal state. |
| Outcome: | The proposed model outperforms existing methods and the proposed modules in a virtual environment that simulates complex tasks based on real-world scenarios. |
KAPA: A Deliberative Agent Framework with Tree-Structured Knowledge Base for Multi-Domain User Intent Understanding (2025.findings-acl)
Copied to clipboard
Jiakai Tang, Shiqi Shen, ZhipengWang ZhipengWang, Gong Zhi, Xueyang Feng, Zexu Sun, Haoran Tan, Xu Chen
| Challenge: | Existing studies on the use of LLMs for estimating user intents are either too far from real human thought processes or require labeled samples. |
| Approach: | They propose a deliberative agent framework that leverages human thought process to build high-level domain knowledge and a tree-structured knowledge base to store refined experience and data. |
| Outcome: | The proposed framework is able to build high-level domain knowledge and efficiently store it across multiple steps. |
RASD: Retrieval-Augmented Speculative Decoding (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for generating draft tokens rely on lightweight draft models or additional model structures to generate tokens and retrieve context from databases. |
| Approach: | They propose to use a pruning method to enhance model-based speculative decoding by combining the best-fit model with the best retrieval tree. |
| Outcome: | The proposed method achieves state-of-the-art inference acceleration across tasks such as DocQA, Summary, Code, and In-Domain QA. |
FRAG: A Flexible Modular Framework for Retrieval-Augmented Generation based on Knowledge Graphs (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for retrieval-augmented generation struggle with a trade-off between flexibility and retrieval quality. |
| Approach: | They propose a flexible modular KG-RAG framework that uses query text instead of KGs . they propose to use query text to infer the structural information of reasoning paths . |
| Outcome: | The proposed method achieves state-of-the-art performance with high efficiency and low resource consumption. |
Reefknot: A Comprehensive Benchmark for Relation Hallucination Evaluation, Analysis and Mitigation in Multimodal Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing research focuses on object-level or attribute-level hallucinations, neglecting the more complex relation hallucinosities. |
| Approach: | They propose a comprehensive benchmark targeting relation hallucinations comprising over 20,000 real-world samples and a confidence-based mitigation strategy which reduces the halluciation rate by an average of 9.75% across three datasets. |
| Outcome: | The proposed approach reduces the hallucination rate by an average of 9.75% across three datasets, including Reefknot. |
Blessing of Multilinguality: A Systematic Analysis of Multilingual In-Context Learning (2025.findings-acl)
Copied to clipboard
| Challenge: | In-context learning (ICL) is a widely adopted technique for learning large language models . however, there is little systematic understanding of when and why it works well . |
| Approach: | They analyze multilingual in-context learning using demonstrations in HRLs to enhance cross-lingual transfer. |
| Outcome: | The proposed method outperforms English-only models on high-resource languages . the study shows that the presence of irrelevant non-English sentences in the prompt yields measurable gains . |
SEK: Self-Explained Keywords Empower Large Language Models for Code Generation (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) have achieved impressive performance in code generation. |
| Approach: | They propose a technique that extracts and explicates the key terms in the problem description with the LLM itself. |
| Outcome: | The proposed technique improves the Pass@1 of DeepSeek-Coder-V2-Instruct from 85.4% to 93.3% on the humaneval benchmark. |
Why Not Act on What You Know? Unleashing Safety Potential of LLMs via Self-Aware Guard Enhancement (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have demonstrated impressive capabilities across various tasks but are vulnerable to meticulously crafted jailbreak attacks. |
| Approach: | They propose a training-free defense strategy to align LLMs’ strong safety discrimination performance with their relatively weaker safety generation ability. |
| Outcome: | The proposed strategy achieves an average 99% success rate against numerous complex and covert jailbreak methods while maintaining helpfulness on general benchmarks. |
Explorer: Scaling Exploration-driven Web Trajectory Synthesis for Multimodal Web Agents (2025.findings-acl)
Copied to clipboard
Vardaan Pahuja, Yadong Lu, Corby Rosset, Boyu Gou, Arindam Mitra, Spencer Whitehead, Yu Su, Ahmed Hassan Awadallah
| Challenge: | Recent success in large multimodal models (LMMs) has sparked promising applications of agents capable of autonomously completing complex web tasks. |
| Approach: | They propose a scalable recipe to synthesize the largest and most diverse trajectory-level dataset to date. |
| Outcome: | The proposed model synthesizes the largest and most diverse trajectory-level dataset to date, with 94K successful multimodal web trajectories, 720K screenshots, and 33M web elements. |
Advancing General Multimodal Capability of Vision-language Models with Pyramid-descent Visual Position Encoding (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods to encode visual positions inhibit the performance of vision-language Models (VLMs) however, language constitutes only one aspect of communication. |
| Approach: | They propose a method to assign visual position indexes from the periphery to the center and expand the central receptive field incrementally to enhance the perception of visual tokens within VLMs. |
| Outcome: | The proposed method reduces the relative distance between interrelated visual elements and instruction tokens, promoting a more rational allocation of attention weights and allowing for a multi-granularity perception of visual elements. |
P-React: Synthesizing Topic-Adaptive Reactions of Personality Traits via Mixture of Specialized LoRA Experts (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing studies on personalized large language models focus on modeling explicit character profiles, while ignoring the underlying personality traits that truly shape behaviors and decision-making. |
| Approach: | They propose a personalized large language model (LLM) that captures implicit Big Five personality traits and integrates a Personality Specialization Loss to capture individual trait expressions. |
| Outcome: | The proposed model improves on Big Five personality traits and integrates a Personality Specialization Loss (PSL) to capture individual trait expressions. |
EssayJudge: A Multi-Granular Benchmark for Assessing Automated Essay Scoring Capabilities of Multimodal Large Language Models (2025.findings-acl)
Copied to clipboard
Jiamin Su, Yibo Yan, Fangteng Fu, Zhang Han, Jingheng Ye, Xiang Liu, Jiahao Huo, Huiyu Zhou, Xuming Hu
| Challenge: | Automated Essay Scoring (AES) systems face three major challenges: reliance on handcrafted features that limit generalizability, difficulty in capturing fine-grained traits like coherence and argumentation, and inability to handle multimodal contexts. |
| Approach: | They propose a multimodal benchmark to evaluate AES capabilities across lexical-, sentence-, and discourse-level traits without manual feature engineering. |
| Outcome: | The proposed system can evaluate AES capabilities across lexical-, sentence-, and discourse-level traits without manual feature engineering. |
Streamlining the Collaborative Chain of Models into A Single Forward Pass in Generation-Based Tasks (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent advances in "Chain of Models" approach increase resource demands as each model must be deployed separately. |
| Approach: | They propose a prompt-tuning method that enables models to share hidden states . they modify input and attention masks during training to eliminate redundant forward passes . |
| Outcome: | Empirical results show that FTHSS matches the performance of traditional model chains while improving inference efficiency. |
Self-Correction is More than Refinement: A Learning Framework for Visual and Language Reasoning Tasks (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have shown remarkable abilities, but they invariably generate flawed responses. |
| Approach: | They propose a self-correction approach that instructs VLMs to refine their outputs by allowing them to learn from their self-generated self-reference data without external feedback. |
| Outcome: | The proposed approach enables VLMs to learn from their self-generated self-correction data without relying on external feedback, facilitating self-improvement. |
Beyond Reactive Safety: Risk-Aware LLM Alignment via Long-Horizon Simulation (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing alignment methods focus on reactive feedback, where immediate human perception is leveraged to judge sampled model responses as preference data for post-training. |
| Approach: | They propose a proof-of-concept framework that projects how model-generated advice could propagate through societal systems on a macroscopic scale over time, enabling more robust alignment. |
| Outcome: | The proposed framework achieves 20% improvement on existing safety benchmarks and an average win rate exceeding 70% against strong baselines. |
Probability-Consistent Preference Optimization for Enhanced LLM Reasoning (2025.findings-acl)
Copied to clipboard
Yunqiao Yang, Houxing Ren, Zimu Lu, Ke Wang, Weikang Shi, Aojun Zhou, Junting Pan, Mingjie Zhan, Hongsheng Li
| Challenge: | Recent advances in preference optimization have demonstrated significant potential for improving mathematical reasoning capabilities in large language models. |
| Approach: | They propose a framework that establishes two quantitative metrics for preference selection: surface-level answer correctness and intrinsic token-level probability consistency. |
| Outcome: | The proposed framework outperforms existing outcome-only criterion approaches across a diverse range of LLMs and benchmarks. |
IW-Bench: Evaluating Large Multimodal Models for Converting Image-to-Web (2025.findings-acl)
Copied to clipboard
Hongcheng Guo, Wei Zhang, Junhao Chen, Yaonan Gu, Jian Yang, Junjia Du, Shaosheng Cao, Binyuan Hui, Tianyu Liu, Jianxin Ma, Chang Zhou, Zhoujun Li
| Challenge: | Existing models have been introduced to improve image comprehension, but there is no robust benchmark for imagetoweb conversion. |
| Approach: | They propose a benchmark to assess imagetoweb conversion proficiency of large multimodal models . they propose to measure layout information of web pages by parsing the Document Object Model tree . |
| Outcome: | The proposed benchmark measures the layout information of web pages—i.e., the positional relationships between elements—which has been overlooked by prior work. |
TDCSA: LLM-Guided Top-Down Approach for Robust Citation Sentiment Analysis (2025.findings-acl)
Copied to clipboard
| Challenge: | Citation Sentiment Analysis (CSA) is a key part of academic influence and knowledge diffusion. |
| Approach: | They propose a top-down framework that leverages LLMs’ semantic understanding capabilities to enhance PLM-based Citation Sentiment Analysis. |
| Outcome: | The proposed framework outperforms existing methods while maintaining robustness to quadruple quality variations. |
DeepRTL2: A Versatile Model for RTL-Related Tasks (2025.findings-acl)
Copied to clipboard
| Challenge: | Integration of large language models into electronic design automation has been a key driver in eDA. |
| Approach: | They propose a family of large language models that unifies generation- and embedding-based tasks related to RTL. |
| Outcome: | The proposed model achieves state-of-the-art performance across all evaluated tasks. |
The Self-Improvement Paradox: Can Language Models Bootstrap Reasoning Capabilities without External Scaffolding? (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing approaches to self-improvement rely on external supervision signals in the form of seed data and/or assistance from third-party models. |
| Approach: | They propose a framework for generating high-quality synthetic question-answer data in a fully autonomous manner. |
| Outcome: | The proposed framework generates high-quality synthetic question-answer data in a fully autonomous manner. |
Cross-lingual Multimodal Sentiment Analysis for Low-Resource Languages via Language Family Disentanglement and Rethinking Transfer (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing multimodal sentiment analysis methods are limited to textual data and cannot handle multimodal scenarios. |
| Approach: | They propose a transfer learning framework that allows cross-lingual and cross-modal alignments and a language family disentanglement module that enhances the sharing of language universals within families. |
| Outcome: | The proposed method is superior to existing methods and can handle low-resource languages. |
Does Chain-of-Thought Reasoning Really Reduce Harmfulness from Jailbreaking? (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing jailbreak attacks fail against reasoning models enhanced by Chain-of-Thought (CoT) reasoning. |
| Approach: | They propose a jailbreak method that uses Chain-of-Thought reasoning to reduce harmfulness from jailbreaking. |
| Outcome: | The proposed jailbreak method performs well against open AI models and deepseek-R1 reasoning models. |
InternLM-XComposer2.5-Reward: A Simple Yet Effective Multi-Modal Reward Model (2025.findings-acl)
Copied to clipboard
Yuhang Zang, Xiaoyi Dong, Pan Zhang, Yuhang Cao, Ziyu Liu, Shengyuan Ding, Shenxi Wu, Yubo Ma, Haodong Duan, Wenwei Zhang, Kai Chen, Dahua Lin, Jiaqi Wang
| Challenge: | Despite the promising performance of Large Vision Language Models, they sometimes generate incorrect outputs. |
| Approach: | They propose a multi-modal reward model that aligns LVLMs with human preferences. |
| Outcome: | The proposed model achieves excellent results on the latest multi-modal reward model benchmark and shows competitive performance on text-only reward model. |
RATE-Nav: Region-Aware Termination Enhancement for Zero-shot Object Navigation with Vision-Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Object navigation is a fundamental task in embodied artificial intelligence. |
| Approach: | They propose a region-aware Termination-Enhanced method that incorporates visual language models and exploration rates to enable efficient termination. |
| Outcome: | The proposed method achieves a success rate of 67.8% and an SPL of 31.3% on the HM3D dataset. |
RMoA: Optimizing Mixture-of-Agents through Diversity Maximization and Residual Compensation (2025.findings-acl)
Copied to clipboard
| Challenge: | Multi-agent systems based on large language models are limited by high computational overhead, information loss, and robustness. |
| Approach: | They propose a Residual Mixture-of-Agents (RMoA) that integrates residual connections to optimize efficiency and reliability. |
| Outcome: | The proposed model achieves state-of-the-art performance on benchmarks of alignment, mathematical reasoning, code generation, and multitasking understanding, while significantly reducing computational overhead. |
Instruction-Tuning Data Synthesis from Scratch via Web Reconstruction (2025.findings-acl)
Copied to clipboard
Yuxin Jiang, Yufei Wang, Chuhan Wu, Xinyi Dai, Yan Xu, Weinan Gan, Yasheng Wang, Xin Jiang, Lifeng Shang, Ruiming Tang, Wei Wang
| Challenge: | Existing methods for generating and curating high-quality instruction-tuning data rely heavily on the quality of seed data or strong assumptions about the structure and content of web documents. |
| Approach: | They propose a fully automated framework for synthesizing high-quality instruction-tuning (IT) data directly from raw web documents with minimal assumptions. |
| Outcome: | The proposed framework outperforms state-of-the-art baselines by 16.65% across four instruction-following benchmarks. |
RLKGF: Reinforcement Learning from Knowledge Graph Feedback Without Human Annotations (2025.findings-acl)
Copied to clipboard
| Challenge: | Lack of human preference labels remains a significant bottleneck when applying RLHF to a downstream domain. |
| Approach: | They propose a method that leverages human priors encoded in Knowledge Graphs (KGs) to derive RL rewards in the absence of manual annotations. |
| Outcome: | Experiments on three public and one private medical dialogue datasets show that the proposed method outperforms the competitive RLAIF in improving LLM diagnostic accuracy. |
Learning Task Representations from In-Context Learning (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for generalizing tasks to modalities beyond text fail to generalize effectively to linguistic tasks. |
| Approach: | They propose a method for encoding task information in ICL prompts as a function of attention heads within the transformer architecture. |
| Outcome: | The proposed method extracts task-specific information from in-context demonstrations and excels in both text and regression tasks. |
CAVGAN: Unifying Jailbreak and Defense of LLMs via Generative Adversarial Attacks on their Internal Representations (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing studies have isolated LLM jailbreak attacks and defenses . a new framework combines attack and defense to protect against malicious queries . |
| Approach: | They propose a framework that combines attack and defense to protect the Large Language Model (LLM) by embedding harmful problems into the safe area. |
| Outcome: | The proposed framework achieves an average jailbreak success rate of 88.85% across three popular LLMs while the defense success rate reaches an average of 84.17%. |
Firm or Fickle? Evaluating Large Language Models Consistency in Sequential Interactions (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have demonstrated remarkable capabilities across diverse tasks, but their deployment in high-stake domains requires consistent and coherent behavior across multiple rounds of user interaction. |
| Approach: | They propose a framework for evaluating and improving LLM response consistency, and introduce a benchmark dataset to evaluate LLM consistency. |
| Outcome: | The proposed framework improves response stability without sacrificing accuracy, and offers a practical path toward more dependable behavior in critical, real-world deployments. |
OS-Kairos: Adaptive Interaction for MLLM-Powered GUI Agents (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing efforts to build GUI agents focused on the autonomous mode have failed to address the problem of over-execution. |
| Approach: | They propose an adaptive GUI agent that predicts confidence levels at each interaction step and elicits adaptive interaction. |
| Outcome: | The proposed GUI agent outperforms existing models on a complex dataset and on established benchmarks. |
Red-Teaming LLM Multi-Agent Systems via Communication Attacks (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Model-based Multi-Agent Systems (LLM-MAS) have revolutionized complex problem-solving capability by enabling agent collaboration through message-based communications. |
| Approach: | They propose an attack that exploits communication mechanisms in Large Language Model-based Multi-Agent Systems (LLM-MAS) by intercepting and manipulating inter-agent messages. |
| Outcome: | The proposed attack exploits communication mechanisms in large language model-based multi-agent systems by intercepting and manipulating inter-agencies. |
Can We Trust AI Doctors? A Survey of Medical Hallucination in Large Language and Large Vision-Language Models (2025.findings-acl)
Copied to clipboard
Zhihong Zhu, Yunyan Zhang, Xianwei Zhuang, Fan Zhang, Zhongwei Wan, Yuyan Chen, QingqingLong QingqingLong, Yefeng Zheng, Xian Wu
| Challenge: | Hallucination is a critical challenge for large language models and large vision-language models (LVLMs) however, dedicated research on medical hallucinations remains unexplored. |
| Approach: | They provide a unified perspective on medical hallucination for both LLMs and LVLMs, and delve into its causes. |
| Outcome: | The proposed models have demonstrated impressive performance on a variety of medical benchmarks. |
DRT: Deep Reasoning Translation via Long Chain-of-Thought (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent studies have shown the effectiveness of long chain-of-thought (CoT) in reasoning tasks such as math and coding tasks. |
| Approach: | They propose to use Qwen2.5 and LLama-3.1 as backbones to train long thought models to bring the success of long chain-of-thought (CoT) to neural machine translation. |
| Outcome: | The proposed model outperforms vanilla LLMs and LLM models which are fine-tuning on paired sentences without long thought and outperformed vanilla LRMs. |
CTPD: Cross-Modal Temporal Pattern Discovery for Enhanced Multimodal Electronic Health Records Analysis (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for predicting clinical outcomes have focused on capturing temporal interactions within individual samples and fusing multimodal information, overlooking critical temporal patterns across different patients. |
| Approach: | They propose a cross-modal temporal pattern discovery framework to extract temporal patterns from multimodal EHR data. |
| Outcome: | The proposed framework extracts meaningful cross-modal temporal patterns from multimodal EHR data. |
Vision-aided Unsupervised Constituency Parsing with Multi-MLLM Debating (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing approaches require explicit cross-modal alignment, but new approaches address these challenges. |
| Approach: | They propose a framework for vision-aided unsupervised constituency parsing . they leverage multimodal large language models pre-trained on diverse image-text or video-text data . |
| Outcome: | The proposed framework achieves state-of-the-art performance on image-text and video-text datasets, improving robustness and accuracy. |
Inter-Passage Verification for Multi-evidence Multi-answer QA (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing multi-answer question answering systems struggle to retrieve and synthesize a large number of evidence passages. |
| Approach: | They propose a multi-answer question answering framework that generates a large set of passages and then processes each passage individually to generate an initial high-recall but noisy answer set. |
| Outcome: | The proposed framework outperforms baselines on the QAMPARI and RoMQA datasets, achieving an average F1 score improvement of 11.17%. |
PROMTEC: Fast LLM Inference Decoding using Prompt Multi-Lookup with Template Database and Common Sequences (2025.findings-acl)
Copied to clipboard
| Challenge: | Prompt Multi-Lookup, Template Datastore, and Common Sequences methods are used to accelerate the inference of large language models (LLMs). |
| Approach: | They propose a multi-faceted approach to accelerate the inference of large language models by leveraging Prompt Multi-Lookup, Template Datastore, and Common Sequences methods. |
| Outcome: | The proposed method achieves a 3.91 speedup on the miniF2F benchmark and 4.23 on the HumanEval benchmark. |
Logical DA: Enhancing Data Augmentation for Logical Reasoning via a Multi-Agent System (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing data augmentation paradigms isolate data synthesis from label validation, thereby reducing their utility for complex reasoning tasks. |
| Approach: | They propose a framework for enhancing reasoning-focused data augmentation in few-shot learning scenarios that integrates four agents through two synergistic phases: diverse data generation and label verification. |
| Outcome: | The proposed framework achieves the highest average improvement in task accuracy in both fine-tuning and in-context learning paradigms. |
Adapting General-Purpose Embedding Models to Private Datasets Using Keyword-based Retrieval (2025.findings-acl)
Copied to clipboard
| Challenge: | Text embedding models show strong performance on generic benchmarks, but their effectiveness diminishes when applied to private datasets. |
| Approach: | They propose a method for adapting general-purpose text embedding models to private datasets . they construct supervisory signals from the ranking of keyword-based retrieval results . |
| Outcome: | The proposed method improves retrieval performance across domains, datasets, and models. |
SQL Injection Jailbreak: A Structural Disaster of Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods to jailbreak Large Language Models (LLMs) exploited internal properties or capabilities of the model, such as optimization-based jailbreak methods and methods that leveraged the model’s context-learning abilities. |
| Approach: | They propose a new method which injects jailbreak information into user prompts and induces the model to generate harmful content. |
| Outcome: | The proposed method achieves near 100% success rates on open-source models while incurring lower time costs compared to previous methods. |
TAMP: Token-Adaptive Layerwise Pruning in Multimodal Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing pruning methods fail to account for unique token attributes across layers and modalities inherent to MLLMs. |
| Approach: | They propose a pruning framework that takes into account unique token attributes across layers and modalities inherent to MLLMs. |
| Outcome: | The proposed pruning framework outperforms existing pruning techniques on two state-of-the-art MLLMs. |
Generative Music Models’ Alignment with Professional and Amateur Users’ Expectations (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent years have witnessed rapid advances in text-to-music generation using large language models. |
| Approach: | They propose a task to align AI-generated music with human expressions . they use a dataset of over 1.5 million songs to analyze their content . |
| Outcome: | The proposed framework outperforms baseline models and facilitates end-to-end generation of songs audio. |
LLM-Forest: Ensemble Learning of LLMs with Graph-Augmented Prompts for Data Imputation (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing frameworks for missing data imputation are lacking in a finetuning-free process and mitigating biases and uncertainty in LLM outputs. |
| Approach: | They propose a framework for imputation of large language models with a forest of few-shot learning LLM "trees" they use bipartite information graphs to identify relevant neighboring entries with feature and value granularity. |
| Outcome: | The proposed framework is based on a concept of bipartite information graphs to identify high-quality relevant neighboring entries with both feature and value granularity. |
Task Calibration: Calibrating Large Language Models on Inference Tasks (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) have shown impressive zero-shot performance on inference tasks, however, they may suffer from spurious correlations between input texts and output labels, which limits their ability to reason based purely on general language understanding. |
| Approach: | They propose a zero-shot and inference-only calibration method inspired by mutual information which recovers LLM performance through task reformulation. |
| Outcome: | The proposed calibration method improves on 13 benchmarks and prompt templates and can be integrated with other calibration methods. |
MiniELM: A Lightweight and Adaptive Query Rewriting Framework for E-Commerce Search Optimization (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for rewriting query terms struggle with natural language understanding . generative methods face high inference latency and cost in offline settings . |
| Approach: | They propose a hybrid pipeline for rewriting query queries using offline knowledge distillation and online reinforcement learning. |
| Outcome: | The proposed pipeline improves query relevance, diversity, adaptability and cost-effective evaluation without manual annotations on Amazon ESCI dataset. |
Visibility as Survival: Generalizing NLP for Native Alaskan Language Identification (2025.findings-acl)
Copied to clipboard
| Challenge: | Indigenous languages are largely invisible in commercial language identification systems, a stark reality exemplified by Google Translate’s LangID tool, which excludes all 150 Indigenous languages of North America. |
| Approach: | They propose a framework that shows how large language models and specialized classifiers can effectively identify these languages with minimal data. |
| Outcome: | The proposed framework shows that large language models and specialized classifiers can effectively identify these languages with minimal data. |
KodCode: A Diverse, Challenging, and Verifiable Synthetic Dataset for Coding (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing code-focused resources typically fail to ensure either the breadth of coverage or verifiable correctness. |
| Approach: | They propose a synthetic dataset that provides high-quality, verifiable training data for Large Language Models for coding. |
| Outcome: | The proposed dataset surpasses Qwen2.5-Coder-32B-Instruct and DeepSeek-R1-Distill-Llama-70B in performance on coding benchmarks. |
Select, Read, and Write: A Multi-Agent Framework of Full-Text-based Related Work Generation (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for related work generation (RWG) suffer from shallow comprehension due to taking the limited portions of references as input and isolated explanation for each reference due to ineffective capturing the relationships among them. |
| Approach: | They propose a multi-agent framework that takes the limited portions of references papers as input and isolates the relationships between them. |
| Outcome: | The proposed framework outperforms other selectors and improves reading order with constrains of the graph structure. |
Graph-Assisted Culturally Adaptable Idiomatic Translation for Indic languages (2025.findings-acl)
Copied to clipboard
| Challenge: | a single source idiom can have multiple target-language equivalents depending on cultural references and contextual variations. |
| Approach: | They propose an adaptive graph neural network-based method that learns intricate mappings between idiomatic expressions and generalizes to both seen and unseen nodes during training. |
| Outcome: | The proposed method improves translation quality even in resource-constrained settings, facilitating improved idiomatic translation in smaller models. |
Question Answering in Climate Adaptation for Agriculture: Model Development and Evaluation with Expert Feedback (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing domain-specific question answering systems have generative capabilities, but their ability to answer climate adaptation questions remains unclear. |
| Approach: | They propose an iterative framework that enables LLMs to dynamically aggregate information from heterogeneous sources, such as climate literature and structured tabular climate data from climate model projections and historical observations. |
| Outcome: | The proposed framework enables LLMs to dynamically aggregate information from heterogeneous sources, such as text from climate literature and structured tabular climate data from climate model projections and historical observations. |
AGRec: Adapting Autoregressive Decoders with Graph Reasoning for LLM-based Sequential Recommendation (2025.findings-acl)
Copied to clipboard
| Challenge: | Autoregressive decoders in large language models excel at capturing sequential behaviors for generative recommendations, but they lack graph-structured user-item interactions, which are widely recognized as beneficial. |
| Approach: | They propose a novel algorithm that adapts LLMs’ decoders with graph reasoning for recommendation by augmenting the decoding logits with an auxiliary GNN model to optimize token generation. |
| Outcome: | The proposed model outperforms state-of-the-art models in sequential recommendations. |
Causal Denoising Prototypical Network for Few-Shot Multi-label Aspect Category Detection (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent methods that learn robust prototypes to represent aspects with limited support samples address noise categories in the support set that hinder their models from effective prototype generation. |
| Approach: | They propose a causal denoising prototypical network for few-shot MACD by learning robust prototypes to represent categories with limited support samples. |
| Outcome: | The proposed model outperforms baseline models and can prevent models from overly predicting more categories and mitigate semantic ambiguity issues among categories. |
RealHiTBench: A Comprehensive Realistic Hierarchical Table Benchmark for Evaluating LLM-Based Table Analysis (2025.findings-acl)
Copied to clipboard
Pengzuo Wu, Yuhang Yang, Guangcheng Zhu, Chao Ye, Hong Gu, Xu Lu, Ruixuan Xiao, Bowen Bao, Yijing He, Liangyu Zha, Wentao Ye, Junbo Zhao, Haobo Wang
| Challenge: | Existing benchmarks for large language models focus on simple, flat table structures. |
| Approach: | They propose a benchmark to evaluate the performance of both Large Language Models and Multimodal LLMs across a variety of input formats for complex tabular data, including LaTeX, HTML, and PNG. |
| Outcome: | The proposed benchmark evaluates the performance of LLMs and Multimodal LLM models across a variety of input formats for complex tabular data, including LaTeX, HTML, and PNG. |
A Query-Response Framework for Whole-Page Complex-Layout Document Image Translation with Relevant Regional Concentration (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for document image translation rely on the vanilla encoder-decoder paradigm . a novel dynamic aggregation mechanism is designed to enhance the text semantics in query features toward translation. |
| Approach: | They propose a Query-Response DIT framework that reformulates the DIT task into a parallel response/translation process of multiple queries. |
| Outcome: | The proposed framework improves translation quality on four translation directions on three benchmarks. |
DependEval: Benchmarking LLMs for Repository Dependency Understanding (2025.findings-acl)
Copied to clipboard
| Challenge: | a benchmark is designed to evaluate the repository-level dependency understanding of large language models (LLMs) based on 2683 repositories from real-world websites. |
| Approach: | They propose a benchmark to evaluate repository dependency understanding for large language models . DEPENDEVAL evaluates models on three core tasks across 8 programming languages . |
| Outcome: | The benchmark evaluates models on three core tasks across 8 programming languages from real-world repositories. |
A General Knowledge Injection Framework for ICD Coding (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods to improve ICD coding focus on a single type of knowledge and design specialized modules that are complex and incompatible with each other. |
| Approach: | They propose a general knowledge injection framework that integrates three key types of knowledge without specialized design of additional modules. |
| Outcome: | The proposed framework outperforms baseline models and is comparable to models relying on extra human annotations. |
MMUnlearner: Reformulating Multimodal Machine Unlearning in the Era of Multimodal Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent advances in machine learning (MU) have enabled the selective removal of private or sensitive information encoded within deep neural networks. |
| Approach: | They propose to "reformulate" the task of multimodal MU in the era of MLLMs by preserving only the visual patterns associated with a given entity while preserving the corresponding textual knowledge. |
| Outcome: | The proposed method surpasses baselines that finetuned MLLMs with VQA data directly through Gradient Ascent (GA) or Negative Preference Optimization (NPO), across all evaluation dimensions. |
Generating Questions, Answers, and Distractors for Videos: Exploring Semantic Uncertainty of Object Motions (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing video QADs are generated using video captions, incurring significant costs and resulting in bias. |
| Approach: | They propose to use temporal motion to describe video objects to generate diverse QADs focusing on different objects and interactions. |
| Outcome: | The proposed approach improves consistency and diversity of generated QADs on the NExT-QA and Perception Test benchmarks. |
DiffSkip: Differential Layer Skipping in Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing Large Language Models (LLMs) enforce uniform computation across all tokens. |
| Approach: | They propose to dynamically skip FFN blocks based on self-attention difference . they propose to use a lightweight router module to do the same . |
| Outcome: | The proposed model can skip FFN blocks even in token generation tasks. |
Towards Explainable Temporal Reasoning in Large Language Models: A Structure-Aware Generative Framework (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing studies on temporal reasoning models neglect the explainable reasoning processes underlying the results. |
| Approach: | They propose a structure-aware generative framework that integrates Graph structures with text for Explainable TEmporal Reasoning. |
| Outcome: | The proposed framework achieves state-of-the-art performance while also demonstrating robust generalization capabilities. |
A Bounding Box is Worth One Token - Interleaving Layout and Text in a Large Language Model for Document Understanding (2025.findings-acl)
Copied to clipboard
Jinghui Lu, Haiyang Yu, Yanjie Wang, Yongjie Ye, Jingqun Tang, Ziwei Yang, Binghong Wu, Qi Liu, Hao Feng, Han Wang, Hao Liu, Can Huang
| Challenge: | Existing methods for integrating spatial layouts with text have limitations . existing methods produce overly long text sequences or lack autoregressive traits of LLMs . |
| Approach: | They introduce Interleaving Layout and Text in a Large Language Model (LayTextLLM) they use OCR-derived text and spatial layouts to integrate with LLMs for document understanding . |
| Outcome: | The proposed model shows an increase in performance in KIE and VQA tasks. |
Self-Foveate: Enhancing Diversity and Difficulty of Synthesized Instructions from Unsupervised Text via Multi-Level Foveation (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for training large language models rely on human effort for data annotation. |
| Approach: | They propose an unsupervised method that generates unsupervised instruction from unsupervised text using a "Micro-Scatter-Macro" method that excavates fine-grained information embedded in unsupervised texts. |
| Outcome: | The proposed method improves diversity and difficulty of synthesized instructions across multiple unsupervised corpora and diverse model architectures. |
TableDreamer: Progressive and Weakness-guided Data Synthesis from Scratch for Table Instruction Tuning (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for table instruction tuning are limited due to limited data diversity and lack of data quality. |
| Approach: | They propose a weakness-guided data synthesis framework for table instruction tuning that explores the vast input space of table understanding tasks and then iterates through the input space. |
| Outcome: | The proposed framework boosts the average accuracy of Llama3.1-8B-instruct by 11.62% with 27K GPT-4o synthetic data and outperforms state-of-the-art data synthesis baselines which use more training data. |
Konooz: Multi-domain Multi-dialect Corpus for Named Entity Recognition (2025.findings-acl)
Copied to clipboard
| Challenge: | Using the Wojood framework, we compare existing Arabic Named Entity Recognition models with domain and dialect divergence and resource scarcity. |
| Approach: | They propose a multi-dimensional Arabic named entity corpus covering 16 dialects across 10 domains and an annotation scheme using the Wojood guidelines. |
| Outcome: | The proposed model performs better on 16 dialects across 10 domains and 16 domains, while other models struggle with different dialects and domains. |
Self-Rewarding Large Vision-Language Models for Optimizing Prompts in Text-to-Image Generation (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for rewriting text-to-image models require specialized vocabulary . a new approach uses large vision language models to optimize text-based models . |
| Approach: | They propose a prompt optimization framework that rephrases a user prompt into a text-to-image model by using large vision language models as solver and reward model. |
| Outcome: | The proposed model outperforms existing models on two popular datasets. |
CodeV: Issue Resolving with Visual Data (2025.findings-acl)
Copied to clipboard
Linhao Zhang, Daoguang Zan, Quanshun Yang, Zhirong Huang, Dong Chen, Bo Shen, Tianyu Liu, Yongshun Gong, Huang Pengjie, Xudong Lu, Guangtai Liang, Lizhen Cui, Qianxiang Wang
| Challenge: | Large Language Models (LLMs) have expanded to more complex repository-level tasks. |
| Approach: | They propose a first approach to leveraging visual data to enhance the issue-resolving capabilities of Large Language Models (LLMs) they demonstrate the effectiveness of CodeV and provide valuable insights into leveraging visualization to resolve GitHub issues. |
| Outcome: | The proposed approach improves the issue-resolving capabilities of Large Language Models (LLMs) by using visual data. |
A Survey of Large Language Models in Psychotherapy: Current Landscape and Future Directions (2025.findings-acl)
Copied to clipboard
Hongbin Na, Yining Hua, Zimu Wang, Tao Shen, Beibei Yu, Lilin Wang, Wei Wang, John Torous, Ling Chen
| Challenge: | Large language models (LLMs) can handle extensive context and multi-turn reasoning. |
| Approach: | They propose a taxonomy dividing psychotherapy into stages of assessment, diagnosis, and treatment to examine LLM advancements and challenges. |
| Outcome: | The proposed taxonomy reveals imbalances in current research, such as a focus on common disorders, linguistic biases, fragmented methods, and limited theoretical integration. |
Breaking the Reasoning Barrier A Survey on LLM Complex Reasoning through the Lens of Self-Evolution (2025.findings-acl)
Copied to clipboard
Tao He, Hao Li, Jingchang Chen, Runxuan Liu, Yixin Cao, Lizi Liao, Zihao Zheng, Zheng Chu, Jiafeng Liang, Ming Liu, Bing Qin
| Challenge: | OpenAI's O1 and subsequent projects like DeepSeek R1 have significantly advanced research on complex reasoning in LLMs. |
| Approach: | They analyze existing reasoning studies from the perspective of self-evolution and summarize O1-like works from open-source projects like DeepSeek R1 and Kimi-k1.5. |
| Outcome: | The proposed models are based on open-source models and pioneer advanced methodologies like Scaling Reinforcement Learning (RL). |
SEE: Continual Fine-tuning with Sequential Ensemble of Experts (2025.findings-acl)
Copied to clipboard
| Challenge: | Continual fine-tuning of large language models suffers from catastrophic forgetting . some approaches use routers to assign tasks to experts, but continual learning often requires retraining . |
| Approach: | They propose a framework that integrates routing and response mechanisms within each expert . it eliminates the need for an additional router and allows each expert to decide whether a query should be handled . |
| Outcome: | The proposed framework outperforms previous approaches in continual fine-tuning . it can handle learning tasks and out-of-distribution instances, paving the way for distributed model ensembling. |
Boosting Policy and Process Reward Models with Monte Carlo Tree Search in Open-Domain QA (2025.findings-acl)
Copied to clipboard
Chi-Min Chan, Chunpu Xu, Junqi Zhu, Jiaming Ji, Donghai Hong, Pengcheng Wen, Chunyang Jiang, Zhen Ye, Yaodong Yang, Wei Xue, Sirui Han, Yike Guo
| Challenge: | Experimental results show that our approach can effectively improve the performance of both the policy model and the reward model. |
| Approach: | They propose to use Monte Carlo Tree Search for both policy model improvement and reward model improvement to bridge it to more subtle open-domain question answering. |
| Outcome: | The proposed approach surpasses existing methods for annotation and training data with fewer data points and achieves better performance in test-time scaling strategies. |
Investigating and Enhancing Vision-Audio Capability in Omnimodal Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent years have witnessed significant advancements in large language models (LLMs) but still struggle with integrating vision and audio. |
| Approach: | They propose a self-knowledge distillation method to improve vision-audio capabilities of OLLMs by learning from the vision-text components. |
| Outcome: | The proposed method improves vision-audio capabilities of OLLMs by learning from vision-text components, which improves interaction between audio and images and results in improved performance on multimodal tasks. |
OpenHuEval: Evaluating Large Language Model on Hungarian Specifics (2025.findings-acl)
Copied to clipboard
Haote Yang, Xingjian Wei, Jiang Wu, Noémi Ligeti-Nagy, Jiaxing Sun, Yinfan Wang, Győző Zijian Yang, Junyuan Gao, Jingchao Wang, Bowen Jiang, Shasha Wang, Nanjun Yu, Zihao Zhang, Shixin Hong, Hongwei Liu, Wei Li, Songyang Zhang, Dahua Lin, Lijun Wu, Gábor Prószéky, Conghui He
| Challenge: | Recent advances in Large Language Models (LLMs) represent significant strides toward artificial general intelligence (AGI). |
| Approach: | They introduce OpenHuEval, the first benchmark for LLMs focusing on the Hungarian language and specifics. |
| Outcome: | The framework reveals intrinsic patterns and mechanisms of LLMs in non-English languages, with Hungarian serving as an example. |
StructFact: Reasoning Factual Knowledge from Structured Data with Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) have made significant strides in natural language processing by leveraging their ability to comprehend and reason with factual knowledge. |
| Approach: | They propose a benchmark to evaluate the ability of large language models to reason with structured data for knowledge-intensive tasks. |
| Outcome: | Extensive tests on 10 common LLMs show that they struggle with heterogeneity of structured data during reasoning. |
From Imitation to Introspection: Probing Self-Consciousness in Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing language models demonstrate impressive abilities in areas like natural language understanding, content creation, and reasoning. |
| Approach: | They propose a definition of self-consciousness for language models and refine ten core concepts by leveraging structural causal games. |
| Outcome: | The proposed definitions are based on structural causal games and ten core concepts. |
DocFusion: A Unified Framework for Document Parsing Tasks (2025.findings-acl)
Copied to clipboard
Mingxu Chai, Ziyu Shen, Chong Zhang, Yue Zhang, Xiao Wang, Shihan Dou, Jihua Kang, Jiazheng Zhang, Qi Zhang
| Challenge: | Existing methods for document parsing often employ multiple models, limiting performance . Existing models often employ discrete tokens, whereas recognition relies on continuous coordinates . |
| Approach: | They propose a Gaussian-Kernel Cross-Entropy Loss (GK-CEL) that unifies detection and recognition by enabling generative frameworks to handle both tasks simultaneously. |
| Outcome: | The proposed model performs competitively across four core document parsing tasks. |
Hierarchical Safety Realignment: Lightweight Restoration of Safety in Pruned Large Vision-Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent work has shown that pruning can reduce model performance, but it can also lead to degradation in safety performance. |
| Approach: | They propose a hierarchical safety realignment approach to prune large vision-Language Models . they quantify contribution of each attention head to safety and restore neurons . |
| Outcome: | The proposed approach achieves significant safety improvements in LVLMs pruned post pruning. |
LongDPO: Unlock Better Long-form Generation Abilities for LLMs via Critique-augmented Stepwise Information (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent advances in large language models have improved their capacity to handle long text inputs, but current models still exhibit unsatisfactory performance in long-form generation. |
| Approach: | They propose a method to enhance long-form text generation through step-level supervision by leveraging Monte Carlo Tree Search to collect stepwise preference pairs and employ a global memory pool to maintain factual accuracy. |
| Outcome: | The proposed method improves performance on long-form generation benchmarks while maintaining lossless performance on several general benchmarks. |
Reinforcing Compositional Retrieval: Retrieving Step-by-Step for Composing Informative Contexts (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have demonstrated remarkable capabilities across numerous tasks, yet they often rely on external context to handle complex tasks. |
| Approach: | They propose a tri-encoder sequential retriever that models a Markov Decision Process (MDP) this method decomposes the probability of retrieving a set of elements into a sequence of conditional probabilities and allows each retrieval step to be conditioned on previously selected examples. |
| Outcome: | The proposed method outperforms baselines and shows that it can handle multiple pieces of evidence or examples. |
Towards A Better Initial Policy Model For Scalable Long-CoT Reinforcement Learning (2025.findings-acl)
Copied to clipboard
| Challenge: | Long-CoT reasoning and reinforcement learning are demonstrating remarkable performance and scalability, however, there is a lack of systematic guidelines for obtaining a better initial policy model. |
| Approach: | They propose a systematic guideline and a novel Re-RFT method to obtain more efficient reasoning patterns from different initial models. |
| Outcome: | The proposed method surpasses DeepSeek-R1-Distill-Qwen-14B model by 4.6%, demonstrating its effectiveness and superiority. |
Topic Modeling for Short Texts via Optimal Transport-Based Clustering (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing approaches to topic modeling are based on probabilistic graphical models or non-negative matrix factorization techniques. |
| Approach: | They propose a method that aligns global clusters with topics to discover topics and learn document representations in topic space. |
| Outcome: | The proposed method outperforms state-of-the-art techniques in short-text topic modeling across commonly used metrics. |
Lemmatisation & Morphological Analysis of Unedited Greek: Do Simple Tasks Need Complex Solutions? (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing systems for part-of-speech tagging of unedited Greek text have outperformed traditional methods for morphological analysis and lemmatisation. |
| Approach: | They propose to combine nominal features into a single label and combine the three most distinctive features of verbs into another unified label. |
| Outcome: | The proposed models outperform traditional models in lemmatisation and morphological analysis and show that multi-task learning improves performance by transferring parameters. |
FRAME: Feedback-Refined Agent Methodology for Enhancing Medical Research Insights (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing approaches to automate scientific research are limited by human cognitive constraints and timeintensive workflows. |
| Approach: | They propose a framework that enhances medical paper generation through iterative refinement and structured feedback. |
| Outcome: | The proposed framework achieves significant improvements over conventional methods across multiple models and evaluation dimensions. |
Chain-of-Scrutiny: Detecting Backdoor Attacks for Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have demonstrated impressive capabilities across various domains, but are vulnerable to backdoor attacks. |
| Approach: | They propose a chain-of-scrutiny approach which leverages LLMs’ unique reasoning abilities to mitigate backdoor attacks. |
| Outcome: | The proposed model is well-suited for the popular API-only LLM deployments, enabling detection at minimal cost and with little data. |
Relevance Scores Calibration for Ranked List Truncation via TMP Adapter (2025.findings-acl)
Copied to clipboard
| Challenge: | ranked list truncation methods struggle with limited capacity, unstable training and inconsistency of selected thresholds. |
| Approach: | They propose a new approach that incorporates the Threshold Margin Penalty as an additive loss function to calibrate ranking model relevance scores for ranked list truncation. |
| Outcome: | The proposed method improves on retrieval datasets and offers theoretical and practical benefits. |
Neuron Activation Modulation for Text Style Transfer: Guiding Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Text style transfer (TST) aims to flexibly adjust the style of text while preserving its core content. |
| Approach: | They propose a method that aligns activation values of style-related neurons with those of the target style to guide the model in performing the transfer. |
| Outcome: | The proposed method significantly improves style transfer quality while preserving core content. |
MTVQA: Benchmarking Multilingual Text-Centric Visual Question Answering (2025.findings-acl)
Copied to clipboard
Jingqun Tang, Qi Liu, Yongjie Ye, Jinghui Lu, Shu Wei, An-Lan Wang, Chunhui Lin, Hao Feng, Zhen Zhao, Yanjie Wang, Yuliang Liu, Hao Liu, Xiang Bai, Can Huang
| Challenge: | Text-Centric Visual Question Answering (TEC-VQA) is a text-centric visual task understanding tool. |
| Approach: | They introduce a benchmark that features human expert annotations across 9 languages . they prioritize the text in question-answer pairs while disregarding visual text in images . |
| Outcome: | The proposed benchmarks prioritize the text in question-answer pairs while disregarding visual text in images. |
HICD: Hallucination-Inducing via Attention Dispersion for Contrastive Decoding to Mitigate Hallucinations in Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) often generate hallucinations, producing outputs that are contextually inaccurate or factually incorrect. |
| Approach: | They propose a method that selects attention heads crucial to the model's prediction as inducing heads and induces hallucinations by dispersing attention of these inducers. |
| Outcome: | The proposed method significantly improves performance on tasks requiring contextual faithfulness, reading comprehension, and question answering. |
Understanding the Repeat Curse in Large Language Models from a Feature Perspective (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models suffer from repetitive text generation, a phenomenon we refer to as the ”Repeat Curse”. |
| Approach: | They propose a method to induce and analyze the Repeat Curse in large language models by using mechanistic interpretability. |
| Outcome: | The proposed method induces and analyzes the Repeat Curse in large language models using mechanistic interpretability. |
Code-Switching Curriculum Learning for Multilingual Transfer in LLMs (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) exhibit near human-level performance in various tasks, but performance drops after a handful of high-resource languages due to the imbalance in pre-training data. |
| Approach: | They propose a code-switching curriculum learning model to enhance cross-lingual transfer for LLMs by progressively training models with a curriculum consisting of token-level code-changing, sentence-level codeswitching, and monolingual corpora. |
| Outcome: | The proposed model improves language transfer to Korean, with significant gains in Japanese and Indonesian . the proposed model mitigates spurious correlations between language resources and safety alignment . |
A Mousetrap: Fooling Large Reasoning Models for Jailbreak with Chain of Iterative Chaos (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Reasoning Models (LRMs) have advanced beyond traditional Large Language Models, yet they pose heightened safety risks. |
| Approach: | They propose a first jailbreak attack targeting Large Reasoning Models . they exploit a Chaos Machine component to transform attack prompts with diverse one-to-one mappings based on the reasoning chain . |
| Outcome: | The proposed attack exploits the unique vulnerabilities of LRMs by integrating a Chaos Machine. success rates of the mousetrap attack are as high as 96%, 86% and 98% respectively. |
Tag-Evol: Achieving Efficient Instruction Evolving via Tag Injection (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods rely on a fixed set of strategies to evolve, which requires manual design and is monolithic in form. |
| Approach: | They propose a method that uses diverse and specific knowledge tags to achieve controlled evolution by injecting different combinations of tags into original instructions. |
| Outcome: | The proposed method generates better evolved data than existing methods and is more diverse and challenging. |
Breaking the Ceiling: Exploring the Potential of Jailbreak Attacks through Expanding Strategy Space (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods to exploit black-box jailbreaks fail to capture key attack patterns . a novel framework decomposes jailbreak strategies into essential components . |
| Approach: | They propose a framework that decomposes jailbreak strategies into essential components and develops genetic-based optimization with intention evaluation mechanisms. |
| Outcome: | The proposed framework achieves 90% success rate on Claude-3.5, where prior methods completely fail . it also surpasses specialized safeguard models in evaluation accuracy . |
GeNRe: A French Gender-Neutral Rewriting System Using Collective Nouns (2025.findings-acl)
Copied to clipboard
| Challenge: | Gender rewriting is an NLP task that uses gendered forms to mitigate gender biases. |
| Approach: | They propose a French gender-neutral rewriting system using collective nouns, which are gender-fixed in French. |
| Outcome: | The proposed system detects gendered forms and replaces them with neutral or opposite forms. |
LGAR: Zero-Shot LLM-Guided Neural Ranking for Abstract Screening in Systematic Literature Reviews (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for abstract screening focus on binary classification settings; existing question answering (QA) based ranking approaches suffer from error propagation. |
| Approach: | They propose a systematic literature review (SLR) method that uses large language models to evaluate the SLR's inclusion and exclusion criteria. |
| Outcome: | The proposed method outperforms existing question answering (QA) based methods by 5-10 pp. in mean precision. |
LCHAIM - Investigating Long Context Reasoning in Hebrew (2025.findings-acl)
Copied to clipboard
| Challenge: | a dataset designed to evaluate long premises and complex reasoning models in Hebrew is being used in research. |
| Approach: | They introduce a dataset to evaluate Hebrew NLI models on long premises and complex reasoning tasks. |
| Outcome: | The proposed dataset evaluates models on tasks involving long premises and complex reasoning. |
CLeVeR: Multi-modal Contrastive Learning for Vulnerability Code Representation (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for detecting code capture the overall semantics of the code rather than its intrinsic vulnerability-specific semantics. |
| Approach: | They propose an approach that leverages contrastive learning to generate precise vulnerability code representations under the supervision of vulnerability descriptions. |
| Outcome: | The proposed approach outperforms state-of-the-art methods in vulnerability detection tasks by 11.85% and 13.61%. |
MEMIT-Merge: Addressing MEMIT’s Key-Value Conflicts in Same-Subject Batch Editing for LLMs (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing knowledge editing techniques that modify models’ internal knowledge without full model retraining have gained significant attention. |
| Approach: | They propose an enhanced approach that merges value computation processes for facts sharing the same subject to improve editing efficiency. |
| Outcome: | The proposed method maintains a 98% editing success rate on same-subject and distinct-sub subject datasets while the original success rate drops to 46%. |
Large Language Models for Predictive Analysis: How Far Are They? (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing studies on LLMs do not evaluate their capability in predictive analysis. |
| Approach: | They propose a benchmark to evaluate Large Language Models (LLMs) they integrate 1130 queries from 44 real-world datasets of 8 different fields to evaluate their capability . |
| Outcome: | The proposed benchmark evaluates 12 renowned LLMs from 44 real-world datasets . results offer insights into their practical use in predictive analysis . |
Think More, Hallucinate Less: Mitigating Hallucinations via Dual Process of Fast and Slow Thinking (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing studies have focused on the issue of hallucination in large language models. |
| Approach: | They propose a framework that allows an explicit slow thinking generation process for mitigating hallucinations during inference. |
| Outcome: | The proposed framework outperforms baseline approaches on English and Chinese datasets. |
Towards Adaptive Memory-Based Optimization for Enhanced Retrieval-Augmented Generation (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for enhancing response accuracy and accuracy struggle with open-domain QA tasks because they perform independent retrieval operations without maintaining a summarizing memory or using adaptive retrieval strategies. |
| Approach: | They propose a method that integrates non-parametric knowledge from external knowledge bases into models to enhance response accuracy while mitigating factual errors and hallucinations. |
| Outcome: | The proposed method improves on open-domain QA datasets and reduces noise and hallucinations due to redundant information and insufficient information integration. |
Enhancing Cross-Tokenizer Knowledge Distillation with Contextual Dynamical Mapping (2025.findings-acl)
Copied to clipboard
| Challenge: | Knowledge distillation (KD) approaches focus on homogeneous architectures with identical tokenizers, constraining their applicability in cross-architecture scenarios. |
| Approach: | They propose a framework that uses contextual information to enhance sequence alignment precision and dynamically improves vocabulary mapping. |
| Outcome: | The proposed framework shows significant advantages over existing methods for model compression . it can be used across multiple model families and across multiple benchmarks . |
A Semantic-Aware Layer-Freezing Approach to Computation-Efficient Fine-Tuning of Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing work on how to finetune but neglects the issue of where to fine-tune language models is expensive. |
| Approach: | They propose to use transition traces of latent representation to compute deviations (or loss) and then estimate the gain of each layer in reducing deviation (or gain). |
| Outcome: | The proposed approach outperforms baseline methods and is cost-benefit balanced. |
CNNSum: Exploring Long-Context Summarization with Large Language Models in Chinese Novels (2025.findings-acl)
Copied to clipboard
| Challenge: | Currently, long-context summarization mainly relies on memory ability. |
| Approach: | They propose a multi-scale long-context summarization benchmark based on Chinese novels . they use human-driven annotations to analyze long-constituency models . |
| Outcome: | The proposed benchmark features human-driven annotations across four subsets with lengths ranging from 16k to 128k. |
Document Segmentation Matters for Retrieval-Augmented Generation (2025.findings-acl)
Copied to clipboard
Zhitong Wang, Cheng Gao, Chaojun Xiao, Yufei Huang, Shuzheng Si, Kangyang Luo, Yuzhuo Bai, Wenhao Li, Tangjian Duan, Chuancheng Lv, Guoshan Lu, Gang Chen, Fanchao Qi, Maosong Sun
| Challenge: | Existing rule-based chunking methods lead to suboptimal splits, where overly large chunks introduce irrelevant information and small chunks lack semantic coherence. |
| Approach: | They propose a method that leverages document summaries as pseudo-instructions to guide chunking by computing semantic similarity between sentences and the summary. |
| Outcome: | Experiments on multiple open-domain question-answering benchmarks show that PIC significantly improves retrieval accuracy (Hits@k) and end-to-end QA performance (Exact Match) without any additional training. |
UBench: Benchmarking Uncertainty in Large Language Models with Multiple Choice Questions (2025.findings-acl)
Copied to clipboard
Xunzhi Wang, Zhuowei Zhang, Gaonan Chen, Qiongyu Li, Bitong Luo, Zhixin Han, Haotian Wang, Zhiyu Li, Hang Gao, Mengting Hu
| Challenge: | Existing methods for benchmarking the uncertainty of large language models face challenges . existing methods require internal model access, additional training, or high computational costs . |
| Approach: | They propose a new benchmark for evaluating the uncertainty of large language models based on confidence intervals . UBench encompasses 11,978 multiple choice questions spanning knowledge, language, understanding, and reasoning capabilities. |
| Outcome: | The proposed method outperforms existing methods for benchmarking the uncertainty of large language models. |
Embracing Large Language Models in Traffic Flow Forecasting (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods to predict future traffic flows capture spatio-temporal dependencies, but they fail to adapt to test-time environmental changes. |
| Approach: | They propose to use large language models to help traffic flow forecasting by capturing spatio-temporal dependencies and using a large language model to select the most likely result. |
| Outcome: | The proposed method is based on large language models (LLMs) and an LLM-based selector. |
Flow2Code: Evaluating Large Language Models for Flowchart-based Code Generation Capability (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing code generation benchmarks neglect flowchart-based code generation . existing benchmarks lack flowcharting-based evaluation, limiting the potential of large language models and minimizing human error. |
| Approach: | They propose to use flowcharts to evaluate existing LLMs' code generation capabilities. |
| Outcome: | The proposed benchmarks show that the supervised fine-tuning technique contributes greatly to the models’ performance. |
Smarter, Not Harder: Training-Free Adaptive Computation for Transformers (2025.findings-acl)
Copied to clipboard
| Challenge: | Adaptive Computation in Transformers (ACT) aims to reduce computation for simpler inferences while enhancing performance by allocating more computation to complex inference steps. |
| Approach: | They propose a method that perturbs network weights rather than input embeddings to improve performance. |
| Outcome: | The proposed method outperforms beam search and hesitation-based methods but suffers from inefficiency and instability due to its reliance on randomness. |
UCS-SQL: Uniting Content and Structure for Enhanced Semantic Bridging In Text-to-SQL (2025.findings-acl)
Copied to clipboard
Zhenhe Wu, Zhongqiu Li, JieZhangChinaTele JieZhangChinaTele, Zhongjiang He, Jian Yang, Yu Zhao, Ruiyu Fang, Bing Wang, Hongyan Xie, Shuangyong Song, Zhoujun Li
| Challenge: | Existing methods overlook the challenge of effectively transforming structure information from NL to SQL. |
| Approach: | They propose a text-to-SQL framework that unites content and structure pipes to bridge the gap between NL and SQL. |
| Outcome: | The proposed framework bridges the gap between natural language questions and SQL by combining content and structure pipes. |
CodePRM: Execution Feedback-enhanced Process Reward Model for Code Generation (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent advances in code generation focus on optimizing the thought process, but lack effective process supervision, making it difficult to optimize the thoughts. |
| Approach: | They propose a method that leverages the code execution feedback to build a code PRM by collecting a large dataset of thought traces and then training it to take both the reasoning process and code execution as input. |
| Outcome: | The proposed approach outperforms baselines and strong LLMs in the inference stage. |
STEM-POM: Evaluating Language Models Math-Symbol Reasoning in Document Parsing (2025.findings-acl)
Copied to clipboard
| Challenge: | Advances in large language models have spurred research into enhancing their reasoning capabilities, particularly in math-rich STEM documents. |
| Approach: | They propose a benchmark dataset to evaluate LLMs’ reasoning abilities on math symbols within contextual scientific text. |
| Outcome: | The proposed dataset demonstrates that state-of-the-art LLMs achieve an average accuracy of 20-60% under in-context learning and 50-60% with fine-tuning, highlighting a substantial gap in their ability to classify mathematical symbols. |
Retrieval Visual Contrastive Decoding to Mitigate Object Hallucinations in Large Vision-Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Vision Language Models are not free from the issue of Object Hallucination (OH) OH is a phenomenon where LVLMs generate hallucinated objects and descriptions in their outputs. |
| Approach: | They propose a method to suppress OH by referencing images from AI-generated images at the logit level. |
| Outcome: | The proposed method significantly improves existing methods on visual contrast decoding. |
Leveraging LLMs for Bangla Grammar Error Correction: Error Categorization, Synthetic Data, and Model Evaluation (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) perform exceedingly well in Natural Language Understanding tasks for many languages including English. |
| Approach: | They propose to use a rule-based noise injection method to create grammatically incorrect sentences . they categorize 12 error classes in Bangla and take a survey of native speakers . |
| Outcome: | The proposed method improves performance of LLMs in Bangla by 3-7 percentage points compared to zero-shot setting . human errors are still superior in error correction, the authors show . |
Think Both Ways: Teacher-Student Bidirectional Reasoning Enhances MCQ Generation and Distractor Quality (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for generating high-quality MCQs struggle with contextual relevance and plausible distractors. |
| Approach: | They propose a framework that integrates bidirectional reasoning perspectives to generate contextually relevant questions and plausible distractors while student reasoning evaluates question clarity and the misleading nature of distractors. |
| Outcome: | The proposed framework outperforms existing methods in generating text-grounded questions and high-quality distractors for narrative contexts. |
mmE5: Improving Multimodal Multilingual Embeddings via High-quality Synthetic Data (2025.findings-acl)
Copied to clipboard
| Challenge: | Multimodal embedding models encode multimedia inputs into latent vector representations. |
| Approach: | They propose to synthesize multimodal multilingual data using a multimodal large language model . they identify three criteria for high-quality synthetic multimodal data . |
| Outcome: | The proposed model outperforms existing models on the MMEB Benchmark and the XTD benchmark. |
Word2Passage: Word-level Importance Re-weighting for Query Expansion (2025.findings-acl)
Copied to clipboard
| Challenge: | Retrieval-augmented generation (RAG) enhances the quality of LLM generation by providing relevant chunks, but retrieving accurately from external knowledge remains challenging due to missing contextually important words in query expansion. |
| Approach: | They propose a method that generates word, sentence, and passage references for query expansion and assigns distinct importance scores to words based on their origin and characteristics. |
| Outcome: | The proposed method outperforms existing methods across datasets and LLM configurations, effectively enhancing retrieval accuracy and generation quality. |
MECoT: Markov Emotional Chain-of-Thought for Personality-Consistent Role-Playing (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing Large Language Models struggle to maintain emotionally consistent and psychologically plausible character personalities. |
| Approach: | They propose a framework that enhances LLMs’ ability to generate authentic personality-driven dialogues through stochastic emotional transitions. |
| Outcome: | The proposed framework achieves 93.3% emotional accuracy on the RAPD dataset and significantly outperforms existing approaches. |
FiDeLiS: Faithful Reasoning in Large Language Models for Knowledge Graph Question Answering (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing retrieval-based or agent-based methods are prone to generating erroneous or hallucinated outputs. |
| Approach: | They propose a framework to leverage knowledge graphs as external knowledge sources to improve the factuality of LLM responses by anchoring answers to verifiable reasoning steps retrieved from KGs. |
| Outcome: | The proposed framework improves factuality and interpretability across benchmarks and reduces computational costs. |
REALM: A Dataset of Real-World LLM Use Cases (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing studies on LLM adoption and their social implications lack empirical grounding, weakening their validity. |
| Approach: | They propose to integrate a dataset of over 94,000 LLM use cases collected from Reddit and news articles to provide insights into LLM adoption across different domains. |
| Outcome: | The proposed dataset includes over 94,000 LLM use cases collected from Reddit and news articles. |
BABELEDITS: A Benchmark and a Modular Approach for Robust Cross-lingual Knowledge Editing of Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for cross-lingual knowledge editing are limited in their effectiveness and robustness. |
| Approach: | They propose a new CKE benchmark that accounts for the rich variety of entity aliases within and across languages. |
| Outcome: | The proposed method is more effective than state-of-the-art methods and robust against model collapse when subjected to multiple edits. |
CDS: Data Synthesis Method Guided by Cognitive Diagnosis Theory (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have demonstrated remarkable capabilities across diverse tasks, but the complexity of emerging tasks and higher performance demands highlight the need for continuous improvement. |
| Approach: | They propose a method that refines evaluation results and characterizes model profiles at the knowledge component level. |
| Outcome: | The proposed method improves performance across multiple benchmarks and academic exams. |
Problem-Solving Logic Guided Curriculum In-Context Learning for LLMs Complex Reasoning (2025.findings-acl)
Copied to clipboard
| Challenge: | In-context learning (ICL) can enhance the complex reasoning capabilities of large language models. |
| Approach: | They propose a curriculum ICL strategy guided by problem-solving logic to select demonstration examples based on curriculum learning. |
| Outcome: | The proposed strategy outperforms existing methods on multiple benchmarks. |
BESSTIE: A Benchmark for Sentiment and Sarcasm Classification for Varieties of English (2025.findings-acl)
Copied to clipboard
| Challenge: | despite large language models showing bias against non-mainstream varieties, there are no labeled datasets for sentiment analysis of English. |
| Approach: | They propose a benchmark for sentiment and sarcasm classification for three varieties of English . they manually annotate the datasets with sentiment and the sarcasmatic labels . |
| Outcome: | The proposed benchmark is based on a web-based content from Google Place reviews and Reddit comments. |
NavRAG: Generating User Demand Instructions for Embodied Navigation through Retrieval-Augmented LLM (2025.findings-acl)
Copied to clipboard
| Challenge: | High-performance vision-and-language navigation models require large amounts of training data, the high cost of manual annotating has seriously hindered this field. |
| Approach: | They propose a retrieval-augmented generation framework that generates user demand instructions for vision-and-language navigation. |
| Outcome: | The proposed model achieves SOTA performance on the REVERIE benchmark. |
SQLForge: Synthesizing Reliable and Diverse Data to Enhance Text-to-SQL Reasoning in LLMs (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing closed-source LLMs have a performance gap in text-to-SQL reasoning tasks. |
| Approach: | They propose a SQL-based approach to synthesize reliable data to enhance text-to-SQL reasoning in LLMs. |
| Outcome: | The proposed model achieves state-of-the-art accuracy on the widely recognized Spider and BIRD benchmarks, significantly narrowing the performance gap with closed-source methods. |
Retrieval-Augmented Process Reward Model for Generalizable Mathematical Reasoning (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) have advanced mathematical reasoning, but they still struggle with out-of-distribution (OOD) issues. |
| Approach: | They propose a framework to evaluate the logical validity of reasoning steps . they retrieves semantically similar questions and steps for PRM as a warmup . |
| Outcome: | The proposed framework outperforms baseline models on multiple real-world datasets. |
Contrastive Learning for Task-Independent SpeechLLM-Pretraining (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models excel in speech processing tasks but their reliance on written text limits their application in real-world scenarios. |
| Approach: | They propose a task-independent speech pretraining stage and task-specific fine-tuning stage to adapt LLMs to speech processing tasks. |
| Outcome: | The proposed model outperforms models specialized on speech translation and question answering while being trained on 10% of the task-specific data. |
QiMeng-Attention: SOTA Attention Operator is generated by SOTA Attention Algorithm (2025.findings-acl)
Copied to clipboard
Qirui Zhou, Shaohui Peng, Weiqiang Xiong, Haixin Chen, Yuanbo Wen, Haochen Li, Ling Li, Qi Guo, Yongwei Zhao, Ke Gao, Ruizhi Chen, Yanjun Wu, Zhao Chen, Yunji Chen
| Challenge: | Existing LLMs cannot comprehend the complex data flow and computation process of the attention operator and utilize low-level primitive to exploit GPU performance. |
| Approach: | They propose an LLM-friendly Thinking Language (LLM-TL) that can decouple the generation of high-level optimization logic and low-level implementation on GPU and enhance LLMs’ understanding of attention operator. |
| Outcome: | The proposed method outshines existing LLMs on A100, RTX8000, and T4 GPUs, achieving a speed-up of up to 35.16. |
ALW: Adaptive Layer-Wise contrastive decoding enhancing reasoning ability in Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing research has demonstrated that contrast decoding of two different models can improve text quality in open-ended text generation but with limited gains on reasoning tasks. |
| Approach: | They propose a framework that dynamically disentangles noise in shallow layers from critical signals in deep layers to enhance reasoning ability. |
| Outcome: | The proposed framework improves answer accuracy while maintaining inference efficiency. |
Mixture of Decoding: An Attention-Inspired Adaptive Decoding Strategy to Mitigate Hallucinations in Large Vision-Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Vision-Language Models (LVLMs) have impressive capabilities across visual tasks, yet they remain hindered by the persistent challenge of hallucinations. |
| Approach: | They propose a novel approach that dynamically adapts decoding strategies by evaluating the correctness of the model’s attention on image tokens to distinguish the correct attention. |
| Outcome: | Extensive experiments show that the proposed approach outperforms existing decoding methods across multiple mainstream benchmarks, effectively mitigating hallucinations in LVLMs. |
VidCapBench: A Comprehensive Benchmark of Video Captioning for Controllable Text-to-Video Generation (2025.findings-acl)
Copied to clipboard
Xinlong Chen, Yuanxing Zhang, Chongling Rao, Yushuo Guan, Jiaheng Liu, Fuzheng Zhang, Chengru Song, Qiang Liu, Di Zhang, Tieniu Tan
| Challenge: | Existing studies have not identified a link between video caption evaluation and T2V generation. |
| Approach: | They propose a video caption evaluation scheme specifically designed for T2V generation that integrates video annotation with caption evaluation. |
| Outcome: | The proposed system is agnostic to any particular caption format and can be used for training. |
Mitigating Demonstration Bias through Global Coevolutionary Reasoning (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for chain-of-thought prompting rely on manual demonstrations . experimental results show that GCR outperforms baseline methods without performance degradation . |
| Approach: | They propose a method that uses random samples to generate demonstrations in zero-shot settings. |
| Outcome: | The proposed method outperforms baseline methods on ten datasets without demonstration bias. |
A Representation Level Analysis of NMT Model Robustness to Grammatical Errors (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing work on robustness failures or improving robustness has focused on documenting failures . however, there has been limited analysis of model representations in response to noise. |
| Approach: | They perform Grammatical Error Detection probing and representational similarity analysis to examine model representations of ungrammatical inputs and how they evolve through model layers. |
| Outcome: | The proposed model detects and corrects the grammatical error by moving its representation toward the correct form. |
T2DR: A Two-Tier Deficiency-Resistant Framework for Incomplete Multimodal Learning (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing incomplete multimodal learning frameworks are inadequate for integrating multimodal data. |
| Approach: | They propose a framework for incomplete multimodal learning that is deficiency-resistant and provides two modules to address fine-grained deficiencies. |
| Outcome: | The proposed framework outperforms the SOTA models on two well-known multimodal benchmarks. |
From Specific-MLLMs to Omni-MLLMs: A Survey on MLLMs Aligned with Multi-modalities (2025.findings-acl)
Copied to clipboard
Shixin Jiang, Jiafeng Liang, Jiyuan Wang, Xuan Dong, Heng Chang, Weijiang Yu, Jinhua Du, Ming Liu, Bing Qin
| Challenge: | MLLMs are able to integrate multiple modalities into a single model to tackle complex tasks in real-world scenarios. |
| Approach: | They propose a comprehensive survey of Omni-MLLMs to address the challenges and opportunities of multimodal modeling. |
| Outcome: | The proposed model can integrate multiple modalities into a single model and provide novel perspectives. |
Analyzing the Effect of Linguistic Similarity on Cross-Lingual Transfer: Tasks and Experimental Setups Matter (2025.findings-acl)
Copied to clipboard
| Challenge: | Prior work on cross-lingual transfer often focuses on a small set of languages from a few language families and/or a single task. |
| Approach: | They analyze cross-lingual transfer for 263 languages from a wide variety of language families . they include three popular NLP tasks: POS tagging, dependency parsing, topic classification . |
| Outcome: | The proposed approach is based on linguistic similarity measures for 263 languages . the results show that the effect of linguistic similarities on transfer performance depends on a range of factors . |
Agents generalize to novel levels of abstraction by using adaptive linguistic strategies (2025.findings-acl)
Copied to clipboard
| Challenge: | Abstract: Abstracts are fundamental to building well-generalizing models. |
| Approach: | They propose to use a concept-level reference game to generalize concepts . they find that agents can learn robust concepts based on which they can generalize . |
| Outcome: | The proposed model can generalize from generic to very specific concepts, while reusing many messages from training. |
The Linguistic Connectivities Within Large Language Models (2025.findings-acl)
Copied to clipboard
Dan Wang, Boxi Cao, Ning Bian, Xuanang Chen, Yaojie Lu, Hongyu Lin, Jia Zheng, Le Sun, Shanshan Jiang, Bin Dong, Xianpei Han
| Challenge: | Recent studies have discovered notable disparities in their performance across different languages. |
| Approach: | They conduct a systematic investigation into the behaviors of large language models across 27 different languages on 3 different scenarios and reveals a Linguistic Map correlates with the richness of available resources and linguistic family relations. |
| Outcome: | The proposed model demonstrates that there are significant disparities in performance across languages across 27 different languages on 3 different scenarios. |
XFinBench: Benchmarking LLMs in Complex Financial Problem Solving and Reasoning (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing large language models (LLMs) lack advanced capabilities such as temporal reasoning, future forecasting, and numerical modeling. |
| Approach: | They propose a benchmarking tool to evaluate LLMs' ability to solve complex financial problems across diverse graduate-level finance topics with multi-modal context. |
| Outcome: | The proposed model improves on the o1 model but still lags behind human experts in temporal reasoning and scenario planning capabilities. |
Align2LLaVA: Cascaded Human and Large Language Model Preference Alignment for Multi-modal Instruction Curation (2025.findings-acl)
Copied to clipboard
Hongzhe Huang, Jiang Liu, Zhewen Yu, Li Cai, Dian Jiao, Wenqiao Zhang, Siliang Tang, Juncheng Li, Hao Jiang, Haoyuan Li, Yueting Zhuang
| Challenge: | Recent advances in Multi-modal Large Language Models (MLLMs) introduce significant variability in data quality. |
| Approach: | They propose to use human and LLM preference alignment to compress large corpus of machine-generated multimodal instructions into a compact and high-quality form. |
| Outcome: | The proposed algorithm outperforms LLaVA-series models in MLLM benchmarks by 90% . it uses human and LLM preference alignment to compress a large dataset . |
Achieving binary weight and activation for LLMs using Post-Training Quantization (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for quantizing large language models suffer from performance degradation when weights are quantized to 1 bit. |
| Approach: | They propose a post-training quantization framework with W(1+1)A(14) configuration . they propose utilizing Hessian-aware fine-grained grouping along with an EM-based quantization scheme . |
| Outcome: | The proposed method surpasses state-of-the-art (SOTA) LLM quantization baselines on W2A4 across multiple tasks. |
Mitigating Negative Interference in Multilingual Knowledge Editing through Null-Space Constraints (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing monolingual knowledge editing methods are expensive and require multiple models to maintain factual consistency. |
| Approach: | They propose a null-space constrained framework to precisely isolate language-specific knowledge updates that can be mapped onto other languages’ subspaces. |
| Outcome: | The proposed framework can project parameter updates for each language onto the orthogonal complement of other languages’ subspaces while preserving multilingual generalization capabilities. |
From Awareness to Adaptability: Enhancing Tool Utilization for Scientific Reasoning (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing approaches enhance reasoning through Chain-of-Thought, Program-ofThough, and Tool-Integration. |
| Approach: | They propose a tool-awareness training method that leverages both forward and backward data generation strategies to strengthen the model’s conscious and selective tool utilization in multi-step reasoning tasks. |
| Outcome: | The proposed method improves the model's tool utilization capabilities, including proactivity and execution success rates. |
AMoPO: Adaptive Multi-objective Preference Optimization without Reward Models and Reference Models (2025.findings-acl)
Copied to clipboard
Qi Liu, Jingqing Ruan, Hao Li, Haodong Zhao, Desheng Wang, Jiansong Chen, Wan Guanglu, Xunliang Cai, Zhi Zheng, Tong Xu
| Challenge: | Existing multi-objective preference alignment methods for large language models face limitations such as auxiliary reward/reference models and computational complexity. |
| Approach: | They propose a framework that achieves dynamic balance across preference dimensions by using dimension-aware generation metrics as implicit rewards. |
| Outcome: | Empirical results show that AMoPO outperforms state-of-the-art methods by 28.5% . |
Supervised Optimism Correction: Be Confident When LLMs Are Sure (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent advances in Large Language Models (LLMs) have demonstrated remarkable success across diverse tasks such as instruction following, code generation, and medical diagnosis. |
| Approach: | They propose a supervised fine-tuning-based auxiliary loss for Q-value estimations during supervised refinement. |
| Outcome: | The proposed method outperforms beam search on GSM8K, MATH, and GAOKAO on reasoning benchmarks. |
Offline Reinforcement Learning for LLM Multi-step Reasoning (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) are increasingly applied to complex tasks requiring multi-step reasoning. |
| Approach: | They propose an offline method for enhancing multi-step reasoning by optimizing the soft Bellman Equation by combining a policy model and a value function. |
| Outcome: | The proposed method surpasses existing methods on multi-step reasoning benchmarks and can be extended to multi-iteration frameworks when additional resources are available. |
Sampling-based Pseudo-Likelihood for Membership Inference Attacks (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) are trained on large-scale web data, which makes it difficult to grasp the contribution of each text. |
| Approach: | They propose a membership-inference attack method that uses only the input text to detect leaks. |
| Outcome: | The proposed method performs on par with existing likelihood-based methods even without likelihoods. |
AgentStore: Scalable Integration of Heterogeneous Agents As Specialized Generalist Computer Assistant (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing agents lack generalization and specialization capabilities for open-ended tasks . specialized generalists are often underdeveloped in real-world environments . |
| Approach: | They propose a platform to dynamically integrate heterogeneous agents for automating computer tasks . they propose specialized generalist agent MetaAgent with the AgentToken strategy . |
| Outcome: | The proposed platform expands capabilities of existing agents in generalization and specialization . it can be used to automate open-ended tasks in real-world environments . |
Boosting Vulnerability Detection of LLMs via Curriculum Preference Optimization with Synthetic Reasoning Data (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) are capable of detecting software vulnerabilities, but lack of reasoning data hinders their ability to capture underlying vulnerability patterns. |
| Approach: | They propose a framework that excels at mining vulnerability patterns through reasoning data synthesizing and vulnerability-specific preference optimization. |
| Outcome: | The proposed framework improves on SVEN and PrimeVul datasets and improves 12.24%-22.77% accuracy. |
GA-S3: Comprehensive Social Network Simulation with Group Agents (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing social network simulations focus on discrete events or system dynamics instead of elucidating underlying mechanisms or causal relationships. |
| Approach: | They propose a Social network simulation system that leverages newly designed Group Agents to make intelligent decisions regarding various online events. |
| Outcome: | The proposed system can make intelligent decisions regarding online events at a manageable cost. |
M-RangeDetector: Enhancing Generalization in Machine-Generated Text Detection through Multi-Range Attention Masks (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing supervised methods for text detection are overfitting within their training domains. |
| Approach: | They propose a method that integrates four distinct attention masking strategies into a Multi-Range Attention module to learn various writing strategies for machine-generated text detection. |
| Outcome: | The proposed method improves the generalization capability of existing detectors on three datasets. |
Does Your Voice Assistant Remember? Analyzing Conversational Context Recall and Utilization in Voice Interaction Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent advances in multi-turn voice interaction models have improved user-model communication, but whether open-source models share this ability remains unexplored. |
| Approach: | They propose to use ContextDialog to evaluate open-source interaction models' ability to recall past utterances to identify key limitations. |
| Outcome: | The proposed model retains and recalls past utterances better than closed-source models, but still struggles with questions about past . findings highlight key limitations in open-source model and suggest ways to improve memory retention and retrieval robustness. |
NeuronMerge: Merging Models via Functional Neuron Groups (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent studies show that task arithmetic improves performance by combining model parameters with output features. |
| Approach: | They propose a neuron-based task arithmetic merging method that improves model linearity . they group neurons by function and propose combining them with existing models . |
| Outcome: | The proposed method improves performance across tasks and scales. |
HellaSwag-Pro: A Large-Scale Bilingual Benchmark for Evaluating the Robustness of LLMs in Commonsense Reasoning (2025.findings-acl)
Copied to clipboard
Xiaoyuan Li, Moxin Li, Rui Men, Yichang Zhang, Keqin Bao, Wenjie Wang, Fuli Feng, Dayiheng Liu, Junyang Lin
| Challenge: | Existing studies show that large language models are robust in commonsense reasoning . however, some variations in questions can lead to incorrect responses . |
| Approach: | They propose a large-scale bilingual benchmark consisting of 11,200 cases . they conduct extensive experiments on 41 representative LLMs . |
| Outcome: | The proposed benchmark systematically evaluates the robustness of large language models in commonsense reasoning. |
Self-Steering Optimization: Autonomous Preference Optimization for Large Language Models (2025.findings-acl)
Copied to clipboard
Hao Xiang, Bowen Yu, Hongyu Lin, Keming Lu, Yaojie Lu, Xianpei Han, Ben He, Le Sun, Jingren Zhou, Junyang Lin
| Challenge: | Prior research focused on developing data generation methods, while insufficient attention has been paid to quality control mechanisms and often produces inaccurate and unhelpful data. |
| Approach: | They propose an algorithm that automatically generates high-quality preference data, eliminating manual annotation requirements. |
| Outcome: | The proposed algorithm outperforms baselines in human preference alignment and reward optimization. |
LIME: Less Is More for MLLM Evaluation (2025.findings-acl)
Copied to clipboard
King Zhu, Qianbo Zang, Shian Jia, Siwei Wu, Feiteng Fang, Yizhi Li, Shuyue Guo, Tianyu Zheng, Jiawei Guo, Bo Li, Haoning Wu, Xingwei Qu, Jian Yang, Ruibo Liu, Xiang Yue, Jiaheng Liu, Chenghua Lin, Hamid Alinejad-Rokny, Min Yang, Shiwen Ni, Wenhao Huang, Ge Zhang
| Challenge: | Existing MLLM benchmarks and unified evaluation frameworks cannot accurately and efficiently reflect the ability of MLMLs. |
| Approach: | They propose a semi-automated benchmark curated using a pipeline that filters out uninformative samples and eliminates answer leakage by focusing on tasks that require image-based understanding. |
| Outcome: | The proposed benchmark reduces the number of samples by 76% and evaluation time by 77% while it can more effectively distinguish different models’ abilities. |
Debate, Reflect, and Distill: Multi-Agent Feedback with Tree-Structured Preference Optimization for Efficient Language Model Enhancement (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for enhancing small models struggle to yield substantial and lasting performance gains. |
| Approach: | They propose a Debate and Reflect framework that orchestrates multi-turn debates between smaller models and stronger teacher models. |
| Outcome: | The proposed framework outperforms existing methods by a large margin in smaller models. |
CodeReviewQA: The Code Review Comprehension Assessment for Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | State-of-the-art large language models (LLMs) have demonstrated impressive code generation capabilities but struggle with real-world software engineering tasks such as revising source code to address code reviews. |
| Approach: | They propose a benchmark to evaluate large language models' ability to bridge both technical and conversational contexts by decomposing the generation task of code refinement into three essential reasoning steps. |
| Outcome: | The proposed benchmark exposes specific model weaknesses in code review comprehension disentangled from their generative automated code refinement results. |
Narrative Media Framing in Political Discourse (2025.findings-acl)
Copied to clipboard
| Challenge: | Narrative frames are a powerful way of conceptualizing and communicating complex ideas. |
| Approach: | They propose a framework which formalizes and operationalizes elements of narrative framing . they annotate news articles in the climate change domain and test their framework . |
| Outcome: | The proposed framework formalizes and operationalizes elements of narrative framing . it is applied to climate change crisis data, showing generalizability of the framework . |
MHALO: Evaluating MLLMs as Fine-grained Hallucination Detectors (2025.findings-acl)
Copied to clipboard
| Challenge: | Hallucination remains a critical challenge for multimodal large language models, undermining their reliability in real-world applications. |
| Approach: | They propose a benchmark specifically designed for evaluating MLLMs’ capability in performing token-level hallucination detection (FHD) . they use curated training data to train a specialized model that significantly outperforms existing models. |
| Outcome: | The proposed model outperforms existing models in the evaluation of 9 MLLMs and reaches an average F1IoU of 40.59%. |
Semantic Topology: a New Perspective for Communication Style Characterization (2025.findings-acl)
Copied to clipboard
| Challenge: | a new framework for discourse analysis uses Circuit Topology to quantify the semantic arrangement of sentences in textual structure. |
| Approach: | They propose a framework that leverages Circuit Topology to quantify the semantic arrangement of sentences in a text. |
| Outcome: | The proposed framework can quantify the semantic arrangement of sentences in a text. |
Decoding LLM Personality Measurement: Forced-Choice vs. Likert (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent research has focused on investigating the psychological characteristics of Large Language Models (LLMs), emphasizing the importance of comprehending their behavioral traits. |
| Approach: | They evaluated six Large Language Models: Llama-3.1-8B, GLM-4-9B, Claude-3.5-sonnet, and Deepseek-V3 and used the forced-choice test to assess their personality traits. |
| Outcome: | The forced-choice test is more reliable and more accurate than the likert scale and forced-CHOICE test results for LLMs' Big Five personality scores. |
MultiMSD: A Corpus for Multilingual Medical Text Simplification from Online Medical References (2025.findings-acl)
Copied to clipboard
| Challenge: | Medical texts contain technical terms, and non-experts often cannot use information effectively. |
| Approach: | They propose a method for training medical text simplification models to actively paraphrase medical terms. |
| Outcome: | The proposed method improves the performance of medical text simplification in nine languages. |
BadWindtunnel: Defending Backdoor in High-noise Simulated Training with Confidence Variance (2025.findings-acl)
Copied to clipboard
| Challenge: | Current backdoor attack defenders in NLP typically involve data reduction or model pruning, risking losing crucial information. |
| Approach: | They propose a backdoor defender that allows precise control over training conditions to model backdoor learning behavior without affecting the final model. |
| Outcome: | The proposed model reduces the backdoor learning behavior without affecting the final model. |
Multimodal Machine Translation with Text-Image In-depth Questioning (2025.findings-acl)
Copied to clipboard
| Challenge: | Multimodal machine translation (MMT) models focus on intermodal interactions, but focus on simple interactions between nouns and entities in image, overlooking global semantic alignment. |
| Approach: | They propose a Text-Image In-depth Questioning method to deepen interactions and optimize translations by utilizing visual data to capture global semantic alignment. |
| Outcome: | The proposed method achieves state-of-the-art results on five translation directions of Multi30K and AmbigCaps, with +2.35 BLEU on the challenging MSCOCO benchmark. |
ReKG-MCTS: Reinforcing LLM Reasoning on Knowledge Graphs via Training-Free Monte Carlo Tree Search (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing approaches to combining knowledge graphs with large language models face limitations in path exploration strategies or excessive computational overhead. |
| Approach: | They propose a training-free framework that synergizes Monte Carlo Tree Search with LLM capabilities to enable dynamic reasoning over KGs. |
| Outcome: | The proposed framework outperforms existing training-free methods and achieves competitive performance compared to fine-tuned baselines. |
HTML: Hierarchical Topology Multi-task Learning for Semantic Parsing in Knowledge Base Question Answering (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing approaches struggle with mapping questions to precise logical forms . Existing frameworks struggle with complex mapping of questions to logical form . |
| Approach: | They propose a framework that leverages a hierarchical multi-task learning paradigm to enhance the performance of logical form generation. |
| Outcome: | The proposed framework outperforms supervised fine-tuning methods and training-free ones on large language models. |
StructFlowBench: A Structured Flow Benchmark for Multi-turn Instruction Following (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing evaluation benchmarks focus on fine-grained constraint satisfaction and domain-specific capability assessment, yet overlook the crucial structural dependencies between dialogue turns that distinguish multi-turn from single-turn interactions. |
| Approach: | They propose a multi-turn instruction following benchmark with structural flow modeling that defines an innovative structural flow framework with six fundamental inter-turn relationships. |
| Outcome: | The proposed model is based on a framework with six fundamental inter-turn relationships and is able to analyze and generate specific dialogue flows tailored to specific scenarios. |
CMIE: Combining MLLM Insights with External Evidence for Explainable Out-of-Context Misinformation Detection (2025.findings-acl)
Copied to clipboard
| Challenge: | Multimodal large language models have demonstrated impressive capabilities in visual reasoning and text generation. |
| Approach: | They propose a multimodal large language model that captures deeper relationships between images and text . they propose CMIE, which uses a Coexistence Relationship Generation strategy and an AS mechanism to detect misinformation. |
| Outcome: | The proposed framework outperforms existing methods in detecting out-of-context misinformation. |
EtiCor++: Towards Understanding Etiquettical Bias in LLMs (2025.findings-acl)
Copied to clipboard
| Challenge: | Etiquettes are region-specific and are an essential part of the culture of a region. |
| Approach: | They introduce EtiCor++, a corpus of etiquettes worldwide, to evaluate LLMs for their knowledge about etiques across regions. |
| Outcome: | The proposed corpus of etiquettes shows that LLMs are biased towards certain regions. |
FinRipple: Aligning Large Language Models with Financial Market for Event Ripple Effect Awareness (2025.findings-acl)
Copied to clipboard
| Challenge: | Financial markets exhibit complex dynamics where localized events trigger ripple effects across entities. |
| Approach: | They propose a framework that empowers large language models to analyze ripple effects . they use financial theory-guided large-scale reinforcement learning to align LLMs with the market . |
| Outcome: | The proposed framework allows LLMs to analyze ripple effects through financial theory-guided large-scale reinforcement learning. |
Beyond Decoder-only: Large Language Models Can be Good Encoders for Machine Translation (2025.findings-acl)
Copied to clipboard
Yingfeng Luo, Tong Zheng, Yongyu Mu, Bei Li, Qinghong Zhang, Yongqi Gao, Ziqiang Xu, Peinan Feng, Xiaoqian Liu, Tong Xiao, JingBo Zhu
| Challenge: | Recent advances in machine translation have focused on a single pre-trained decoder . encoder-decoder architectures have received relatively little attention in NMT . |
| Approach: | They propose a method that leverages LLMs as MT encoders and pairs them with lightweight decoders to develop universal translation models. |
| Outcome: | The proposed method matches or surpasses baselines in terms of translation quality but achieves 75% reduction in memory footprint of the KV cache. |
EC-RAFT: Automated Generation of Clinical Trial Eligibility Criteria through Retrieval-Augmented Fine-Tuning (2025.findings-acl)
Copied to clipboard
| Challenge: | Eligibility criteria (EC) are critical components of clinical trial design, specifying parameters for participant inclusion and exclusion. |
| Approach: | They propose a method that utilizes Retrieval-Augmented Fine-Tuning to generate structured and cohesive EC directly from clinical trial titles and descriptions. |
| Outcome: | The proposed method outperforms Llama-3.1-8B-Instruct and Llm-as-a-Judge models in BERTScore and EC score. |
Pitfalls of Scale: Investigating the Inverse Task of Redefinition in Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have shown remarkable results in several linguistic, reasoning and knowledge retrieval tasks. |
| Approach: | They propose to scale Large Language Models (LLMs) to scale up to reveal potential reasoning gaps as LLMs scale up. |
| Outcome: | The proposed redefinition task shows that model performance degrades with scale, and false confidence rises. |
Implicit Reasoning in Transformers is Reasoning through Shortcuts (2025.findings-acl)
Copied to clipboard
| Challenge: | Language models can perform step-by-step reasoning and achieve high accuracy in both in-domain and out-of-domain tests via implicit reasoning. |
| Approach: | They train GPT-2 from scratch on a curated multi-step mathematical reasoning dataset and conduct analytical experiments to investigate how language models perform implicit reasoning in multi- step tasks. |
| Outcome: | The proposed model performs better on multi-step tasks than the explicit reasoning model. |
Learning to Align Multi-Faceted Evaluation: A Unified and Robust Framework (2025.findings-acl)
Copied to clipboard
Kaishuai Xu, Tiezheng Yu, Yi Cheng, Wenjun Hou, Liangyou Li, Xin Jiang, Lifeng Shang, Qun Liu, Wenjie Li
| Challenge: | Existing methods for fine-tuning open-source LLMs are limited to text-based analysis under predefined general criteria. |
| Approach: | They propose a framework that fine-tunes LLMs to replicate the evaluation explanations and judgments of proprietary models. |
| Outcome: | The proposed evaluation framework outperforms existing fine-tuned evaluation methods in effectiveness and robustness. |
CortexDebate: Debating Sparsely and Equally for Multi-Agent Debate (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods to improve the reasoning performance of LLMs suffer from two major shortcomings: too lengthy input contexts and overconfidence dilemma. |
| Approach: | They propose a method to debating among LLM agents using a sparse debator graph . they use a module called McKinsey-based Debate Matter to optimize the debators . |
| Outcome: | The proposed method has been well demonstrated across eight datasets from four task types. |
PAP2PAT: Benchmarking Outline-Guided Long-Text Patent Generation with Patent-Paper Pairs (2025.findings-acl)
Copied to clipboard
| Challenge: | In patents, the description constitutes more than 90% of the document on average, yet its automatic generation remains understudied. |
| Approach: | They propose a method to generate patent documents using a research paper as an invention specification. |
| Outcome: | The proposed model can generate 1.8k patent-paper pairs describing the same inventions, but it's difficult to provide the level of detail required. |
Debt Collection Negotiations with Large Language Models: An Evaluation System and Optimizing Decision Making with Multi-Agent (2025.findings-acl)
Copied to clipboard
| Challenge: | Debt collection negotiations (DCN) are vital for managing non-performing loans (NPLs) prior systems lacking dynamic negotiation and real-time decision-making capabilities. |
| Approach: | They propose a framework for debt negotiation that incorporates planning and judging modules to improve decision rationality. |
| Outcome: | The proposed framework improves decision rationality and integrates planning and judging modules to improve decision rationalness. |
Focused-DPO: Enhancing Code Generation Through Focused Preference Optimization on Error-Prone Points (2025.findings-acl)
Copied to clipboard
| Challenge: | Current code generation models produce errors concentrated at specific error-prone points, affecting accuracy of code. |
| Approach: | They propose a framework that focuses preference optimization on error-prone areas . focused-DPO improves the accuracy and reliability of code generation by reducing common errors . |
| Outcome: | The proposed framework improves code generation by focusing on error-prone areas. |
Supervised and Unsupervised Probing of Shortcut Learning: Case Study on the Emergence and Evolution of Syntactic Heuristics in BERT (2025.findings-acl)
Copied to clipboard
| Challenge: | Contemporary language models (LMs) rely on shortcut learning, using superficial cues that are spuriously correlated with labels. |
| Approach: | They propose to use syntactic heuristics to learn shortcuts in BERT when performing a task in Natural Language Understanding to investigate where these shortcuts emerge, how they evolve and how they impact the latent knowledge of the LM. |
| Outcome: | The proposed model rely on syntactic heuristics when performing a task in Natural Language Understanding. |
GIMMICK: Globally Inclusive Multimodal Multitask Cultural Knowledge Benchmarking (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing studies on Large Vision-Language Models (LVLMs) focus on a narrow range of cultures, focus on only a small number of cultural aspects or evaluate a limited selection of models on ONE task only. |
| Approach: | They propose a multimodal benchmark to assess a broad spectrum of cultural knowledge across 144 countries representing six global macro-regions. |
| Outcome: | The proposed benchmark examines cultural knowledge across 144 countries across six global macro-regions. |
R-VLM: Region-Aware Vision Language Model for Precise GUI Grounding (2025.findings-acl)
Copied to clipboard
Joonhyung Park, Peng Tang, Sagnik Das, Srikar Appalaraju, Kunwar Yashraj Singh, R. Manmatha, Shabnam Ghadar
| Challenge: | Existing vision-only GUI agents ground elements from large and cluttered screenshots, requiring them to process substantial irrelevant information that compromises their accuracy. |
| Approach: | They propose a visual agent model for GUI automation that leverages zoomed-in region proposals for precise element localization. |
| Outcome: | The proposed approach improves state-of-the-art grounding accuracy by 13% across diverse GUI platforms on the GUI grounding benchmarks ScreenSpot and AgentStudio. |
Perspective Transition of Large Language Models for Solving Subjective Tasks (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) have revolutionized the field of natural language processing . performance of LLMs on subjective tasks is limited, authors say . |
| Approach: | They propose a method that allows LLMs to select between direct, role, and third-person perspectives for best way to solve corresponding subjective problem. |
| Outcome: | The proposed method outperforms widely used single fixed perspective based methods on 12 subjective tasks. |
TripTailor: A Real-World Benchmark for Personalized Travel Planning (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing evaluation metrics for travel planning rely on unrealistic simulated data . fewer than 10% of the itineraries generated by the latest state-of-the-art LLMs achieve human-level performance. |
| Approach: | They propose a benchmark for personalized travel planning in real-world scenarios . they identify several critical challenges in travel planning including feasibility and rationality . |
| Outcome: | The proposed benchmarks show that fewer than 10% of the itineraries generated by the latest state-of-the-art LLMs achieve human-level performance. |
Random Splitting Negatively Impacts NER Evaluation: Quantifying and Eliminating the Overestimation of NER Performance (2025.findings-acl)
Copied to clipboard
| Challenge: | Named entity recognition models are evaluated on their ability to identify entity mentions in text. |
| Approach: | They propose a method to reduce the amount of entity contamination in NER datasets by a minimum cut algorithm. |
| Outcome: | The proposed method minimizes train-test entity leakage while ensuring near zero entity contamination. |
Structure-adaptive Adversarial Contrastive Learning for Multi-Domain Fake News Detection (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing models for fake news detection capture domain-shared semantic features but fail to generalize well due to poor adaptability. |
| Approach: | They propose a framework to enable structure knowledge transfer between multiple domains . they compare content-only and propagation-rich data to preserve structural patterns . |
| Outcome: | The proposed framework can learn semantic and structural features across domains. |
BiasGuard: A Reasoning-Enhanced Bias Detection Tool for Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for identifying bias in LLM-generated content face limitations . existing methods rely on pattern-based learning, which makes it challenging to understand intentions . |
| Approach: | They propose a bias detection tool that explicitly analyzes inputs and reasons through fairness specifications to provide accurate judgments. |
| Outcome: | The proposed tool outperforms existing tools and improves accuracy and reduces over-fairness misjudgments. |
Qorǵau: Evaluating Safety in Kazakh-Russian Bilingual Contexts (2025.findings-acl)
Copied to clipboard
Maiya Goloburda, Nurkhan Laiyk, Diana Turmakhan, Yuxia Wang, Mukhammed Togmanov, Jonibek Mansurov, Askhat Sametov, Nurdaulet Mukhituly, Minghan Wang, Daniil Orel, Zain Muhammad Mujahid, Fajri Koto, Timothy Baldwin, Preslav Nakov
| Challenge: | Large language models (LLMs) have the potential to generate harmful content, posing risks to users. |
| Approach: | They propose a dataset specifically designed for safety evaluation in Kazakh and Russian . they use a bilingual context in Kazakhstan where both Kazakh (a low-resource language) and Russian (a high-resourced language) |
| Outcome: | The proposed dataset is designed for safety evaluation in Kazakh and Russian . it shows that both multilingual and language-specific LLMs perform better than others . |
MMXU: A Multi-Modal and Multi-X-ray Understanding Dataset for Disease Progression (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing datasets and models fail to consider critical aspects of medical diagnostics, authors argue . MMXU enables multi-image questions incorporating both current and historical patient data. |
| Approach: | They propose a dataset for MedVQA that focuses on identifying changes in specific regions between two patient visits. |
| Outcome: | The proposed dataset improves diagnostic accuracy by 20% by integrating historical data. |
Tree-of-Code: A Self-Growing Tree Framework for End-to-End Code Generation and Execution in Complex Tasks (2025.findings-acl)
Copied to clipboard
| Challenge: | Effectively and efficiently handling complex realworld problems has become a key focus across industry and academia. |
| Approach: | They propose a tree-of-code framework that generates nodes through self-supervision and combines prompt and model exploration in a GT-free setting. |
| Outcome: | Experiments on two datasets with ten popular zero-shot LLMs show that Tree-of-Code boosts accuracy by nearly 20% over CodeAct with fewer than 1/4 turns. |
Akan Cinematic Emotions (ACE): A Multimodal Multi-party Dataset for Emotion Recognition in Movie Dialogues (2025.findings-acl)
Copied to clipboard
David Sasu, Zehui Wu, Ziwei Gong, Run Chen, Pengyuan Shi, Lin Ai, Julia Hirschberg, Natalie Schluter
| Challenge: | Akan Cinematic Emotions (AkaCE) is the first multimodal emotion dialogue dataset for an African language . it contains 385 emotion-labeled dialogues and 6162 utterances across audio, visual, and textual modalities, along with word-level prosodic prominence annotations. |
| Approach: | They propose to use AkaCE to analyze African cinematic emotions using word-level prosodic prominence annotations. |
| Outcome: | The Akan Cinematic Emotions (AkaCE) dataset addresses the significant lack of resources for low-resource languages in emotion recognition research. |
A Cognitive Writing Perspective for Constrained Long-Form Text Generation (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models struggle to generate high-quality long-form text in a single pass . a new framework that trains LLMs to write human-like writing capabilities is needed . |
| Approach: | They propose a framework that equips large language models with human-like cognitive writing capabilities . they use a planning agent and multiple Generation Agents to generate long-form text in parallel . |
| Outcome: | CogWriter surpasses GPT-4o by 22% in complex instruction completion accuracy . the framework can generate coherent text in a single pass with fluency that rivals human writers . |
Migician: Revealing the Magic of Free-Form Multi-Image Grounding in Multimodal Large Language Models (2025.findings-acl)
Copied to clipboard
You Li, Heyu Huang, Chi Chen, Kaiyu Huang, Chao Huang, Zonghao Guo, Zhiyuan Liu, Jinan Xu, Yuhua Li, Ruixuan Li, Maosong Sun
| Challenge: | Existing MLLMs still struggle to achieve precise grounding in multi-image scenarios. |
| Approach: | They propose a Chain-of-Thought framework that integrates single-image grounding with multi-image comprehension to address this challenge. |
| Outcome: | The proposed model outperforms existing models in multi-image grounding tasks by 24.94% and surpasses larger 70B models. |
SIKeD: Self-guided Iterative Knowledge Distillation for Mathematical Reasoning (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) can generate intermediate reasoning process for multistep reasoning tasks. |
| Approach: | They propose a distillation method that teaches the model to approach a task using different strategies and the model uses its self-generated on-policy outputs to choose the most suitable strategy. |
| Outcome: | The proposed method significantly outperforms distillation techniques on large models of different sizes. |
Chain of Attack: Hide Your Intention through Multi-Turn Interrogation (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing jailbreak attacks focus on single-turn dialogue scenarios, leaving vulnerabilities in multi-turn contexts inadequately explored. |
| Approach: | They propose an optimal interrogation principle to conceal the jailbreak intent and introduce a multi-turn attack chain generation strategy called CoA. |
| Outcome: | The proposed method shows that black-box LLMs exhibit insufficient resistance under multi-turn interrogation, with more advantages (ASR, 83% vs 64%) |
MIG: Automatic Data Selection for Instruction Tuning by Maximizing Information Gain in Semantic Space (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for instruction-tuning datasets prioritize instance quality and use heuristic rules to maintain diversity. |
| Approach: | They propose a method that quantifies diversity based on the distribution of information within a label graph. |
| Outcome: | The proposed method outperforms state-of-the-art methods on 5% Tulu3 datasets and base models. |
Enhancing Automatic Term Extraction with Large Language Models via Syntactic Retrieval (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) have improved IE, but their potential for ATE has not been explored. |
| Approach: | They propose a retrieval-based prompting strategy that selects demonstrations according to syntactic rather than semantic similarity in a few-shot setting. |
| Outcome: | The proposed method improves performance on three specialized ATE benchmarks. |
Explainable Depression Detection in Clinical Interviews with Personalized Retrieval-Augmented Generation (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing systems rely on black-box neural networks, which lack interpretability, which is crucial in mental health contexts. |
| Approach: | They propose a Retrieval-augmented generation framework for Explainable depression detection that retrieves evidence from clinical interview transcripts, providing explanations for predictions. |
| Outcome: | The proposed framework retrieves evidence from clinical interview transcripts, providing explanations for predictions. |
EMPEC: A Comprehensive Benchmark for Evaluating Large Language Models Across Diverse Healthcare Professions (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent advances in Large Language Models (LLMs) show their potential in accurately answering biomedical questions, yet current healthcare benchmarks primarily assess knowledge mastered by medical doctors, neglecting other essential professions. |
| Approach: | They evaluated 17 LLMs including proprietary and open-source models and found they struggled with specialized fields and alternative medicine. |
| Outcome: | The examinations for medical PErsonnel in Chinese (EMPEC) features 157,803 exam questions across 124 subjects and 20 healthcare professions. |
Beyond Numeric Rewards: In-Context Dueling Bandits with LLM Agents (2025.findings-acl)
Copied to clipboard
| Challenge: | In-Context Reinforcement Learning (ICRL) is a frontier paradigm for RL problems . authors find that LLMs can generalize cross-domain to perform ICRL on a stateless preference-based RL problem. |
| Approach: | They propose an agentic-flow framework that integrates off-the-shelf DB algorithm support with LLM agents through fine-grained adaptive interplay. |
| Outcome: | The proposed framework can generalize cross-domain to perform ICRL on a stateless preference-based RL problem. |
“Well, Keep Thinking”: Enhancing LLM Reasoning with Adaptive Injection Decoding (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) exhibit strong reasoning abilities, often attributed to few-shot or zero-shot Chain-of-Thought (CoT) prompting. |
| Approach: | They propose a decoding strategy that nudges LLMs to continue reasoning, thereby preventing immature reasoning processes. |
| Outcome: | The proposed method significantly improves LLM reasoning capabilities on diverse reasoning benchmarks. |
SpeechT-RAG: Reliable Depression Detection in LLMs with Retrieval-Augmented Generation Using Speech Timing Information (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have been extensively utilized for health-related tasks, yet their performance in depression detection remains limited when relying solely on text input. |
| Approach: | They propose a system that leverages speech timing features for depression detection and reliable confidence estimation. |
| Outcome: | The proposed system outperforms text-based RAG systems in depression detection and confidence estimation. |
Fine-grained Knowledge Enhancement for Retrieval-Augmented Generation (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing studies rely on semantic similarity to retrieve knowledge but ignore fine-grained information within documents. |
| Approach: | They propose a fine-grained knowledge enhancement method to fill knowledge gaps with retrieved external information by a Chain-of-Thought prompting procedure and a decoding enhancement strategy to constrain the document-based decoding process. |
| Outcome: | The proposed method can be applied in a plug-and-play manner to enhance its performance with no additional modules or training process. |
Bayesian Optimization for Controlled Image Editing via LLMs (2025.findings-acl)
Copied to clipboard
Chengkun Cai, Haoliang Liu, Xu Zhao, Zhongyu Jiang, Tianfang Zhang, Zongkai Wu, John Lee, Jenq-Neng Hwang, Lei Li
| Challenge: | achieving precise control over generated content and maintaining semantic consistency remain significant limitations, particularly concerning grounding techniques and the necessity for model fine-tuning. |
| Approach: | They propose an off-the-shelf approach that integrates Large Language Models with Bayesian Optimization to facilitate precise and user-friendly image editing. |
| Outcome: | The proposed approach outperforms existing methods in editing accuracy and semantic preservation, as validated using different LLMs including Claude3 and GPT-4. |
SPOT: Zero-Shot Semantic Parsing Over Property Graphs (2025.findings-acl)
Copied to clipboard
| Challenge: | Knowledge Graphs (KGs) are becoming increasingly popular as a means of storing structured data. |
| Approach: | They propose a method to generate training data for semantic parsing over Property Graphs without human annotations by matching tree patterns to the KG and paraphrasing the query program with an LLM. |
| Outcome: | The proposed method generates training data for parsing over Property Graphs without human annotations on two property graph benchmarks utilizing the Cypher query language. |
Reasoning Circuits in Language Models: A Mechanistic Interpretation of Syllogistic Inference (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent studies on reasoning in language models have sparked a debate on whether they can learn systematic inferential principles or merely exploit superficial patterns in the training data. |
| Approach: | They propose a method for circuit discovery aimed at interpreting syllogistic inference . they uncover a circuit involving middle-term suppression that elucidates how LMs transfer information to derive valid conclusions from premises. |
| Outcome: | The proposed method elucidates how LMs transfer information to derive valid conclusions from premises. |
Multi-Hop Question Generation via Dual-Perspective Keyword Guidance (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing work fails to fully utilize the guiding potential of keywords and neglect to differentiate the distinct roles of question-specific and document-specific keywords. |
| Approach: | They propose a dual-perspective keyword-guided framework that integrates question and document keywords into the multi-hop question generation process. |
| Outcome: | The proposed framework integrates question and document keywords into the multi-hop question generation process. |
LoRMA: Low-Rank Multiplicative Adaptation for LLMs (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models have shown impressive generalization capabilities, but can be expensive to fine-tune due to high computational costs. |
| Approach: | They propose a low-rank multiplicative Adaptation technique that shifts the paradigm of additive updates to a richer space of matrix multiplicative transformations. |
| Outcome: | The proposed approach overcomes computational complexity and rank bottlenecks in terms of matrix multiplication metrics. |
DI-BENCH: Benchmarking Large Language Models on Dependency Inference with Testable Repositories at Scale (2025.findings-acl)
Copied to clipboard
Linghao Zhang, Junhao Wang, Shilin He, Chaoyun Zhang, Yu Kang, Bowen Li, Jiaheng Wen, Chengxing Xie, Maoquan Wang, Yufan Huang, Elsie Nallipogu, Qingwei Lin, Yingnong Dang, Saravan Rajmohan, Dongmei Zhang, Qi Zhang
| Challenge: | Existing studies highlight that dependency-related issues cause over 40% of observed runtime errors on the generated repository. |
| Approach: | They propose a large-scale benchmark and evaluation framework specifically designed to assess LLMs’ capability on dependency inference. |
| Outcome: | The proposed model achieves only a 48% execution pass rate on Python, indicating room for improvement. |
Weak-to-Strong Honesty Alignment via Learning-to-Rank Supervision (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing approaches to enhance honesty with prompt engineering and fine-tuning are limited by annotated data. |
| Approach: | They propose a framework that enhances honesty through weak-to-strong generalization by training weak LLMs under weak supervision to improve their honesty. |
| Outcome: | The proposed framework improves honesty in large models even with limited label data. |
MultiHoax: A Dataset of Multi-hop False-premise questions (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing benchmarks focus on single-hop FPQs, but real-world reasoning often requires multi-hop inference . state-of-the-art LLMs struggle to detect false premises across different countries, knowledge categories, and multi-step reasoning types. |
| Approach: | They propose a benchmark to evaluate Large Language Models' ability to handle false premises in complex, multi-step reasoning tasks. |
| Outcome: | The proposed tests show that state-of-the-art LLMs struggle to detect false premises across different countries, knowledge categories, and multi-hop reasoning types. |
Learning to Play Like Humans: A Framework for LLM Adaptation in Interactive Fiction Games (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing approaches prioritize task-specific performance metrics over human-like comprehension of narrative context and gameplay logic. |
| Approach: | They propose a framework that guides Large Language Models to learn and play IF games systematically. |
| Outcome: | The proposed framework aligns LLMs-based agents’ behavior with narrative intent and commonsense constraints to deliver more interpretable, human-like performance. |
STATE ToxiCN: A Benchmark for Span-level Target-Aware Toxicity Extraction in Chinese Hate Speech Detection (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing studies on Chinese hate speech detection lack span-level fine-grained annotations. |
| Approach: | They construct a Span-level target-aware Toxicity Extraction dataset and evaluate existing models for Chinese hateful slang. |
| Outcome: | The proposed dataset is the first span-level Chinese hate speech dataset and evaluates the ability of existing models to understand hate semantics. |
RelEdit: Evaluating Conceptual Knowledge Editing in Language Models via Relational Reasoning (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing knowledge editing methods struggle to reason about related conceptual knowledge effectively, despite a lack of model-level relational reasoning. |
| Approach: | They propose a benchmark to assess concept-level and instance-level relational reasoning abilities of edited models. |
| Outcome: | The proposed model obtains the best scores on the memory-based in-context editing baseline, MICE, suggesting a promising direction for model editing. |
Unlocking Speech Instruction Data Potential with Query Rewriting (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing LLMs lack datasets and biased training tasks to follow speech instructions. |
| Approach: | They propose a query rewriting framework that uses multiple agents to annotate and validate the synthesized speech. |
| Outcome: | The proposed framework can transform text instructions into distributions more suitable for TTS models for speech synthesis without human annotation. |
From Evasion to Concealment: Stealthy Knowledge Unlearning for LLMs (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing approaches to unlearning often treat nonsensical responses or template-based refusals as the unlearning target, making the process even more vulnerable to attacks and jailbreaks. |
| Approach: | They propose a method that uses inverted facts to remove the need for auxiliary models or retaining data while avoiding leakage. |
| Outcome: | Evaluated on the ToFU Knowledge Unlearning dataset using Llama2-7B-Chat and Phi-1.5, MEOW outperforms baselines in forgetting quality while preserving model utility. |
Context-DPO: Aligning Language Models for Context-Faithfulness (2025.findings-acl)
Copied to clipboard
Baolong Bi, Shaohan Huang, Yiwei Wang, Tianchi Yang, Zihan Zhang, Haizhen Huang, Lingrui Mei, Junfeng Fang, Zehao Li, Furu Wei, Weiwei Deng, Feng Sun, Qi Zhang, Shenghua Liu
| Challenge: | Context-DPO is the first alignment method specifically designed to enhance contextfaithfulness for large language models. |
| Approach: | They propose a benchmark that simulates Retrieval-Augmented Generation scenarios with knowledge conflicts to evaluate context-faithfulness. |
| Outcome: | The proposed method improves LLMs' context-faithfulness by 35% to 280% over open-source models. |
Reasoning Does Not Necessarily Improve Role-Playing Ability (2025.findings-acl)
Copied to clipboard
| Challenge: | a study compares zero-shot role-playing, reasoning-optimized LLMs, and reasoning-based LLM. |
| Approach: | They propose to use reasoning-optimized LLMs to improve role-playing performance . they propose to develop a chain-of-thought-based learning system that can be used to improve LLM performance if reasoning is used . |
| Outcome: | The proposed research compares zero-shot role-playing, role-playering with Chain-of-Thought, and reasoning-optimized LLMs. |
TableLLM: Enabling Tabular Data Manipulation by LLMs in Real Office Usage Scenarios (2025.findings-acl)
Copied to clipboard
Xiaokang Zhang, Sijia Luo, Bohan Zhang, Zeyao Ma, Jing Zhang, Yang Li, Guanlin Li, Zijun Yao, Kangli Xu, Jinchang Zhou, Daniel Zhang-Li, Jifan Yu, Shu Zhao, Juanzi Li, Jie Tang
| Challenge: | TableLLM is a robust large language model capable of handling tabular data manipulation tasks. |
| Approach: | They propose a distant supervision method for training which includes a reasoning process extension strategy and a cross-way validation strategy. |
| Outcome: | The proposed model has 8 billion parameters and is capable of handling tabular data tasks. |
A Survey of LLM-based Agents in Medicine: How far are we from Baymax? (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) are transforming healthcare through their ability to understand and assist with medical tasks. |
| Approach: | They analyze system profiles, clinical planning, medical reasoning frameworks, and external capacity enhancement. |
| Outcome: | The findings highlight the future directions in medical reasoning, physical system integration, and training simulations. |
Context-Robust Knowledge Editing for Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing knowledge editing methods assess success by considering only edited knowledge without preceding contexts. |
| Approach: | They propose a method to strengthen context robustness by minimizing context-sensitive variance in hidden states of the model. |
| Outcome: | The proposed method improves the success rate in situations where a preceding context is present and preserves the overall capabilities of the model. |
Multi-Agent Collaboration via Cross-Team Orchestration (2025.findings-acl)
Copied to clipboard
Zhuoyun Du, Chen Qian, Wei Liu, Zihao Xie, YiFei Wang, Rennai Qiu, Yufan Dang, Weize Chen, Cheng Yang, Ye Tian, Xuantang Xiong, Lei Han
| Challenge: | Large Language Models (LLMs) have significantly impacted various domains, especially through organized LLM-driven autonomous agents. |
| Approach: | They propose a framework that enables orchestrated teams to jointly propose various task-oriented solutions and interact with their insights in a self-independence while cross-team collaboration environment for superior solutions generation. |
| Outcome: | Experiments show that the framework can generate better software quality compared to state-of-the-art frameworks. |
Semantic Evaluation of Multilingual Data-to-Text Generation via NLI Fine-Tuning: Precision, Recall and F1 scores (2025.findings-acl)
Copied to clipboard
| Challenge: | KG-to-Text models are prone to errors like Additions and Omissions, and few languages are taken into account since both train and test data are not readily available. |
| Approach: | They propose a multilingual evaluation framework that is reference-less . it allows estimating how much a KG-to-Text Model under- (omission) or over- (addition) generates. |
| Outcome: | The proposed evaluation framework outperforms prior reference-less metrics in correlation with human judgments and provides scores for precision and recall. |
Optimized Text Embedding Models and Benchmarks for Amharic Passage Retrieval (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for multilingual and cross-lingual retrieval are lacking in low-resource, morphologically rich languages such as Amharic. |
| Approach: | They propose to train Amharic-specific dense retrieval models based on pre-trained Amharican BERT and RoBERTa backbones. |
| Outcome: | The proposed model achieves 17.6% improvement in MRR@10 and 9.86% gain in Recall@10 over the strongest multilingual baseline, Arctic Embed 2.0. |
Enhancing Transformation from Natural Language to Signal Temporal Logic Using LLMs with Diverse External Knowledge (2025.findings-acl)
Copied to clipboard
| Challenge: | Temporal Logic (STL) is a formal specification tool for cyber-physical systems . but it is difficult to transform ambiguous and complex data into STL, a paper argues . |
| Approach: | They propose a NL-STL dataset with 16,000 samples enriched with diverse patterns . they propose KGST framework to transform natural language into STL using a generate-then-refine process . |
| Outcome: | The proposed dataset outperforms baseline models in diversity and accuracy . the proposed dataset contains 16,000 samples enriched with diverse patterns . |
DAGS: A Dependency-Based Dual-Attention and Global Semantic Improvement Framework for Metaphor Recognition (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for metaphor recognition ignore interference caused by literal annotations . et al., 2018: Metaphor recognition plays an important role in cognition and communication . |
| Approach: | They propose a dependency-based Dual-Attention and Global Semantic Improvement framework to improve metaphor recognition. |
| Outcome: | The proposed framework can extract features from multiple information sources while improving on mainstream metaphor datasets. |
ESF: Efficient Sensitive Fingerprinting for Black-Box Tamper Detection of Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) are increasingly utilized in diverse applications, including code generation, legal document analysis, medical diagnosis, and decision-making. |
| Approach: | They propose a fingerprinting method tailored for black-box tamper detection of large language models. |
| Outcome: | The proposed method detects tampering with a 99.2% detection rate using 5 fingerprint samples across state-of-the-art LLMs. |
The Lessons of Developing Process Reward Models in Mathematical Reasoning (2025.findings-acl)
Copied to clipboard
Zhenru Zhang, Chujie Zheng, Yangzhen Wu, Beichen Zhang, Runji Lin, Bowen Yu, Dayiheng Liu, Jingren Zhou, Junyang Lin
| Challenge: | a recent study shows that process reward models can make mistakes, leading to wrong conclusions. |
| Approach: | They propose a consensus filtering mechanism that integrates MC estimation with LLM-as-a-judge to improve model performance and data efficiency. |
| Outcome: | The proposed model outperforms existing open-source alternatives and provides practical guidelines for future research. |
MinosEval: Distinguishing Factoid and Non-Factoid for Tailored Open-Ended QA Evaluation with LLMs (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing evaluation methods overlook the distinction between factoid and non-factoidic questions. |
| Approach: | They propose a method that distinguishes open-ended questions and ranks candidate answers . they propose QA requires longer answer statements and nuanced reasoning processes . |
| Outcome: | The proposed method better aligns with human annotations and offers more interpretable results. |
Towards Conditioning Clinical Text Generation for User Control (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models exhibit hallucinations and factual inconsistencies necessitating human oversight. |
| Approach: | They propose to use Large Language Models as human proxies to condition LLMs for clinician control without increasing cognitive workload. |
| Outcome: | The proposed approach yields 9% relative improvement without augmented training and up to 34% with dataset augmentation. |
CoDet-M4: Detecting Machine-Generated Code in Multi-Lingual, Multi-Generator and Multi-Domain Settings (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have revolutionized code generation but have significant consequences for programming skills, ethics, and assessment integrity. |
| Approach: | They propose a framework capable of distinguishing between human-written and LLM-generated program code across multiple programming languages, code generators, and domains. |
| Outcome: | The proposed framework distinguishes between human-written and LLM-generated program code across multiple programming languages, code generators, and domains. |
Q-Mamba: Towards more efficient Mamba models via post-training quantization (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing studies show that Mamba architectures have room for further optimization in linear projections and state caches. |
| Approach: | They propose a decoupled scale quantization scheme to mitigate outliers in states and channels by applying separate quantization scales. |
| Outcome: | The proposed method reduces memory consumption by 50% across various quantization settings, model sizes, and generation and zero-shot tasks. |
P²Net: Parallel Pointer-based Network for Key Information Extraction with Complex Layouts (2025.findings-acl)
Copied to clipboard
Kaiwen Wei, Jie Yao, Jiang Zhong, Yangyang Kang, Jingyuan Zhang, Changlong Sun, Xin Zhang, Fengmao Lv, Li Jin
| Challenge: | Existing methods for key information extraction are based on a limited set of entity categories and fixed layouts. |
| Approach: | They propose a large-scale, human-annotated dataset for key information extraction . it is based on a human-annotated layout and 1,162 entity categories . they propose 'parallel pointer-based network' that leverages implicit relationships . |
| Outcome: | Experiments on widely-used datasets show that the proposed model outperforms state-of-the-art methods while maintaining fast inference speeds. |
Refining Sentence Embedding Model through Ranking Sentences Generation with Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Sentence embedding is essential for many NLP tasks, but reliance on manual labels limits scalability. |
| Approach: | They propose a method for controlling the generation direction of large language models in the latent space by integrating ranking information and semantic information. |
| Outcome: | The proposed method achieves new SOTA performance with a modest cost in ranking sentence synthesis. |
RQT: Hierarchical Residual Quantization for Multi-Model Compression (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for decomposing fine-tuned LLMs are sensitive to the magnitude of delta values. |
| Approach: | They propose a hierarchical quantization framework that shares low-bit integer weights across similar models. |
| Outcome: | The proposed framework achieves an average accuracy degradation of approximately 3% on fine-tuned models across mathematics, coding, chatbot, and Chinese LLMs. |
taz2024full: Analysing German Newspapers for Gender Bias and Discrimination across Decades (2025.findings-acl)
Copied to clipboard
| Challenge: | a large corpus of German newspaper articles is available for free in other languages, such as English. |
| Approach: | They propose to use taz2024full to analyse gender representation across four decades of reporting. |
| Outcome: | The proposed corpus supports a wide range of applications from diachronic language analysis to critical media studies. |
LCFO: Long Context and Long Form Output Dataset and Benchmarking (2025.findings-acl)
Copied to clipboard
Marta R. Costa-jussà, Pierre Andrews, Mariano Coria Meglioli, Joy Chen, Joe Chuang, David Dale, Christophe Ropers, Alexandre Mourachko, Eduardo Sánchez, Holger Schwenk, Tuan A. Tran, Arina Turkatenko, Carleigh Wood
| Challenge: | Using long text outputs to evaluate progress in summarization and summary expansion tasks is challenging. |
| Approach: | They propose a framework for assessing gradual summarization and summary expansion capabilities across diverse domains. |
| Outcome: | The proposed framework provides alignments between specific QA pairs and corresponding summaries in 7 domains. |
Span-based Semantic Role Labeling as Lexicalized Constituency Tree Parsing (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing models for semantic role labeling fail to capture the relationship between syntax and semantics. |
| Approach: | They propose a lexicalized tree representation for span-based SRL that integrates constituency and dependency parsing to explicitly model predicate-argument structures. |
| Outcome: | The proposed model achieves competitive performance on standard English benchmarks. |
Learning from Negative Samples in Biomedical Generative Entity Linking (2025.findings-acl)
Copied to clipboard
| Challenge: | Generative models are usually trained only with positive samples and do not explicitly learn from hard negative samples, which are entities that look similar but have different meanings. |
| Approach: | They propose a framework that trains generative BioEL models using negative samples to learn from hard negative samples. |
| Outcome: | The proposed framework outperforms baseline models by up to an average top-1 accuracy of 1.4% on five benchmarks. |
Self-play through Computational Runtimes improves Chart Reasoning (2025.findings-acl)
Copied to clipboard
| Challenge: | Vision-language models (VLMs) achieve impressive zero-shot performance on multimodal reasoning tasks. |
| Approach: | They propose a self-play programming interface which leverages visual reasoning models to generate code to decompose a complex visual reasoning task in sub-tasks and use it as a tool to solve decomposed tasks. |
| Outcome: | The proposed approach outperforms baselines on difficult chart reasoning benchmarks. |
Towards Better Chain-of-Thought: A Reflection on Effectiveness and Faithfulness (2025.findings-acl)
Copied to clipboard
| Challenge: | Chain-of-thought (CoT) prompting demonstrates varying performance under different reasoning tasks. |
| Approach: | They propose to recall extra information from the question to enhance CoT generation and evaluate CoTs based on their information gain. |
| Outcome: | The proposed method improves both the faithfulness and effectiveness of CoT and evaluates it based on their information gain. |
A Couch Potato is not a Potato on a Couch: Prompting Strategies, Image Generation, and Compositionality Prediction for Noun Compounds (2025.findings-acl)
Copied to clipboard
| Challenge: | a new method to predict the compositionality of English noun compounds is proposed . |
| Approach: | They propose a visual modality and vision transformers to predict the compositionality of English noun compounds. |
| Outcome: | The proposed method compared with a state-of-the-art text-based approach reveals complementary contributions regarding features and degrees of abstractness in English noun compounds. |
A Rose by Any Other Name: LLM-Generated Explanations Are Good Proxies for Human Explanations to Collect Label Distributions on NLI (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent research has shown that explanations provide valuable information for understanding human label variation (HLV) Large language models (LLMs) can approximate HJD from a few human-provided label-explanation pairs, but collecting explanations for every label is still time-consuming. |
| Approach: | They propose to use Large Language Models (LLMs) as annotators to generate model explanations for a few given human labels. |
| Outcome: | The proposed models can generate human-provided explanations from human labels, but they are still time-consuming. |
Measuring What Matters: Evaluating Ensemble LLMs with Label Refinement in Inductive Coding (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) are prone to inconsistencies and individual biases, limiting their reliability. |
| Approach: | They propose a framework that combines ensemble methods with code refinement methodology to address these challenges. |
| Outcome: | The proposed framework outperforms large language models and LLMs with a low-rank averaging and a moderator-based mechanism to simulate human consensus. |
Dynamic Evil Score-Guided Decoding: An Efficient Decoding Framework For Red-Team Model (2025.findings-acl)
Copied to clipboard
Cong Gao, Bo Zhang, Linkang Yang, Minghao Hu, Zhunchen Luo, Xiaoying Bai, Guotong Geng, Jun Zhang, Yunhua Xue
| Challenge: | Existing red-teaming methods require expensive fine-tuning, especially for large LLMs. |
| Approach: | They propose a red-teaming method that uses an ‘evil score’ to evaluate the potential of tokens to contribute to harmful outputs during decoding. |
| Outcome: | The proposed method achieves an ASR of 92.83% on the Llama-3.2-3B-Instruct model, compared to 83.48% with adversarial fine-tuning while using less computational resources. |
CoMuMDR: Code-mixed Multi-modal Multi-domain corpus for Discourse paRsing in conversations (2025.findings-acl)
Copied to clipboard
| Challenge: | Discourse parsing datasets based on conversations are restricted to a single domain . a lack of discourse structures in audio-based conversations is a challenge . |
| Approach: | They introduce CoMuMDR: Code-mixed Multi-modal Multi-domain corpus for Discourse parsing in conversations. |
| Outcome: | The proposed corpus is code-mixed in Hindi and English and annotated with nine discourse relations. |
Multi-word Measures: Modeling Semantic Change in Compound Nouns (2025.findings-acl)
Copied to clipboard
| Challenge: | Compound words provide a multifaceted challenge for diachronic models of semantic change . novel sense-targeting approach targets both noun compounds and their constituent parts . |
| Approach: | They propose a dataset of relatedness judgements of noun compounds in English and german . they use contrasting vector representations to evaluate their ability to cluster example sentence pairs . |
| Outcome: | The proposed approach captures diachronic meaning changes for multi-word expressions without condensing individual senses into an aggregate value. |
Bridge-Coder: Transferring Model Capabilities from High-Resource to Low-Resource Programming Language (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) excel at generating code for high-resource programming languages (HRPLs) however, they struggle significantly with low-resourced programming languages such as D, exacerbating the digital divide. |
| Approach: | They propose a method to generate LRPL data using LLM's general knowledge, HRPL proficiency, and in-context learning capabilities. |
| Outcome: | The proposed method improves on R, D, Racket, and Bash, while maintaining the same quality. |
ProBench: Judging Multimodal Foundation Models on Open-ended Multi-domain Expert Tasks (2025.findings-acl)
Copied to clipboard
| Challenge: | Solving expert-level multimodal tasks requires strong user query understanding, domain-specific knowledge, and advanced reasoning abilities. |
| Approach: | They propose a benchmark of open-ended user queries encapsulating professional expertise and advanced reasoning. |
| Outcome: | The proposed benchmark is publicly accessible at TBC. |
2M-BELEBELE: Highly Multilingual Speech and American Sign Language Comprehension Dataset Download PDF (2025.findings-acl)
Copied to clipboard
Marta R. Costa-jussà, Bokai Yu, Pierre Andrews, Belen Alastruey, Necati Cihan Camgoz, Joe Chuang, Jean Maillard, Christophe Ropers, Arina Turkatenko, Carleigh Wood
| Challenge: | We extend the BELEBELE dataset to speech and sign, and extend the Automatic Speech Recognition Benchmark, FLEURS, by 20%. |
| Approach: | They extend the BELEBELE and FLEURS speech comprehension datasets to speech and sign . they evaluate the datasets for 5-shot and zero-shot settings and find that the accuracy is 10% lower than reading comprehension. |
| Outcome: | The proposed dataset covers 91 spoken languages and one sign language (ASL) it also extends the Automatic Speech Recognition Benchmark, FLEURS, by 20% across languages. |
LSC-Eval: A General Framework to Evaluate Methods for Assessing Dimensions of Lexical Semantic Change Using LLM-Generated Synthetic Data (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for measuring Lexical Semantic Change are lacking historical benchmarks. |
| Approach: | They propose a three-stage general-purpose evaluation framework that simulates theory-driven LSC using In-Context Learning and a lexical database. |
| Outcome: | The proposed framework evaluates the sensitivity of computational methods to synthetic change and their suitability for detecting change in specific dimensions and domains. |
Chain-of-Jailbreak Attack for Image Generation Models via Step by Step Editing (2025.findings-acl)
Copied to clipboard
Wenxuan Wang, Kuiyi Gao, Youliang Yuan, Jen-tse Huang, Qiuzhi Liu, Shuai Wang, Wenxiang Jiao, Zhaopeng Tu
| Challenge: | Text-based image generation models, such as Stable Diffusion and DALL-E 3, hold significant potential in content creation and publishing workflows . however, considerable efforts are being made to prevent the generation of harmful content, such abusive, violent, or pornographic material. |
| Approach: | They propose a chain-of-jailbreak method which decomposes malicious queries into multiple sub-queries and iteratively edits images based on these sub-questions. |
| Outcome: | The proposed method can bypass safeguards of image generation models for over 60% cases, significantly outperforms other jailbreaking methods (14%) |
Tokenization is Sensitive to Language Variation (2025.findings-acl)
Copied to clipboard
| Challenge: | Variation in language is often linked to regional, social, and contextual factors. |
| Approach: | They propose a method to estimate tokenizer impact on downstream LLM performance . they pre-train BERT models with the popular Byte-Pair Encoding algorithm . |
| Outcome: | The proposed model improves on Rényi efficiency and other metrics on language variation. |
WirelessMathBench: A Mathematical Modeling Benchmark for LLMs in Wireless Communications (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have demonstrated impressive results across a broad array of tasks, yet their capacity for complex, domain-specific mathematical reasoning remains underexplored. |
| Approach: | They propose a benchmark to evaluate Large Language Models on mathematical modeling challenges to wireless communications engineering. |
| Outcome: | The proposed benchmark evaluates LLMs on mathematical modeling challenges to wireless communications engineering. |
Self-Improvement Towards Pareto Optimality: Mitigating Preference Conflicts in Multi-Objective Alignment (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing approaches to optimize large language models with human preferences suffer from preference conflicts in the data. |
| Approach: | They propose to construct Pareto-optimal responses to resolve preference conflicts by using a self-improving DPO framework that enables LLMs to self-generate and select Paret-optimized responses. |
| Outcome: | The proposed framework achieves superior Pareto Front performance over baselines on two datasets. |
Investigating and Scaling up Code-Switching for Multilingual Language Model Pre-Training (2025.findings-acl)
Copied to clipboard
Zhijun Wang, Jiahuan Li, Hao Zhou, Rongxiang Weng, Jingang Wang, Xin Huang, Xue Han, Junlan Feng, Chao Deng, Shujian Huang
| Challenge: | Large language models (LLMs) exhibit remarkable multilingual capabilities despite the extreme language imbalance in the pre-training data. |
| Approach: | They investigate the existence of code-switching in the pre-training corpus and categorize it into four types within two quadrants. |
| Outcome: | The proposed approach improves performance across benchmarks and representation space. |
User Behavior Prediction as a Generic, Robust, Scalable, and Low-Cost Evaluation Strategy for Estimating Generalization in LLMs (2025.findings-acl)
Copied to clipboard
| Challenge: | We argue that knowledge-retrieval and reasoning tasks are not ideal for measuring generalization, as LLMs are not trained for specific tasks. |
| Approach: | They propose a statistically motivated framework using personalization to assess generalization in Large Language Models. |
| Outcome: | The proposed framework outperforms existing models on movie and music recommendation datasets, but all models have room for improvement, especially Llama. |
Beyond Browsing: API-Based Web Agents (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing web agents use browsers to facilitate human activities such as online shopping, online planning, and other work-related tasks. |
| Approach: | They propose to use web browsers as an interface to interact with online content through application programming interfaces (APIs) they propose to call APIs and use Hybrid Agents to perform online tasks. |
| Outcome: | The proposed agents outperform web Browsing Agents on a widely-used and realistic benchmark for web navigation tasks. |
MiLiC-Eval: Benchmarking Multilingual LLMs for China’s Minority Languages (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models excel in high-resource languages but struggle with low-resourced languages . minority languages such as Tibetan, Uyghur, Kazakh, and Mongolian are marginalized in NLP research due to limited digital representation and the scarcity of training data. |
| Approach: | They propose a benchmark for minority languages in China that tracks the progress of large language models on low-resource languages. |
| Outcome: | The proposed benchmark focuses on underrepresented writing systems and syntax-intensive tasks. |
ArgInstruct: Specialized Instruction Fine-Tuning for Computational Argumentation (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) have been trained to follow instructions for many NLP tasks, including several tasks from computational argumentation (CA), the computational analysis and synthesis of natural language arguments. |
| Approach: | They propose a specialized instruction fine-tuning for the domain of computational argumentation (CA) they synthesized 52k CA-related instructions and used them to train a CA-specialized instruction-following LLM. |
| Outcome: | The proposed benchmarks show that the LLMs can tackle unseen and seen tasks while maintaining generalization capabilities. |
Crabs: Consuming Resource via Auto-generation for LLM-DoS Attack under Black-box Settings (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing studies on white-box attacks focus on black-box LLMs, leaving black- box scenarios underexplored. |
| Approach: | They propose an automated algorithm designed for black-box LLMs that constructs the DoS Attack Tree and expands the node coverage to achieve effectiveness under black- box conditions. |
| Outcome: | The proposed algorithm can be used to build a DoS Attack Tree and expand the node coverage to achieve effectiveness under black-box conditions. |
Probabilistic Aggregation and Targeted Embedding Optimization for Collective Moral Reasoning in Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have impressive moral reasoning abilities, yet they often diverge when confronted with complex, multi-factor moral dilemmas. |
| Approach: | They propose a framework that synthesizes multiple LLMs’ moral judgments into a collectively formulated moral judgment, realigning models that deviate significantly from this consensus. |
| Outcome: | The proposed framework synthesizes multiple LLMs’ moral judgments into a collectively formulated moral judgment, realigning models that deviate significantly from this consensus. |
Unlocking Recursive Thinking of LLMs: Alignment via Refinement (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for recursive reasoning are limited due to lack of expert-curated data. |
| Approach: | They propose a method that unlocks the potential of Large Language Models for recursive reasoning through long-form Chain of Thought. |
| Outcome: | The proposed method outperforms preference optimization methods on the openAI o1-series models by 20% on 3k synthetic samples. |
CitaLaw: Enhancing LLM with Citations in Legal Domain (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing benchmarks have focused on enabling large language models (LLMs) to generate citationsupported outputs. |
| Approach: | They propose to use a citation-based framework to evaluate LLMs' ability to produce legally sound responses with appropriate citations. |
| Outcome: | The proposed framework enables LLMs to retrieve supporting citations from the reference corpus and align these citation with the corresponding sentences in their responses. |
MEGen: Generative Backdoor into Large Language Models via Model Editing (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for training large language models are limited to yes-or-no discriminative tasks, leading users to underestimate the potential risks. |
| Approach: | They propose an editing-based generative backdoor that expands the backdoor to generative tasks in a unified format of any text-to-any text. |
| Outcome: | The proposed model achieves high attack success rate by adjusting only a small set of local parameters with few-shot samples. |
Social Bias Benchmark for Generation: A Comparison of Generation and QA-Based Evaluations (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for assessing social bias in large language models (LLMs) do not capture nuanced and context-dependent nature of natural language generation. |
| Approach: | They propose a Bias Benchmark for Generation (BBG) that evaluates social bias in long-form generation by having LLMs generate continuations of story prompts. |
| Outcome: | The proposed benchmark is based on the English BBQ and Korean BBQ datasets and compares it with multiplechoice BBQ evaluation. |
Generating Pedagogically Meaningful Visuals for Math Word Problems: A New Benchmark and Analysis of Text-to-Image Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Math word problems (MWPs) describe mathematical scenarios through text, requiring learners to interpret both linguistic and numerical information to derive mathematical expressions. |
| Approach: | They propose a framework for generating pedagogically meaningful visuals from MWP text descriptions using a pre-defined visual language and a design space grounded in interviews with math teachers. |
| Outcome: | The proposed framework illustrates the core mathematical relationships in math word problems. |
RASPberry: Retrieval-Augmented Monte Carlo Tree Self-Play with Reasoning Consistency for Multi-Hop Question Answering (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for generating and analyzing multiple document knowledge are not effective for multi-hop question answering. |
| Approach: | They propose a Monte Carlo tree-based approach to inference-time scaling using RASPberry. |
| Outcome: | Experimental results show that the proposed method achieves better inference-time scaling on smaller LLMs. |
All That Glitters is Not Gold: Improving Robust Retrieval-Augmented Language Models with Fact-Centric Preference Alignment (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods to learn adaptive retrieval for noisy documents lack prior filtering and may lead to the loss of crucial information. |
| Approach: | They propose a method to improve retrieval performance without prior filtering . they use LLMs self-generated synthetic data as training data without manual annotation . |
| Outcome: | The proposed method performs positive document mining based on factual consistency and uses LLMs self-generated synthetic data as training data without manual annotation. |
FairSteer: Inference Time Debiasing for LLMs with Dynamic Activation Steering (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing prompt-based debiasing methods exhibit instability due to sensitivity to prompt changes . fine-tuning-based techniques incur substantial computational overhead and catastrophic forgetting . |
| Approach: | They propose a debiasing framework that encodes fairness-related features into separable directions in the hidden activation space. |
| Outcome: | The proposed framework performs inference-time debiasing without requiring retraining or prompt design . it detects bias signatures in activations and then computes debiased steering vectors . the proposed framework is available to download in the u.s. |
Listen, Watch, and Learn to Feel: Retrieval-Augmented Emotion Reasoning for Compound Emotion Generation (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods to assess human emotion are limited by the subjective nature of emotion perception, limiting the robustness of existing models. |
| Approach: | They propose a plug-and-play module that enhances MLLMs’ ability to tackle compound and context-rich emotion tasks. |
| Outcome: | The proposed framework improves MLLMs' ability to tackle compound and context-rich emotion tasks and the Compound Emotion QA dataset shows it performs well across both benchmarks and evaluation frameworks. |
GLTW: Joint Improved Graph Transformer and LLM via Three-Word Language for Knowledge Graph Completion (2025.findings-acl)
Copied to clipboard
Kangyang Luo, Yuzhuo Bai, Cheng Gao, Shuzheng Si, Zhu Liu, Yingli Shen, Zhitong Wang, Cunliang Kong, Wenhao Li, Yufei Huang, Ye Tian, Xuantang Xiong, Lei Han, Maosong Sun
| Challenge: | Existing knowledge graphs lack the ability to integrate structural information into LLMs and output predictions deterministically. |
| Approach: | They propose a method which encodes structural information of KGs and merges it with LLMs to enhance KGC performance. |
| Outcome: | The proposed method improves the performance of KG Completion datasets on KGs by integrating structural information with LLMs. |
Learning to Select In-Context Demonstration Preferred by Large Language Model (2025.findings-acl)
Copied to clipboard
| Challenge: | In-context learning (ICL) enables large language models to perform tasks with only a few examples as demonstrations. |
| Approach: | They propose a generative preference learning framework that leverages LLM feedback to directly optimize demonstration selection for ICL. |
| Outcome: | Experiments on 19 datasets across 11 task categories show that GenICL achieves superior performance than existing methods in selecting the most effective demonstrations. |
Beyond the Spelling Miracle: Investigating Substring Awareness in Character-Blind Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Current Pre-trained Language Models are character-blind and struggle in spelling tasks . ability to identify characters and substrings within words is trivial but fundamental to robust language understanding. |
| Approach: | They propose to evaluate pre-trained language models with a binary substring identification task . they propose to examine where, when, and how a PLMs develop awareness of characters and substrings . |
| Outcome: | The proposed model identifies characters and substrings in a binary substring identification task. |
DEMO: Reframing Dialogue Interaction with Fine-grained Element Modeling (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) enabled dialogue systems are one of the central modes in human-machine interaction. |
| Approach: | They propose a benchmark task for dialogue element MOdeling and Element Awareness and a new benchmark for dialogue agent interaction that allows the agent to model dialogue elements via imitation learning. |
| Outcome: | The proposed agent performs well in both dialogue element modeling and out-of-domain tasks. |
InfiniteICL: Breaking the Limit of Context Window Size via Long Short-term Memory Transformation (2025.findings-acl)
Copied to clipboard
| Challenge: | InfiniteICL is a framework that parallels context and parameters in large language models with short- and long-term memory in human cognitive systems. |
| Approach: | They propose a framework that parallels context and parameters in large language models with short- and long-term memory in human cognitive systems and enables infinite context integration. |
| Outcome: | The proposed framework reduces context length by 90% while achieving 103% average performance of full-context prompting across fact recall, grounded reasoning, and skill acquisition tasks. |
M3HG: Multimodal, Multi-scale, and Multi-type Node Heterogeneous Graph for Emotion Cause Triplet Extraction in Conversations (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for ECAC focus on textual contexts, overlooking other modalities. |
| Approach: | They propose a multimodal, multi-scenario MECTEC dataset that captures emotional and causal contexts and effectively fuses contextual information at different levels. |
| Outcome: | The proposed model captures emotional and causal contexts and effectively fuses contextual information at both inter- and intra-utterance levels. |
Large Language Models Are Natural Video Popularity Predictors (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) can better capture cultural and social factors such as viewing intensity and geographic spread of video content. |
| Approach: | They propose to use Large Language Models to capture cultural and social factors that influence video popularity and generate interpretable, attribute-based explanations. |
| Outcome: | The proposed model captures both engagement intensity and geographic spread on 13,639 popular videos, while the neural network's predictions reach 82% without fine-tuning. |
DELMAN: Dynamic Defense Against Large Language Model Jailbreaking with Model Editing (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing safety mechanisms for Large Language Models (LLMs) are inadequate to protect against jailbreak attacks, resulting in performance degradation on general tasks. |
| Approach: | They propose a method that directly updates a minimal set of relevant parameters to neutralize harmful behaviors while preserving the model’s utility. |
| Outcome: | The proposed model outperforms baseline methods in mitigating jailbreak attacks while preserving the model’s utility. |
You need to MIMIC to get FAME: Solving Meeting Transcript Scarcity with Multi-Agent Conversations (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing tools for meeting summarization are limited due to privacy and expensive manual annotation. |
| Approach: | They propose a meeting synthesis framework that generates meeting transcripts on a given knowledge source by defining psychologically grounded participant profiles, outlining the conversation, and orchestrating a large language model (LLM) debate. |
| Outcome: | The proposed framework generates meeting transcripts on a given knowledge source by defining psychologically grounded participant profiles, outlining the conversation, and orchestrating a large language model debate. |
Code-Switching and Syntax: A Large-Scale Experiment (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing theories of code-switching (CS) have been refuted in subsequent investigations. |
| Approach: | They propose to use syntactic information to predict where bilinguals switch languages . they find that syntax alone is sufficient for an automatic system to distinguish between sentences in minimal pairs of CS, to the same degree as bilingual humans. |
| Outcome: | The proposed model can explain why bilinguals switch languages more often than in others, but there is no large-scale, multi-language, cross-phenomena experiment that tests this claim. |
Optima: Optimizing Effectiveness and Efficiency for LLM-Based Multi-Agent System (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have emerged as powerful tools for a wide range of tasks, from * Equal Contribution. |
| Approach: | They propose a framework that enhances communication efficiency and task effectiveness in LLM-based multi-agent systems through training. |
| Outcome: | The proposed framework improves communication efficiency and task effectiveness on multi-agent tasks with 2.8x performance gain with less than 10% tokens on tasks requiring heavy information exchange. |
Generating Domain-Specific Knowledge Graphs from Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) have shown impressive world knowledge across different benchmarks and domains but their knowledge is inconveniently scattered across their billions of parameters. |
| Approach: | They propose a prompt-based method to extract knowledge solely from LLMs’ parameters to construct domain-specific KGs by a schema-based process. |
| Outcome: | The proposed method generates large domain-specific KGs containing tens of thousands of entities and relations, and then evaluates against Wikidata, an open-source human-created KG. |
Large Language Models are Miscalibrated In-Context Learners (2025.findings-acl)
Copied to clipboard
| Challenge: | In-context Learning and Supervised Fine-Tuning have emerged as pre-dominant methodologies for machine learning and NLP. |
| Approach: | They propose to use self-ensembling to improve both performance and calibration of language models. |
| Outcome: | The proposed learning paradigms can achieve better calibration and better performance than the previous learning paradigm. |
STeCa: Step-level Trajectory Calibration for LLM Agent Learning (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing work focuses on behavior cloning from expert demonstrations or preference learning through exploratory trajectory sampling, but these methods often struggle to address long-horizon tasks where suboptimal actions accumulate step by step, causing agents to deviate from correct task trajectories. |
| Approach: | They propose a framework for LLM-based agent learning that identifies suboptimal actions through a step-level reward comparison during exploration and constructs calibrated trajectories using LLM reflection. |
| Outcome: | The proposed framework outperforms existing methods in long-horizon tasks where suboptimal actions accumulate step by step, causing agents to deviate from correct task trajectories. |
LEMMA: Learning from Errors for MatheMatical Advancement in LLMs (2025.findings-acl)
Copied to clipboard
Zhuoshi Pan, Yu Li, Honglin Lin, Qizhi Pei, Zinan Tang, Wei Wu, Chenlin Ming, H. Vicky Zhao, Conghui He, Lijun Wu
| Challenge: | Existing approaches focus on improving the quality of correct training data, neglecting the value contained in error data, thereby hindering the model’s reflective ability. |
| Approach: | They propose to improve LLM's reasoning ability by learning from error data and a grounded mistake augmentation method to collect representative errors. |
| Outcome: | The proposed model achieves significant performance improvements over other strong models with less than 90k data. |
Voting or Consensus? Decision-Making in Multi-Agent Debate (2025.findings-acl)
Copied to clipboard
| Challenge: | Increasing the number of agents improves performance, while more discussion rounds before voting reduces it. |
| Approach: | They propose two new methods to improve multi-agent debates by increasing agent diversity and reducing discussion rounds before voting. |
| Outcome: | The proposed methods improve task performance by up to 3.3% with AAD and up to 7.4% with CI. |
Rhetorical Device-Aware Sarcasm Detection with Counterfactual Data Augmentation (2025.findings-acl)
Copied to clipboard
| Challenge: | Sarcasm is a complex form of sentiment expression widely used in human daily life. |
| Approach: | They propose a device-aware sarcasm dataset with counterfactually augmented data to capture its complexity. |
| Outcome: | The proposed dataset shows that it is more balanced than zero-shot models. |
Selecting Demonstrations for Many-Shot In-Context Learning via Gradient Matching (2025.findings-acl)
Copied to clipboard
| Challenge: | In-Context Learning (ICL) empowers Large Language Models for rapid task adaptation without fine-tuning. |
| Approach: | They propose a method that aligns fine-tuning gradients between entire training set and selected examples to enable in-context learning and fine-uning. |
| Outcome: | The proposed method outperforms random selection on large LLMs from 4-shot to 128-shot scenarios across 9 datasets. |
Cheap Character Noise for OCR-Robust Multilingual Embeddings (2025.findings-acl)
Copied to clipboard
| Challenge: | Optical character recognition (OCR) is a key component of the digitization of historical documents. |
| Approach: | They propose a method that fine-tunes existing multilingual models using noisy texts and a contrastive loss. |
| Outcome: | The proposed model improves on the training data of existing models using noisy texts and a contrastive loss. |
Physics: Benchmarking Foundation Models on University-Level Physics Problem Solving (2025.findings-acl)
Copied to clipboard
| Challenge: | a benchmark for university-level physics problem solving contains 1,297 expert-annotated problems . a proprietary model, o3-mini, achieves only 59.9% accuracy, highlighting fundamental weaknesses in scientific reasoning, conceptual understanding, and mathematical precision. |
| Approach: | They introduce Physics, a benchmark for university-level physics problem solving. |
| Outcome: | The proposed model achieves only 59.9% accuracy on the most advanced model, o3-mini . the proposed model is a powerful tool for evaluating models on advanced problems . |
DOVE: A Large-Scale Multi-Dimensional Predictions Dataset Towards Meaningful LLM Evaluation (2025.findings-acl)
Copied to clipboard
Eliya Habba, Ofir Arviv, Itay Itzhak, Yotam Perlitz, Elron Bandel, Leshem Choshen, Michal Shmueli-Scheuer, Gabriel Stanovsky
| Challenge: | Recent work found that LLMs are sensitive to arbitrary prompt dimensions . this challenges traditional single-prompt evaluation practices . |
| Approach: | They present a large-scale dataset containing prompt perturbations of various evaluation benchmarks . they examine LLM sensitivity from an holistic perspective and assess the joint effects of perturbations along various dimensions . |
| Outcome: | The proposed dataset aims to democratize evaluation research and enable robust protocols . it includes more than 250M prompt perturbations and model outputs . |
ALPS: Attention Localization and Pruning Strategy for Efficient Adaptation of Large Language Models (2025.findings-acl)
Copied to clipboard
Hao Chen, Haoze Li, Zhiqing Xiao, Lirong Gao, Qi Zhang, Xiaomeng Hu, Ningtao Wang, Xing Fu, Junbo Zhao
| Challenge: | Prior research has focused on optimizing general-purpose large language models to downstream tasks . however, these approaches inherently introduce data dependency, which hinders generalization and reusability. |
| Approach: | They propose an algorithm that localizes the most task-sensitive attention heads and prunes by restricting attention training updates to these heads, thereby reducing alignment costs. |
| Outcome: | The proposed algorithm achieves 2% performance improvement over baselines on three tasks while localizing the most task-sensitive attention heads. |
DeTAM: Defending LLMs Against Jailbreak Attacks via Targeted Attention Modification (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing defense methods rely on fine-tuning or input modification, which suffer from limited generalization and reduced utility. |
| Approach: | They propose a finetuning-free approach that improves the defensive capabilities against jailbreak attacks of LLMs via targeted attention modification. |
| Outcome: | The proposed approach outperforms baselines in jailbreak defense and exhibits robust generalization across attacks and models, maintaining its effectiveness even on in-the-wild jailbreak data. |
A Survey of Mathematical Reasoning in the Era of Multimodal Large Language Model: Benchmark, Method & Challenges (2025.findings-acl)
Copied to clipboard
Yibo Yan, Jiamin Su, Jianxiang He, Fangteng Fu, Xu Zheng, Yuanhuiyi Lyu, Kun Wang, Shen Wang, Qingsong Wen, Xuming Hu
| Challenge: | This survey provides **the first comprehensive analysis of mathematical reasoning in the era of multimodal large language models** . integrating large language model with mathematical reasoning tasks is becoming significant as AI advances . |
| Approach: | They review over 200 studies published since 2021 and examine the state-of-the-art developments in Math-LLMs . they identify five major challenges hindering the realization of AGI in this domain . |
| Outcome: | The authors examine the state-of-the-art developments in Math-LLMs with a focus on multimodal settings. |
Fast-and-Frugal Text-Graph Transformers are Effective Link Predictors (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods that encode textual and structural information for inductive link prediction are frugal and fast at training and inference time. |
| Approach: | They propose a Transformer-based framework that unifies textual and structural information for inductive link prediction in text-attributed knowledge graphs by encoding ego-graphs (1-hop neighbourhoods). |
| Outcome: | The proposed framework can achieve superior performance on three popular datasets and reduce the reliance on resource-intensive encoders. |
NeoQA: Evidence-based Question Answering with Generated News Events (2025.findings-acl)
Copied to clipboard
| Challenge: | Evaluating Retrieval-Augmented Generation (RAG) in large language models is challenging because benchmarks can quickly become stale. |
| Approach: | They propose a benchmark to evaluate Retrieval-Augmented Generation (RAG) in large language models (LLMs) using timelines and knowledge bases of fictional news events and entities to prevent LLMs from leveraging pretraining knowledge. |
| Outcome: | The proposed benchmark prevents LLMs from leveraging pretraining knowledge and ensures that no prior evidence exists in their training data. |
ChatMap: Mining Human Thought Processes for Customer Service Chatbots via Multi-Agent Collaboration (2025.findings-acl)
Copied to clipboard
Xinyi Jiang, Tianyi Hu, Yuheng Qin, Guoming Wang, Zhou Huan, Kehan Chen, Gang Huang, Rongxing Lu, Siliang Tang
| Challenge: | Existing methods for enhancing dialogue performance rely on summarizing behavior . e-commerce chatbots need to align their dialogue strategies with human behavior to achieve coherent, human-like conversations with customers. |
| Approach: | They propose a method to extract core patterns from dialogue data and integrate them into models by mining service thought processes using a multi-agent aPproach. |
| Outcome: | The proposed method outperforms manual methods and outperfies baselines on Taobao in China. |
P3: Prompts Promote Prompting (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent advances in prompt optimization have shown effectiveness of using multiple components to optimize models . however, such unilateral approaches often yield suboptimal results due to interdependent nature of these components. |
| Approach: | They propose a self-improvement framework that optimizes both system and user prompts . they use offline optimized prompts to promote online prompt optimization . |
| Outcome: | The proposed framework improves performance on general and reasoning tasks. |
VAQUUM: Are Vague Quantifiers Grounded in Visual Data? (2025.findings-acl)
Copied to clipboard
| Challenge: | a dataset containing 20,300 human ratings on quantified statements is used to evaluate the appropriateness of vague quantifiers in visual contexts. |
| Approach: | They use a visual-language-models-based dataset to evaluate the appropriateness of vague quantifiers. |
| Outcome: | The proposed model is based on a visual-visual-language-model-based dataset . it shows that the model is compatible with humans when producing or judging vague quantifiers . |
Forgotten Polygons: Multimodal Large Language Models are Shape-Blind (2025.findings-acl)
Copied to clipboard
William Rudman, Michal Golovanevsky, Amir Bar, Vedant Palit, Yann LeCun, Carsten Eickhoff, Ritambhara Singh
| Challenge: | Multimodal Large Language Models struggle with visual reasoning, despite strong performance on vision-language tasks. |
| Approach: | They propose a visually cued chain-of-thought prompting that enhances multi-step mathematical reasoning by explicitly referencing visual annotations in diagrams. |
| Outcome: | The proposed model improves GPT-4o's accuracy on an irregular polygon side-counting task from 7% to 93%. |
MindBridge: Scalable and Cross-Model Knowledge Editing via Memory-Augmented Modality (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing knowledge editing methods overfit to specific models, causing edited knowledge to be discarded during each LLM update and requiring frequent re-editing. |
| Approach: | They propose a solution that allows editors to edit knowledge in multiple LLMs at the same time. |
| Outcome: | The proposed solution performs better even in editing tens of thousands of knowledge entries and can adapt to different LLMs. |
FIHA: Automated Fine-grained Hallucinations Evaluations in Large Vision Language Models with Davidson Scene Graphs (2025.findings-acl)
Copied to clipboard
| Challenge: | Current approaches to large vision-language models rely on costly annotations and are not comprehensive in terms of evaluating all aspects. |
| Approach: | They propose an automated method which can access LVLMs hallucination in an LLM-free and annotation-free way and model the dependency between different types of halluciNations. |
| Outcome: | The proposed model can model the dependency between different types of hallucinations and generate Q&A pairs on any image dataset at minimal cost. |
On the Role of Semantic Proto-roles in Semantic Analysis: What do LLMs know about agency? (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing studies on large language models (LLMs) have not explored their capacity to reason over event structure . et al., 2015, 142: e007-e0027; eugene, 1985; Weiner, 1995; saab, 1985) focus on the role of large language model in decision-making . |
| Approach: | They propose to characterize agents via properties such as "instigation" and "volition" they also examine whether incorporating semantic proto-role labeling context improves SRL performance . |
| Outcome: | The proposed model improves in a zero-shot setting by incorporating proto-role labeling context . the results support previous work showing that LLMs underperform human annotators in complex semantic analysis. |
GeAR: Graph-enhanced Agent for Retrieval-augmented Generation (2025.findings-acl)
Copied to clipboard
Zhili Shen, Chenxin Diao, Pavlos Vougiouklis, Pascual Merita, Shriram Piramanayagam, Enting Chen, Damien Graux, Andre Melo, Ruofei Lai, Zeren Jiang, Zhongyang Li, Ye Qi, Yang Ren, Dandan Tu, Jeff Z. Pan
| Challenge: | Retrieval-augmented Generation (RAG) relies on effective retrieval capabilities, yet traditional sparse and dense retrievers struggle with multi-hop retrieval scenarios. |
| Approach: | They propose a graph expansion mechanism that augments any conventional base retriever and an agent framework that incorporates the resulting graph-based retrieval into a multi-step retrieval framework. |
| Outcome: | The proposed system achieves state-of-the-art results on three multi-hop question answering datasets while consuming fewer tokens and requiring fewer iterations than existing multi-step retrieval systems. |
WebNLG-IT: Construction of an aligned RDF-Italian corpus through Machine Translation techniques (2025.findings-acl)
Copied to clipboard
| Challenge: | Using NMT and hand-written rules, we created the first aligned Italian RDF-to-text corpus . |
| Approach: | They propose to use NMT to create an Italian version of the WebNLG corpus and to refine and improve the quality of the produced resource. |
| Outcome: | The proposed system is the best on the original English version and the best in the second step, it improves and refines the quality of the produced resource. |
Towards Adapting Open-Source Large Language Models for Expert-Level Clinical Note Generation (2025.findings-acl)
Copied to clipboard
Hanyin Wang, Chufan Gao, Bolun Liu, Qiping Xu, Guleid Hussein, Mohamad El Labban, Kingsley Iheasirim, Hariprasad Reddy Korsapati, Chuck Outcalt, Jimeng Sun
| Challenge: | Proprietary Large Language Models (LLMs) have demonstrated promising capabilities in clinical text summarization tasks. |
| Approach: | They propose a domain- and task-specific adaptation process for an open-source LLaMA-2 model . LLama-2 can generate high-quality clinical notes from outpatient patient-doctor dialogues . |
| Outcome: | The proposed model can generate clinical notes comparable to those authored by physicians. |
Bridging Robustness and Generalization Against Word Substitution Attacks in NLP via the Growth Bound Matrix Approach (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent studies have shown that adversarial examples can alter models' predicted sentiment due to their sensitivity to specific word choices. |
| Approach: | They propose a regularization technique to improve NLP model robustness by reducing the impact of input perturbations on model outputs. |
| Outcome: | The proposed method outperforms state-of-the-art methods in adversarial defense. |
Neuro-Symbolic Query Compiler (2025.findings-acl)
Copied to clipboard
| Challenge: | Retrieval-Augmented Generation (RAG) systems are limited in their ability to process information in open-source environments. |
| Approach: | They propose a neuro-symbolic framework inspired by linguistic grammar rules and compiler design to formalize complex queries using a minimal yet sufficient Backus-Naur Form grammar. |
| Outcome: | The proposed framework is based on a backus-naur form grammar and compiler design that maintains completeness while minimizing redundancy. |
Revealing and Mitigating the Local Pattern Shortcuts of Mamba (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent studies show that Mamba excels in tasks that involve localized key information but faces challenges with tasks that require handling distributed key information. |
| Approach: | They propose to introduce a global gate module into Mamba to address this problem by adding 4M extra parameters to the model. |
| Outcome: | The proposed model outperforms attention-based models on synthetic and synthetic tasks with only 4M extra parameters. |
Forget the Token and Pixel: Rethinking Gradient Ascent for Concept Unlearning in Multimodal Generative Models (2025.findings-acl)
Copied to clipboard
Jiaqi Li, Chuanyi Zhang, Miaozeng Du, Hui Zhang, Yongrui Chen, Qianshan Wei, Junfeng Fang, Ruipeng Wang, Sheng Bi, Guilin Qi
| Challenge: | Gradient Ascent (GA) has emerged as a promising approach for concept unlearning in Multimodal Generative Models (MGMs). |
| Approach: | They propose a novel approach that selectively applies GA to targeted Conceptual Knowledge while preserving Natural Knowledge through Gradient Descent (GD). |
| Outcome: | The proposed approach removes Conceptual Knowledge and inadvertently diminishes Natural Knowledge, resulting in utility degradation. |
Slamming: Training a Speech Language Model on One GPU in a Day (2025.findings-acl)
Copied to clipboard
| Challenge: | *Slam* is a recipe for training high-quality Speech Language Models (SLMs) on a single academic GPU in 24 hours. |
| Approach: | They propose a recipe for training high-quality Speech Language Models on a single academic GPU in 24 hours. |
| Outcome: | The proposed training recipe outperforms predicted compute optimal performance, giving an optimistic view to SLM feasibility. |
Boosting LLM Translation Skills without General Ability Loss via Rationale Distillation (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have demonstrated impressive performance across numerous NLP tasks, but fine-tuning them for Machine Translation (MT) often introduces catastrophic forgetting, compromising the broad general abilities of LLMs and introducing potential security risks. |
| Approach: | They propose a method that harnesses the strong generative capabilities of Large Language Models to create rationales for training data, which are then "replayed" to prevent forgetting. |
| Outcome: | The proposed approach harnesses the strong generative capabilities of LLMs to create rationales for training data, which are then “replayed” to prevent forgetting. |
Clarifying Underspecified Discourse Relations in Instructional Texts (2025.findings-acl)
Copied to clipboard
| Challenge: | Discourse relations can be optionally realized through explicit connectives such as “but” and “while”. |
| Approach: | They build a corpus of 4,274 text revisions in which a connective was explicitly inserted . they collect plausibility annotations on other connectives to check whether they represent suitable alternatives . |
| Outcome: | The proposed model predicts plausibility of individual connectives with up to 66% accuracy, but is not reliable when multiple relations are plausible. |
WMT24++: Expanding the Language Coverage of WMT24 to 55 Languages & Dialects (2025.findings-acl)
Copied to clipboard
Daniel Deutsch, Eleftheria Briakou, Isaac Rayburn Caswell, Mara Finkelstein, Rebecca Galor, Juraj Juraska, Geza Kovacs, Alison Lui, Ricardo Rei, Jason Riesa, Shruti Rijhwani, Parker Riley, Elizabeth Salesky, Firas Trabelsi, Stephanie Winkler, Biao Zhang, Markus Freitag
| Challenge: | In order to evaluate large language models (LLMs), it is important to collect benchmark datasets in order to assess their multilingual performance. |
| Approach: | They extend the WMT24 dataset to cover 55 languages by collecting new human-written references and post-edits for 46 new languages/dialects. |
| Outcome: | The proposed dataset covers 55 languages and provides best-performing MT systems in all 55 languages. |
Exploring Graph Representations of Logical Forms for Language Modeling (2025.findings-acl)
Copied to clipboard
| Challenge: | Graph-based formal-logical distributional semantics models are more data-efficient than textual counterparts. |
| Approach: | They propose a pretrained language model over graph representations of logical forms as a proof-of-concept. |
| Outcome: | The proposed model outperforms textual, transformer LMs on downstream tasks . the model is likely to scale with additional parameters and pretraining data . |
SEA-HELM: Southeast Asian Holistic Evaluation of Language Models (2025.findings-acl)
Copied to clipboard
Yosephine Susanto, Adithya Venkatadri Hulagadri, Jann Railey Montalan, Jian Gang Ngui, Xianbin Yong, Wei Qi Leong, Hamsawardhini Rengarajan, Peerat Limkonchotiwat, Yifan Mai, William Chandra Tjhi
| Challenge: | Existing LLM benchmarks are capable of evaluating specific capabilities in English as well as in various mid- to low-resource languages, but a comprehensive and culturally representative evaluation suite for the SEA languages has not been developed thus far. |
| Approach: | They propose a holistic linguistic and cultural LLM evaluation suite that emphasizes SEA languages and introduces a leaderboard that allows users to understand models’ multilingual and multicultural performance. |
| Outcome: | The proposed evaluation suite emphasizes SEA languages and supports Filipino, Indonesian, Tamil, Thai, and Vietnamese. |
TRANS-ZERO: Self-Play Incentivizes Large Language Models for Multilingual Translation Without Parallel Data (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have reshaped machine translation, but multilingual MT still relies heavily on parallel data for supervised fine-tuning. |
| Approach: | They propose a framework that leverages only monolingual data and the intrinsic multilingual knowledge of Large Language Models (LLMs). |
| Outcome: | The proposed framework matches models trained on large-scale parallel data and excels in non-English translation directions. |
A Conformal Risk Control Framework for Granular Word Assessment and Uncertainty Calibration of CLIPScore Quality Estimates (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing caption evaluation metrics lack granular assessments for errors within captions . lack of uncertainty quantification can give way to misleading scores, reducing user trust . |
| Approach: | They propose a conformal risk control framework to generate and calibrate CLIPScore distributions . they use a model-agnostic conformal framework to detect erroneous words . |
| Outcome: | The proposed method detects erroneous words while providing formal guarantees aligned with desired risk levels. |
SGDPO: Self-Guided Direct Preference Optimization for Language Model Alignment (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for aligning Large Language Models with human values are limited and results of DPO are not resilient. |
| Approach: | They propose a self-guided direct preference optimization algorithm that incorporates a pilot term to steer the gradient flow during the optimization process. |
| Outcome: | The proposed method can generate human-preferred response up to 9.19% higher than previous methods. |
Socratic Style Chain-of-Thoughts Help LLMs to be a Better Reasoner (2025.findings-acl)
Copied to clipboard
| Challenge: | Synthetic data generation is a promising approach to enhance reasoning capabilities of large language models. |
| Approach: | They propose a multi-agent debate framework based on the Socratic questioning strategy . they use socratic questions to deepen the thinking process and broaden it to motivate self-reflection . |
| Outcome: | The proposed framework outperforms existing methods on math and code generation tasks while maintaining affordable costs. |
Quantile Regression with Large Language Models for Price Prediction (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing approaches to structured prediction tasks focus on point estimates and lack systematic comparison across different methods. |
| Approach: | They propose a novel quantile regression approach that enables LLMs to produce full predictive distributions, improving upon traditional point estimates. |
| Outcome: | The proposed model outperforms encoder architectures, embedding-based methods, and few-shot learning methods in prediction accuracy and distributional calibration. |
Training Turn-by-Turn Verifiers for Dialogue Tutoring Agents: The Curious Case of LLMs as Your Coding Tutors (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing studies have focused on coding tutoring, but their capabilities in guiding users to solve complex tasks remain underexplored. |
| Approach: | They propose a novel agent workflow, Trace-and-Verify, which combines knowledge tracing to estimate a student’s knowledge state and turn-by-turn verification to ensure effective guidance toward task completion. |
| Outcome: | The proposed agent workflow achieves significantly higher success rates than existing tutoring agents. |
AIGuard: A Benchmark and Lightweight Detection for E-commerce AIGC Risks (2025.findings-acl)
Copied to clipboard
Wenhua Zhang, Weicheng Li, Xuanrong Rao, Lixin Zou, Xiangyang Luo, Chubin Zhuang, Yongjie Hong, Zhen Qin, Hengyu Chang, Chenliang Li, Bo Zheng
| Challenge: | Existing detection methods lack real-world scenarios and corresponding risk datasets . current MLLMs lack knowledge and have limited capability to detect the risk of AIGC content. |
| Approach: | They propose a benchmark for AIGC risk detection in real-world e-commerce . it includes 253,420 image-text pairs across four critical categories . |
| Outcome: | The proposed method achieves 9.68% higher recall than leading multimodal models while using only 25% of training resources. |
A2ATS: Retrieval-Based KV Cache Reduction via Windowed Rotary Position Embedding and Query-Aware Vector Quantization (2025.findings-acl)
Copied to clipboard
Junhui He, Junna Xing, Nan Wang, Rui Xu, Shangyu Wu, Peng Zhou, Qiang Liu, Chun Jason Xue, Qingan Li
| Challenge: | Long context large language models (LLMs) pose significant challenges for efficient serving due to the large memory footprint and high access overhead of KV cache. |
| Approach: | They propose a retrieval-based method to reduce the memory footprint of LLMs . they propose Windowed Rotary Position Embedding and query-aware vector quantization . |
| Outcome: | The proposed method can achieve lower performance degradation with lower overhead compared to existing methods . it can reduce the memory footprint and access overhead of long context large language models . |
TransBench: Breaking Barriers for Transferable Graphical User Interface Agents in Dynamic Digital Environments (2025.findings-acl)
Copied to clipboard
Yuheng Lu, Qian Yu, Hongru Wang, Zeming Liu, Wei Su, Yanping Liu, Yuhang Guo, Maocheng Liang, Yunhong Wang, Haifeng Wang
| Challenge: | Existing GUI agents struggle to adapt to dynamic and interconnected nature of real-world digital environments, authors show . |
| Approach: | They propose a benchmark to evaluate the transferability of GUI agents across three key dimensions . transBench includes 15 app categories with diverse functionalities . |
| Outcome: | The proposed benchmark shows that existing GUI agents struggle to adapt to dynamic, interconnected environments. |
Order Matters: Investigate the Position Bias in Multi-constraint Instruction Following (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing large language models struggle to follow multi-constraint instructions in real-world applications. |
| Approach: | They propose to quantify the difficulty distribution of constraints by a novel Difficulty Distribution Index (CDDI) they find that LLMs are more performant when presented with constraints in a “hard-to-easy” order. |
| Outcome: | The proposed model is more performant when presented with constraints in a “hard-to-easy” order, compared with existing models with different architectures and sizes of parameters. |
CoT-VTM: Visual-to-Music Generation with Chain-of-Thought Reasoning (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for visual-to-music generation lack large-scale, high-quality visual-music paired datasets and lack of direct semantic correspondence between visuals and music. |
| Approach: | They propose a framework that distills Chain-of-Thought reasoning to enable visual-to-music generation without paired data. |
| Outcome: | The proposed framework achieves optimal performance on image-to-music and video-to music tasks. |
A Tale of Evaluating Factual Consistency: Case Study on Long Document Summarization Evaluation (2025.findings-acl)
Copied to clipboard
| Challenge: | Despite the recent progress for summarization models in producing fluent summaries, they still encounter challenges when long sequences of generated texts and inputs (over thousands of words) need to be evaluated. |
| Approach: | They conduct a systematic analysis of factual-consistency evaluation systems across four long-document datasets and examine the relationship between sentence-level and summary-level model performance. |
| Outcome: | The proposed models can achieve higher recall in error detection for older summaries, yet struggle with false positives and fine-grained error detection. |
Evaluating Pretrained Causal Language Models for Synonymy (2025.findings-acl)
Copied to clipboard
| Challenge: | Despite the scaling of causal language models, the underlying basis of complex skills remains unclear. |
| Approach: | They propose that subjacent skills such as synonymy might be explained using linguistic concepts. |
| Outcome: | The proposed model recognizes synonymy but struggles to generate synonyms when prompted with relevant context. |
MDIT-Bench: Evaluating the Dual-Implicit Toxicity in Large Multimodal Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Multimodal Models (LMMs) have raised concerns about model toxicity. |
| Approach: | They propose a model to measure the toxicity gap between models and their hard level to determine whether they can handle dual-implicit toxicity. |
| Outcome: | The proposed model can handle dual-implicit toxicity effectively on 13 prominent LMMs, but its performance drops significantly in hard level. |
CoVE: Compressed Vocabulary Expansion Makes Better LLM-based Recommender Systems (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing approaches to align LLMs with recommendation tasks do not fully leverage their sequential information processing capabilities. |
| Approach: | They propose a system that allows users to expand their vocabulary by assigning a unique ID to each item within the expanded vocabulary. |
| Outcome: | The proposed system maximizes the sequence understanding abilities of large language models, significantly enhancing their performance on recommendation tasks. |
CtrlA: Adaptive Retrieval-Augmented Generation via Inherent Control (2025.findings-acl)
Copied to clipboard
Liu Huanshuo, Hao Zhang, Zhijiang Guo, Jing Wang, Kuicai Dong, Xiangyang Li, Yi Quan Lee, Cong Zhang, Yong Liu
| Challenge: | Existing methods focus on detecting LLM’s confidence via statistical uncertainty. |
| Approach: | They propose to use a representation perspective to solve adaptive RAG by enabling dynamic retrieval during generation and enabling retrieval only when the query exceeds LLM's internal knowledge. |
| Outcome: | The proposed framework is superior to existing adaptive RAG methods on a diverse set of tasks. |
Maximum Score Routing For Mixture-of-Experts (2025.findings-acl)
Copied to clipboard
| Challenge: | Traditional mixture-of-experts (MoE) networks impose an expert capacity constraint to ensure GPU-friendly computation. |
| Approach: | They propose a routing paradigm that dynamically allocates input tokens to top-k experts through differentiable sparse transformations, enabling scalable model capacity while preserving computational efficiency. |
| Outcome: | The proposed model achieves lower training losses and higher evaluation scores at equivalent FLOPs compared to constrained and unconstrained baselines. |
Time Course MechInterp: Analyzing the Evolution of Components and Knowledge in Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models acquire and store factual knowledge for interpretability, reliability, efficiency . prior work on factual recall focused on localizing knowledge within transformer parameters . |
| Approach: | They analyze the evolution of factual knowledge representation in a large language model by tracking its attention heads and feed forward networks over training. |
| Outcome: | The proposed model acquires and stores factual knowledge over time and is adaptively trained . the proposed model can be pruned, optimized, and transparent . |
Well Begun is Half Done: Low-resource Preference Alignment by Weak-to-Strong Decoding (2025.findings-acl)
Copied to clipboard
| Challenge: | Low-resource methods for LLM alignment have been popular, but still face challenges in obtaining high-quality and aligned content. |
| Approach: | They propose a framework to enhance alignment ability of base models by the guidance of a small aligned model. |
| Outcome: | The proposed framework outperforms baseline methods while avoiding degradation on downstream tasks. |
Disentangling Text and Math in Word Problems: Evidence for the Bidimensional Structure of Large Language Models’ Reasoning (2025.findings-acl)
Copied to clipboard
Pedro Calais, Gabriel Franco, Zilu Tang, Themistoklis Nikas, Wagner Meira Jr., Evimaria Terzi, Mark Crovella
| Challenge: | Existing studies show that LLMs struggle with text interpretation and equation solving, despite distinct proficiencies in textual and mathematical components. |
| Approach: | They disentangle textual interpretation and mathematical solving steps in word problems drawn from Brazil's largest college entrance exam and popular grade school-level benchmark GSM8K. |
| Outcome: | The proposed model outperforms LLMs in Brazil's largest college entrance exam and popular grade school-level benchmark. |
Human-LLM Coevolution: Evidence from Academic Writing (2025.findings-acl)
Copied to clipboard
| Challenge: | a statistical analysis of arXiv paper abstracts shows a marked drop in the frequency of several words previously identified as overused by ChatGPT, such as “delve”, starting soon after they were pointed out in early 2024. |
| Approach: | They report a drop in the frequency of several words previously identified as overused by ChatGPT, such as “delve”, starting soon after they were pointed out in early 2024. |
| Outcome: | The frequency of words previously identified as overused by ChatGPT, such as “delve”, has instead kept increasing. |
Disentangled Multi-span Evolutionary Network against Temporal Knowledge Graph Reasoning (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for temporal knowledge Graphs neglect internal structural interactions between subgraphs and ignore potential smooth features that do not lead to semantic changes. |
| Approach: | They propose to use a disentangled multi-span evolutionary network to capture local neighbor features while perceiving historical neighbor semantic information. |
| Outcome: | Extensive experiments show that the proposed model outperforms the state-of-the-art in TKG reasoning by 22.7%. |
GRAF: Graph Retrieval Augmented by Facts for Romanian Legal Multi-Choice Question Answering (2025.findings-acl)
Copied to clipboard
| Challenge: | Question answering systems have been used for various domains and languages. |
| Approach: | They propose a novel approach for question answering (QA) that combines a dataset of Romanian legal questions with a CROL corpus of laws. |
| Outcome: | The proposed approach achieves competitive results with generally accepted state-of-the-art methods and even exceeds them in most settings. |
Express What You See: Can Multimodal LLMs Decode Visual Ciphers with Intuitive Semiosis Comprehension? (2025.findings-acl)
Copied to clipboard
| Challenge: | Traditional VQA benchmarks encounter a modality gap and over-reliance on language priors, whereas human cognition excels at intuitive semiosis, associating abstract visual symbols to linguistic semantics. |
| Approach: | They propose a task of generating abstract linguistics from emoji sequence images, where such reasoning underpins critical applications in cryptography. |
| Outcome: | The proposed model can generate abstract linguistics from emoji sequence images, challenging MLLMs’ reasoning of decoding complex semantics of visual ciphers. |
ConFit v2: Improving Resume-Job Matching using Hypothetical Resume Embedding and Runner-Up Hard-Negative Mining (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods to model resume-job fit are sparse since job seekers apply to only a few jobs. |
| Approach: | They propose two techniques to enhance the encoder’s contrastive training process by augmenting job data with hypothetical reference resume generated by a large language model and creating high-quality hard negatives from unlabeled resume/job pairs using a novel hard-negative mining strategy. |
| Outcome: | The proposed method outperforms ConFit and prior methods on two real-world datasets and achieves an average improvement of 13.8% in recall and 17.5% in nDCG across job-ranking and resume-ranker tasks. |
Knowing Before Saying: LLM Representations Encode Information About Chain-of-Thought Success Before Completion (2025.findings-acl)
Copied to clipboard
| Challenge: | Using later reasoning steps does not always improve classification, suggesting LLMs encode key information early. |
| Approach: | They propose a method to predict the success of a zero-shot Chain-of-Thought process by using LLM representations that are based on initial steps representations. |
| Outcome: | The proposed method performs well even before a single token is generated, suggesting that crucial information about the reasoning process is already present in the initial steps representations. |
Grounding Task Assistance with Multimodal Cues from a Single Demonstration (2025.findings-acl)
Copied to clipboard
Gabriel Herbert Sarch, Balasaravanan Thoravi Kumaravel, Sahithya Ravi, Vibhav Vineet, Andrew D Wilson
| Challenge: | RGB video often fails to capture fine-grained contextual cues such as intent, safety-critical environmental factors, and subtle preferences embedded in human behavior. |
| Approach: | They propose a framework that integrates eye gaze and speech cues to improve conversational agents for task assistance by integrating eye gaze with speech cuests. |
| Outcome: | The proposed framework captures fine-grained intent and user-specific cues, enabling richer contextual grounding for visual question answering. |
Awes, Laws, and Flaws From Today’s LLM Research (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) are a powerful technology that can follow instructions and output coherent, persuasive text. |
| Approach: | They examine the scientific methodology behind large language model (LLM) research and cross-validate it with arguments at the centre of controversy. |
| Outcome: | The authors cross-validate 2,000 research works released between 2020 and 2024 based on criteria typical of what is considered good research and find that conference checklists are effective at curtailing some of these issues, but balancing velocity and rigour in research cannot solely rely on these. |
Dual Debiasing for Noisy In-Context Learning for Text Generation (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods detect noisy annotations by ranking local perplexities, but this assumption breaks down when the noise ratio is high and many demonstrations are flawed. |
| Approach: | They propose a method that uses synthesized neighbors to explicitly correct perplexity estimates, yielding a robust Sample Cleanliness Score. |
| Outcome: | The proposed method is comparable to a fully clean demonstration corpus and performs well even with noise ratios as high as 0.8. |
DRS: Deep Question Reformulation With Structured Output (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing models like GPT-3 and Instruct-GPT lack the ability to reformulate unanswerable questions. |
| Approach: | They propose a zero-shot method that combines the strengths of LLMs with a DFS-based algorithm to iteratively explore potential entity combinations and constrain outputs using predefined entities. |
| Outcome: | The proposed method outperforms all baselines, including the GPT-3.5 model, on the unanswerable question reformulation task. |
Towards Explainable Hate Speech Detection (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent advances in deep learning have significantly enhanced the efficiency and accuracy of natural language processing (NLP) tasks. |
| Approach: | They propose a model that uses valence, arousal, and dominance (VAD) scores to detect hate speech and a weighted sum of valent, valance, and valency (VA) scores for classification. |
| Outcome: | The proposed model can compete with state-of-the-art models in detecting hate speech and non-hate speech words based on their individual and summed VAD-values. |
BioHopR: A Benchmark for Multi-Hop, Multi-Answer Reasoning in Biomedical Domain (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing benchmarks for multi-hop reasoning in biomedical domain are lacking . bioHopR provides benchmarks to evaluate multi-step reasoning in structured biomedic knowledge graphs . |
| Approach: | They propose a benchmark to evaluate multi-hop, multi-answer reasoning in biomedical knowledge graphs. |
| Outcome: | BioHopR evaluates multi-hop reasoning in biomedical knowledge graphs based on the PrimeKG model . it outperforms proprietary models and open-source biomedal models in 1-hop and 2-hop tasks . |
PipeSpec: Breaking Stage Dependencies in Hierarchical LLM Decoding (2025.findings-acl)
Copied to clipboard
| Challenge: | Speculative decoding is limited by sequential stage dependencies that prevent full hardware utilization. |
| Approach: | They propose a framework that generalizes speculative decoding to use multiple models arranged in a hierarchical pipeline and enables asynchronous execution with lightweight coordination for prediction verification and rollback. |
| Outcome: | The proposed framework achieves 2.25 tokens/unit through pipelined parallelism with multiple models arranged in a hierarchical pipeline. |
LAM SIMULATOR: Advancing Data Generation for Large Action Model Training via Online Exploration and Trajectory Feedback (2025.findings-acl)
Copied to clipboard
Thai Quoc Hoang, Kung-Hsiang Huang, Shirley Kokane, Jianguo Zhang, Zuxin Liu, Ming Zhu, Jake Grigsby, Tian Lan, Michael S Ryoo, Chien-Sheng Wu, Shelby Heinecke, Huan Wang, Silvio Savarese, Caiming Xiong, Juan Carlos Niebles
| Challenge: | Large Action Models (LAMs) face challenges due to the need for high-quality training data, especially for multi-steps tasks that involve planning, executing tool calls, and responding to feedback. |
| Approach: | They propose a framework for online exploration of agentic tasks with high-quality feedback . they use a dynamic task query generator and an extensive collection of tools to create a high-level feedback environment for LLM Agents. |
| Outcome: | The proposed framework achieves 49.3% performance improvement over baselines on toolbench and CRMArena. |
Rank, Chunk and Expand: Lineage-Oriented Reasoning for Taxonomy Expansion (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing taxonomy expansion methods struggle with representation limits and generalization, while generative methods process all candidates at once, introducing noise and exceeding context limits. |
| Approach: | They propose a plug-and-play framework that combines discriminative ranking and generative reasoning for efficient taxonomy expansion. |
| Outcome: | Experiments show that LORex improves accuracy by 12% and similarity by 5% over state-of-the-art methods. |
Probing Subphonemes in Morphology Models (2025.findings-acl)
Copied to clipboard
| Challenge: | morphological inflection models have achieved state-of-the-art performance, yet their ability to generalize across languages and morphology rules remains limited. |
| Approach: | They propose a language-agnostic probing method to investigate phonological feature encoding in transformers trained directly on phonemes and perform it across seven morphologically diverse languages. |
| Outcome: | The proposed method shows that phonological features which are local are captured well in phoneme embeddings, whereas long-distance dependencies like vowel harmony are better represented in the transformer’s encoder. |
Exploiting Instruction-Following Retrievers for Malicious Information Retrieval (2025.findings-acl)
Copied to clipboard
| Challenge: | Instruction-following retrievers are increasingly used in real-world applications, but little research has investigated the safety risks associated with their increasing search capabilities. |
| Approach: | They investigate the ability of retrievers to satisfy malicious queries . they find that for >50% of queries, retrievers can select harmful passages . |
| Outcome: | The findings highlight the safety risks associated with instruction-following retrievers . they show that even safety-aligned LLMs can satisfy malicious requests . |
Improving Causal Interventions in Amnesic Probing with Mean Projection or LEACE (2025.findings-acl)
Copied to clipboard
| Challenge: | Amnesic probing examines the influence of specific linguistic information on the behaviour of a model by identifying and removing a property and then testing whether the model can still perform well on next-word prediction. |
| Approach: | They propose to use iterative nullspace projection to remove information by iterating on the target property and then assessing whether the model's performance changes. |
| Outcome: | The proposed methods remove information in a more targeted manner, thereby enhancing the potential for obtaining behavioural explanations through Amnesic Probing. |
The Threat of PROMPTS in Large Language Models: A System and User Prompt Perspective (2025.findings-acl)
Copied to clipboard
| Challenge: | Prompts are essential for guiding model output and influencing content generation. |
| Approach: | They propose to attack models with prompt leakage and prompt jailbreak attacks . they summarize the experimental setups of these methods and explore the relationship between prompt threats and prompt injection attacks. |
| Outcome: | The proposed methods summarize the experimental setups and examine the relationship between prompt threats and prompt injection attacks. |
RoseRAG: Robust Retrieval-augmented Generation with Small-scale LLMs via Margin-aware Preference Optimization (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) have impressive performance but face high computational costs and latency, limiting their deployment in resource-constrained settings. |
| Approach: | They propose a robust RAG framework for large language models via Margin-aware Preference Optimization to enhance the accuracy and reliability of SLMs. |
| Outcome: | The proposed framework surpasses state-of-the-art benchmarks on three open-domain question answering tasks. |
Instruction-Tuning LLMs for Event Extraction with Annotation Guidelines (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing applications of large language models to IE can be categorized into two lines: prompt engineering-based approaches and instruction-tuning open-weight LLMs. |
| Approach: | They propose to use annotation guidelines to teach large language models for event extraction . they use textual descriptions of event types and arguments to train the models . |
| Outcome: | The proposed approach improves cross-schema generalization and low-frequency event-type performance when there is a decent amount of training data. |
mRAKL: Multilingual Retrieval-Augmented Knowledge Graph Construction for Low-Resourced Languages (2025.findings-acl)
Copied to clipboard
| Challenge: | Knowledge Graphs are structured multirelational graphs that store factual knowledge. |
| Approach: | They introduce a Retrieval-Augmented Generation (mRAKL) based system to perform mKGC. |
| Outcome: | The proposed approach improves over a no-context setting with an idealized retrieval system. |
Mechanistic Interpretability of Emotion Inference in Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing studies on large language models (LLMs) show promising capabilities in predicting human emotions from text. |
| Approach: | They investigate how autoregressive LLMs infer emotions by focusing on appraisal theory . they show that emotion representations are functionally localized to specific regions in the model . |
| Outcome: | The proposed model is functionally localized to specific regions in the model, and the results align with theoretical and intuitive expectations. |
RL-Guider: Leveraging Historical Decisions and Feedback for Drug Editing with Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | RLs can be used to refine drugs by iterative conversations with domain experts . existing methods do not leverage past knowledge, but human experts develop intuition over time through historical experience . |
| Approach: | They propose a reinforcement-learning agent to provide suggestions to large language models . RL-Guider leverages the “world-level” knowledge of LLMs and historical feedback . |
| Outcome: | a new reinforcement-learning agent improves the performance of large language models . the proposed agent leverages the “world-level” knowledge of LLMs and historical feedback . |
BriefMe: A Legal NLP Benchmark for Assisting with Legal Briefs (2025.findings-acl)
Copied to clipboard
| Challenge: | a core part of legal work that has been underexplored in Legal NLP is the writing and editing of legal briefs. |
| Approach: | They propose to use large language models to help legal professionals with writing briefs by capturing and evaluating their abilities in language models. |
| Outcome: | The proposed tasks show that the models perform well on arguments summarization, argument completion, and case retrieval tasks. |
I see what you mean: Co-Speech Gestures for Reference Resolution in Multimodal Dialogue (2025.findings-acl)
Copied to clipboard
| Challenge: | Using representational co-speech gestures, face-to-face interaction participants resolve references to objects using speech and gestures. |
| Approach: | They propose a multimodal reference resolution task centred on representational gestures . they propose 'self-supervised' pre-training approach to gesture representation learning that grounds body movements in spoken language. |
| Outcome: | The proposed approach aligns with expert annotations and has significant predictive power. |
World Knowledge Resolves Some Aspectual Ambiguity (2025.findings-acl)
Copied to clipboard
| Challenge: | Annotating event descriptions with aspectual features is often seen as a pre-requisite to temporal reasoning, however, a recent study has shown that non-experts’ annotations of the aspectual class of English verb phrases can disagree with both expert linguistic annotations and each other. |
| Approach: | They hypothesized that people use their world knowledge to tacitly conjure their own contexts, leading to disagreement between them. |
| Outcome: | The results show that the hypothesis explains some of the disagreement, but outputs from GPT-4 are not an accurate predictor of human answers. |
ACCESS DENIED INC: The First Benchmark Environment for Sensitivity Awareness (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models can process text from various document formats and facilitate user interactions through natural language queries. |
| Approach: | They propose a concept of sensitivity awareness (SA) which enables LLMs to adhere to predefined access rights rules. |
| Outcome: | The proposed model can adhere to predefined access rights rules while managing unauthorized requests while effectively addressing legitimate queries. |
Spatial Coordinates as a Cell Language: A Multi-Sentence Framework for Imaging Mass Cytometry Analysis (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing single-cell LLMs struggle to integrate spatial information into natural language, limiting their ability to capture biological relationships. |
| Approach: | They propose a framework that integrates both single-cell expression and spatial information into natural language using a multi-sentence approach. |
| Outcome: | The proposed framework outperforms existing single-cell LLMs on preprocessed IMC datasets for diabetes and brain tumors while improving interpretability. |
HumanEval Pro and MBPP Pro: Evaluating Large Language Models on Self-invoking Code Generation Task (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing benchmarks for code generation tasks are inadequate, but performance declines on self-invoking tasks. |
| Approach: | They propose a general recipe for generating more challenging versions of existing benchmarks . they propose to use instruction-tuned models to evaluate LLMs on self-invoking code generation tasks . |
| Outcome: | The proposed model improves on humanEval and MBPP but on self-invoking code generation tasks. |
TCSinger 2: Customizable Multilingual Zero-shot Singing Voice Synthesis (2025.findings-acl)
Copied to clipboard
Yu Zhang, Wenxiang Guo, Changhao Pan, Dongyu Yao, Zhiyuan Zhu, Ziyue Jiang, Yuhan Wang, Tao Jin, Zhou Zhao
| Challenge: | Existing zero-shot singing voice synthesis models depend on phoneme and note boundary annotations, limiting their robustness and producing poor transitions between phonemes and notes. |
| Approach: | They propose a multi-task multilingual zero-shot SVS model with style transfer and style control based on various prompts. |
| Outcome: | Experimental results show that TCSinger 2 outperforms baseline models in subjective and objective metrics across multiple related tasks. |
Compute Optimal Scaling of Skills: Knowledge vs Reasoning (2025.findings-acl)
Copied to clipboard
| Challenge: | Scaling laws are a critical component of the LLM development pipeline, but little is known about whether the COs of individual skills such as mathematical reasoning, question answering (QA) or coding, align with these APEs. |
| Approach: | They examine knowledge-based QA and code generation to find out whether skill-dependent scaling is an artefact of the pretraining datamix. |
| Outcome: | The proposed scaling laws are skill-dependent, and knowledge and code exhibit fundamental differences in scaling behaviour when corrected for datamix differences. |
PECAN: LLM-Guided Dynamic Progress Control with Attention-Guided Hierarchical Weighted Graph for Long-Document QA (2025.findings-acl)
Copied to clipboard
| Challenge: | Long-document Question Answering (QA) challenges with large-scale text and long-distance dependencies. |
| Approach: | They propose a method that leverages large language models to control retrieval process . they propose 'attention-based' retrieval methods that construct hierarchical graphs . |
| Outcome: | The proposed method achieves LLM-level performance while maintaining computational complexity comparable to RAG methods. |
Lifelong Model Editing with Graph-Based External Memory (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for post-training model editing suffer from overfitting and catastrophic forgetting. |
| Approach: | They propose a framework that leverages hyperbolic geometry and graph neural networks for precise and stable model edits. |
| Outcome: | Experiments on CounterFact, CounterFACT+, and MQuAKE with GPT2-XL and GPT-J show that HYPE significantly enhances edit stability, factual accuracy, and multi-hop reasoning. |
Multi-Sense Embeddings for Language Models and Knowledge Distillation (2025.findings-acl)
Copied to clipboard
| Challenge: | Transformer-based large language models generate different representations for the same token depending on context . however, words and tokens typically have only a limited number of senses . a knowledge distillation method can be used to learn a smaller student model . |
| Approach: | They propose a multi-sense embedding method that uses a clustering algorithm to generate a sense embeddable dictionary. |
| Outcome: | The proposed method offers significant space and inference time savings while maintaining competitive performance. |
CodeScientist: End-to-End Semi-Automated Scientific Discovery with Code-based Experimentation (2025.findings-acl)
Copied to clipboard
Peter Jansen, Oyvind Tafjord, Marissa Radensky, Pao Siangliulue, Tom Hope, Bhavana Dalvi Mishra, Bodhisattwa Prasad Majumder, Daniel S Weld, Peter Clark
| Challenge: | Automated scientific discovery (ASD) systems are limited in their evaluation of software artifacts and large volumes of research artifs are typically evaluated using conference-style paper review with limited evaluation of code. |
| Approach: | They propose a novel ASD system that frames ideation and experiment construction as a form of genetic search jointly over combinations of research articles and codeblocks defining common actions in a domain. |
| Outcome: | The proposed system returns 19 discoveries on machine-generated ideas in the domain of agents and virtual environments. |
Beyond Factual Accuracy: Evaluating Coverage of Diverse Factual Information in Long-form Text Generation (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing evaluation frameworks for large language models focus on isolated aspects like * Equal contribution. |
| Approach: | They evaluate ICAT, an evaluation framework for measuring coverage of diverse factual information in long-form text generation. |
| Outcome: | The evaluation framework is based on three implementations with different assumptions on availability of aspects and alignment method. |
Continual Quantization-Aware Pre-Training: When to transition from 16-bit to 1.58-bit pre-training for BitNet language models? (2025.findings-acl)
Copied to clipboard
| Challenge: | Quantization-aware training of large language models reduces the precision of model parameters and reduces memory usage and energy consumption at inference time. |
| Approach: | They propose a method where models are first trained with 16-bit precision and then transition to 1.58-bit quantization-aware training. |
| Outcome: | The proposed training strategy reduces memory and energy consumption while maintaining model accuracy while reducing memory and inference time. |
When Detection Fails: The Power of Fine-Tuned Models to Generate Human-Like Social Media Text (2025.findings-acl)
Copied to clipboard
| Challenge: | detecting AI-generated text on social media is difficult due to short text length and informal language of the internet . a recent study shows that detection of AI-generated posts is difficult under assumptions that an attacker has no knowledge of the generating model . |
| Approach: | They use open-source, closed-source and fine-tuned social media to detect AI-generated text . they use assumptions about knowledge of and access to the generating models to test detection . |
| Outcome: | a human study shows that detection of AI-generated social media posts is difficult . the study compared 505,159 posts from open-source, closed-source and fine-tuned models . |
Not quite Sherlock Holmes: Language model predictions do not reliably differentiate impossible from improbable events (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing work has shown that language models can select the most likely or plausible of a set of possible events, but they are far from robust. |
| Approach: | They focus on whether language models can select the most likely or plausible of a set of possibilities and compare them to a broader behavior that humans exhibit largely unconsciously. |
| Outcome: | The proposed models perform worse than expected under certain conditions, compared with Llama 3, Gemma 2, and Mistral NeMo, and they are significantly more sensible than leaves. |
The Rotary Position Embedding May Cause Dimension Inefficiency in Attention Heads for Long-Distance Retrieval (2025.findings-acl)
Copied to clipboard
| Challenge: | We hypothesize that the wide range of rotation angles may prevent LLMs from utilizing certain dimensions. |
| Approach: | They propose to use the Rotary Position Embedding (RoPE) for long context modeling . they hypothesize that the wide range of rotation angles may prevent LLMs from utilizing those dimensions. |
| Outcome: | The proposed model may not be useful for long-context modeling. |
IDEA: Enhancing the Rule Learning Ability of Large Language Model Agent through Induction, Deduction, and Abduction (2025.findings-acl)
Copied to clipboard
| Challenge: | RULEARN is a benchmark to assess the rule-learning abilities of large language models (LLMs) in interactive environments. |
| Approach: | They propose a framework that integrates the process of **I**nduction, **De**duction, and **A**bduction. |
| Outcome: | The proposed framework improves on the baseline and human-like rule learning in real-world scenarios. |
EnigmaToM: Improve LLMs’ Theory-of-Mind Reasoning Capabilities with Neural Knowledge Base of Entity States (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing ToM reasoning methods rely excessively on off-the-shelf LLMs, reducing their efficiency and limiting their applicability to high-order ToM. |
| Approach: | They propose a neuro-symbolic framework that integrates a Neural Knowledge Base of Entity States and knowledge injection to enhance ToM reasoning. |
| Outcome: | The proposed framework improves ToM reasoning on ToMi, HiToM, and FANToM benchmarks. |
ReasonerRank: Redefining Language Model Evaluation with Ground-Truth-Free Ranking Frameworks (2025.findings-acl)
Copied to clipboard
Jiamu Zhang, Jiayi Yuan, Andrew Wen, Hoang Anh Duy Le, Yu-Neng Chuang, Soo-Hyun Choi, Rui Chen, Xia Hu
| Challenge: | Large Language Models (LLMs) are increasingly adopted across real-world applications . traditional evaluations rely on expensive, domain-specific ground-truth labels . obtaining labeled data is expensive, time-consuming, and often requires domain expertise . |
| Approach: | They propose a ground-truth-free evaluation framework focused on reasoning consistency and instruction following. |
| Outcome: | The proposed framework outperforms existing label-free methods, including majority voting, triplet ranking, and peer-review approaches. |
HyGenar: An LLM-Driven Hybrid Genetic Algorithm for Few-Shot Grammar Generation (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing grammar generation models perform sub-optimally, resulting in inconsistent syntactic and semantic accuracy. |
| Approach: | They propose an LLM-driven hybrid genetic algorithm to optimize grammar generation by inferring grammars from a set of examples and generated in Backus-Naur Form. |
| Outcome: | The proposed algorithm improves syntactic and semantic accuracy of generated grammars across LLMs. |
Can Large Language Models Understand Argument Schemes? (2025.findings-acl)
Copied to clipboard
| Challenge: | Argument schemes are stereotypical forms of reasoning that occur in everyday arguments. |
| Approach: | They propose to use large language models (LLMs) to classify argument schemes based on Walton’s taxonomy to employ formal definitions and LLM-generated descriptions to enhance task instructions. |
| Outcome: | The proposed models perform well on annotated and automatically generated arguments, and provide insights for advancing reasoning capabilities in computational argumentation. |
MMInA: Benchmarking Multihop Multimodal Internet Agents (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing benchmarks fail to assess embodied agents in a realistic, evolving environment for compositional Internet tasks. |
| Approach: | They propose a multihop and multimodal benchmark to evaluate embodied agents for compositional Internet tasks. |
| Outcome: | The proposed protocol significantly improves the performance of both the single-hop and multihop web browsing abilities. |
ThinkGuard: Deliberative Slow Thinking Leads to Cautious Guardrails (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing guardrails rely on rule-based filtering or single-pass classification, limiting their ability to handle nuanced safety violations. |
| Approach: | They propose a critique-augmented guardrail model that distills knowledge from high-capacity LLMs by generating structured critiques alongside safety labels. |
| Outcome: | The proposed model outperforms existing guardrail models on multiple safety benchmarks and achieves the highest average F1 and AUPRC. |
Neutralizing Bias in LLM Reasoning using Entailment Graphs (2025.findings-acl)
Copied to clipboard
| Challenge: | Natural Language Inference (NLI) is a foundational understanding task in language understanding. |
| Approach: | They propose a framework to construct counterfactual reasoning data and fine-tune LLMs to reduce attestation bias. |
| Outcome: | The proposed framework reduces hallucinations from attestation bias on original and bias-neutralized datasets while keeping hypotheses unchanged. |
Dynamic Steering With Episodic Memory For Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing activation steering methods apply a single sentence-level steering vector uniformly across all tokens, ignoring LLMs’ token-wise, auto-regressive nature. |
| Approach: | They propose a framework that aligns LLMs to given demonstrations by steering at the token level conditioned on the input query. |
| Outcome: | The proposed framework surpasses baselines across safety, style transfer, and role-playing tasks, demonstrating improved alignment as demonstration scales. |
Eeyore: Realistic Depression Simulation via Expert-in-the-Loop Supervised and Preference Optimization (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have been explored for mental healthcare training and therapy client simulation, but they fail to authentically capture diverse client traits and psychological conditions. |
| Approach: | They propose an 8B model optimized for realistic depression simulation with expert input at every stage. |
| Outcome: | The model outperforms GPT-4o in linguistic authenticity and profile adherence. |
Lost in Translation: Benchmarking Commercial Machine Translation Models for Dyslexic-Style Text (2025.findings-acl)
Copied to clipboard
| Challenge: | Dyslexia affects writing, leading to unique patterns such as letter and homophone swapping. |
| Approach: | They examine the fairness of four commercial machine translation systems towards dyslexic text through a systematic audit using both synthetically generated and real writing from individuals with dyslexia. |
| Outcome: | The proposed system audits show that it is fair to use synthetic and synthetic dyslexic text and real writing from people with dyslexia. |
Divide-Verify-Refine: Can LLMs Self-align with Complex Instructions? (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing research shows LLMs struggle with complex instructions involving multiple constraints. |
| Approach: | They propose a framework to divide complex instructions into single constraints and prepare appropriate tools to verify responses. |
| Outcome: | The proposed framework doubles Llama3.1-8B’s constraint adherence and triples Mistral-7B’ s performance. |
LlamaPIE: Proactive In-Ear Conversation Assistants (2025.findings-acl)
Copied to clipboard
| Challenge: | LlamaPIE is the first real-time proactive assistant designed to enhance human conversations . it provides discreet, concise guidance delivered via hearable devices . traditional language models require explicit user invocation, but the assistant operates in the background . |
| Approach: | They propose a two-model pipeline that decides when to respond and a larger model that generates the response. |
| Outcome: | The proposed approach is effective in providing helpful, unobtrusive assistance on real-world datasets. |
Task-Oriented Automatic Fact-Checking with Frame-Semantics (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing work on automatic fact-checking relies on unstructured data and large language models to produce fact- check verdicts and explanations. |
| Approach: | They propose a new paradigm for automatic fact-checking that leverages frame semantics to enhance the structured understanding of claims and guide the process of fact- checking them. |
| Outcome: | The proposed paradigm improves evidence retrieval and explainability for fact-checking by leveraging frame semantics. |
Craw4LLM: Efficient Web Crawling for LLM Pretraining (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing work discards over 90% of the raw data collected from web crawls, highlighting the inefficiency of current web crawlers in collecting LLM pretraining data. |
| Approach: | They propose a web crawling method that leverages the preference of LLMs as the priority score of the web crawler’s scheduler to obtain high-quality pretraining data. |
| Outcome: | The proposed method achieves high-quality pretraining data on a web graph containing 900 million webpages from a commercial search engine's index with just 21% URLs crawled. |
Be Cautious When Merging Unfamiliar LLMs: A Phishing Model Capable of Stealing Privacy (2025.findings-acl)
Copied to clipboard
| Challenge: | Model merging is a widespread technology in large language models that integrates multiple task-specific LLMs into a unified one. |
| Approach: | They propose a model merging approach that trains a phishing model capable of stealing privacy using a privacy phish instruction dataset. |
| Outcome: | The proposed model cloaking method mimics a specialized capability to conceal attack intent, luring users into merging the phishing model. |
Understand User Opinions of Large Language Models via LLM-Powered In-the-Moment User Experience Interviews (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing large language models (LLMs) are difficult to evaluate and often lack the ability to capture user opinions. |
| Approach: | They propose an LLM-powered interviewer that conducts in-the-moment user experience interviews right after users interact with LLMs and automatically gathers insights about user opinions from massive interview logs. |
| Outcome: | The proposed interviewer captures interesting user opinions, e.g., bipolar views on the displayed reasoning process of DeepSeek-R1 and demands for information freshness and multi-modality. |
HiCOT: Improving Neural Topic Models via Optimal Transport and Contrastive Learning (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent advances in neural topic models (NTMs) have improved topic quality but still face challenges: weak document-topic alignment, high inference costs due to large pretrained language models, and limited modeling of hierarchical topic structures. |
| Approach: | They propose a framework that integrates hierarchical clustering and contrastive learning to refine document-topic relationships using compact PLM-based embeddings. |
| Outcome: | The proposed framework improves topic coherence, topic performance, representation quality and computational efficiency over existing NTMs. |
FLAG-TRADER: Fusion LLM-Agent with Gradient-based Reinforcement Learning for Financial Trading (2025.findings-acl)
Copied to clipboard
Guojun Xiong, Zhiyang Deng, Keyi Wang, Yupeng Cao, Haohang Li, Yangyang Yu, Xueqing Peng, Mingquan Lin, Kaleb E Smith, Xiao-Yang Liu, Jimin Huang, Sophia Ananiadou, Qianqian Xie
| Challenge: | Large language models (LLMs) have impressive reasoning capabilities in financial tasks, but struggle with multi-step, goal-oriented scenarios in interactive financial markets. |
| Approach: | They propose a framework that integrates large language models with gradient-driven reinforcement learning (RL) policy optimization. |
| Outcome: | The proposed framework improves performance in trading and other financial domain tasks. |
The Silent Saboteur: Imperceptible Adversarial Attacks against Black-Box Retrieval-Augmented Generation Systems (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing studies have focused on corpus poisoning, but there are no studies on adversarial attacks on RAG systems. |
| Approach: | They propose a novel imperceptible retrieve-to-generate attack against RAG systems . they propose regenerative reinforcement learning framework that tracks interactions between attacker and target RAG . |
| Outcome: | The proposed framework outperforms existing attacks on factual and non-factual RAG systems with small imperceptible text perturbations. |
CROSSAGENTIE: Cross-Type and Cross-Task Multi-Agent LLM Collaboration for Zero-Shot Information Extraction (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models struggle with producing structured output while maintaining accuracy in zero-shot information extraction (IE) |
| Approach: | They propose a multi-agent framework that enhances zero-shot IE through multi-task collaboration. |
| Outcome: | CROSSAGENTIE outperforms state-of-the-art models in structured prediction . the framework significantly reduces inference cost while preserving accuracy . |
Decoupling Memories, Muting Neurons: Towards Practical Machine Unlearning for Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for MU degrade model utility, especially when accessing the original training data. |
| Approach: | They propose a method that eliminates the influence of unlearned data by modulating the outputs of merely 1% of the neurons in the feed-forward network modules within the Transformer blocks. |
| Outcome: | The proposed method eliminates the influence of unlearned data from Large Language Models by modulating the outputs of 1% of the neurons in the feed-forward network modules within the Transformer blocks, minimizing disruption to the model’s performance. |
Assimilation and Accommodation: Task-Adaptive Hierarchical Abstraction for Solving Web Tasks (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods focus on summarizing workflows, i.e., common sub-routines, which introduce excessive low-level details that distract models. |
| Approach: | They propose a framework that derives task-adaptive hierarchical abstraction from experience to enhance web task reasoning. |
| Outcome: | The proposed framework improves performance with competitive cost-efficiency on Mind2web and Webarena. |
SafeLawBench: Towards Safe Alignment of Large Language Models (2025.findings-acl)
Copied to clipboard
Chuxue Cao, Han Zhu, Jiaming Ji, Qichao Sun, Zhenghao Zhu, Wu Yinyu, Josef Dai, Yaodong Yang, Sirui Han, Yike Guo
| Challenge: | Recent studies indicate that large language models (LLMs) may exhibit risks, including threats to the protection of private data and the generation of hallucinations. |
| Approach: | They propose to evaluate LLMs from a legal perspective using the SafeLawBench benchmark. |
| Outcome: | The proposed framework categorizes safety risks into three levels based on legal standards and includes 24,860 multi-choice questions and 1,106 open-domain question-answering tasks. |
3DM: Distill, Dynamic Drop, and Merge for Debiasing Multi-modal Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent advances in Multi-modal Language Models have shown remarkable performance in multimodal tasks . however, these models often exhibit inherent biases that compromise their reliability and fairness. |
| Approach: | They propose a framework that integrates Distill, Dynamic Drop, and Merge to address these challenges. |
| Outcome: | The proposed framework outperforms existing methods in balancing debiasing and improving performance on the MMSD2.0 sarcasm detection dataset. |
CausalAbstain: Enhancing Multilingual LLMs with Causal Reasoning for Trustworthy Abstention (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods to reduce hallucinations in large language models are inaccurate and inaccuracies in the generated feedback. |
| Approach: | They propose a method that helps LLMs determine whether to utilize multiple generated feedback responses and how to identify the most useful ones. |
| Outcome: | Extensive experiments show that the proposed method outperforms baselines on encyclopedic and commonsense knowledge QA tasks. |
CapArena: Benchmarking and Analyzing Detailed Image Captioning in the LLM Era (2025.findings-acl)
Copied to clipboard
Kanzhi Cheng, Wenpo Song, Jiaxin Fan, Zheng Ma, Qiushi Sun, Fangzhi Xu, Chenyang Yan, Nuo Chen, Jianbing Zhang, Jiajun Chen
| Challenge: | Image captioning has been a challenge for vision-language researchers for decades . current VLMs focus on tasks like visual question answering (YA) but image captioning is not as advanced as expected. |
| Approach: | They evaluate VLMs' performance on image captioning using human annotations . they find that some metrics show high caption-level agreement with humans . |
| Outcome: | The proposed model outperforms open-source models on image captioning . it achieves 93.4% correlation with human rankings at $4 per test . |
LLM-Empowered Class Imbalanced Graph Prompt Learning for Online Drug Trafficking Detection (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing approaches to combat illicit drug trafficking are impractical due to the scarcity of labeled samples and imbalance of classes. |
| Approach: | They propose a Large Language Model-empowered Heterogeneous Graph Prompt Learning framework for illicit drug trafficking detection that leverages LLM to facilitate heterogeneous graph neural networks to effectively identify minority classes. |
| Outcome: | The proposed framework is able to identify minority classes in class-imbalanced scenarios. |
CoLA: Collaborative Low-Rank Adaptation (2025.findings-acl)
Copied to clipboard
| Challenge: | The scaling law of Large Language Models (LLMs) reveals diminishing return on performance as model scale increases. |
| Approach: | They propose a more flexible LoRA architecture with an efficient initialization scheme . they propose combining three collaborative strategies to enhance performance . |
| Outcome: | The proposed model outperforms existing methods in low-sample scenarios. |
GLiM: Integrating Graph Transformer and LLM for Document-Level Biomedical Relation Extraction with Incomplete Labeling (2025.findings-acl)
Copied to clipboard
Hao Fang, Yuejie Zhang, Rui Feng, Yingwen Wang, Qing Wang, Wen He, Xiaobo Zhang, Tao Zhang, Shang Gao
| Challenge: | Document-level relation extraction (DocRE) solves problems of document quality . number of entities and entity-pair relations increases, causing incomplete annotations . |
| Approach: | a framework that reduces the problem space using a graph-enhanced Transformer-based model is proposed . GLiM leverages large language models for reasoning to reduce the problem-space . |
| Outcome: | GLiM boosts average recall and F1 scores on biomedical datasets . compared with existing models, GLim outperforms existing models on biomedicine benchmarks compared to existing models . |
AnalyticKWS: Towards Exemplar-Free Analytic Class Incremental Learning for Small-footprint Keyword Spotting (2025.findings-acl)
Copied to clipboard
| Challenge: | Keyword spotting (KWS) is a useful mechanism to identify spoken commands in voice-enabled systems, but catastrophic forgetting is causing models to lose their ability to recognize earlier keywords. |
| Approach: | They propose an exemplar-free method that updates model parameters without revisiting earlier data. |
| Outcome: | The proposed method outperforms existing continual learning methods on a variety of datasets and settings. |
Sleepless Nights, Sugary Days: Creating Synthetic Users with Health Conditions for Realistic Coaching Agent Interactions (2025.findings-acl)
Copied to clipboard
Taedong Yun, Eric Yang, Mustafa Safdari, Jong Ha Lee, Vaishnavi Vinod Kumar, S. Sara Mahdavi, Jonathan Amar, Derek Peyton, Reut Aharony, Andreas Michaelides PhD, Logan Douglas Schneider, Isaac Galatzer-Levy, Yugang Jia, John Canny, Arthur Gretton, Maja Mataric
| Challenge: | Structured data is generated grounded in health and lifestyle factors and full profiles of synthetic users are developed conditioned on the structured data. |
| Approach: | They propose an end-to-end framework for generating synthetic users for evaluating interactive agents designed to encourage positive behavior changes, such as in health and lifestyle coaching. |
| Outcome: | The proposed framework is validated in the domains of sleep and diabetes coaching using two independently-developed agents for sleep and diabetic coaching as case studies. |
Imagine to Hear: Auditory Knowledge Generation can be an Effective Assistant for Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing approaches to augment language models with audio databases lack the ability to understand auditory signals like humans. |
| Approach: | They propose a method that augments language models with audio databases to generate auditory knowledge . their framework detects multiple audio-related textual spans from the given prompt . |
| Outcome: | The proposed approach achieves state-of-the-art performance on AuditoryBench without external databases. |
SafeEraser: Enhancing Safety in Multimodal Large Language Models through Multimodal Machine Unlearning (2025.findings-acl)
Copied to clipboard
Junkai Chen, Zhijie Deng, Kening Zheng, Yibo Yan, Shuliang Liu, PeiJun Wu, Peijie Jiang, Jia Liu, Xuming Hu
| Challenge: | Existing methods for MU forget quality and model utility are not fully explored for safety in MLLMs. |
| Approach: | They propose a safety unlearning benchmark for MLLMs to measure over-forgetting . they propose MU methods to forget quality and model utility . |
| Outcome: | The proposed method reduces over-forgetting by 79.5% while maintaining forget quality and model utility. |
Prediction-Augmented Generation for Automatic Diagnosis Tasks (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) adopt autoregressive architecture, predicting the next word token based on the preceding context. |
| Approach: | They propose a method that integrates task-specific predictive models as external tools to improve model generation quality and accuracy. |
| Outcome: | The proposed method improves the generation quality and predictive accuracy of large language models in inference-driven tasks. |
FedLEKE: Federated Locate-then-Edit Knowledge Editing for Multi-Client Collaboration (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for updating large language models are inefficient in multi-client scenarios . Existing approaches assume a single-user setting and are ineffective in multiclient scenarios. |
| Approach: | They propose a new task that enables multiple clients to perform LEKE while preserving privacy and reducing computational overhead. |
| Outcome: | The proposed framework outperforms existing LEKE frameworks on two benchmark datasets and retains 96% of performance. |
DiSCo: Device-Server Collaborative LLM-based Text Streaming Services (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) have introduced significant cost and quality of experience (QoE) challenges in serving millions of daily requests. |
| Approach: | They propose a device-server cooperative scheduler that optimizes users’ QoE by adaptively routing requests and migrating response generation between endpoints while maintaining cost constraints. |
| Outcome: | Evaluations on real-world workloads show that the proposed scheduler can reduce tail TTFT (11-52%) and mean TTTT (6-78%) while maintaining comparable QoE levels. |
Customizing In-context Learning for Dynamic Interest Adaption in LLM-based Recommendation (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing Large Language Model (LLM)-based recommender systems face challenges to adapt to dynamic user interests without any model-level updates. |
| Approach: | They propose a framework that establishes recommendation-oriented in-context learning by structuring recent user interactions and current inputs into ICL formats. |
| Outcome: | The proposed model adapts to dynamic user interests without model updates without any model updates and is available online at https://anonymous.4open.science/r/RecICL-8003. |
Robust Data Watermarking in Language Models by Injecting Fictitious Knowledge (2025.findings-acl)
Copied to clipboard
| Challenge: | Data watermarking in language models injects traceable signals, such as specific token sequences or stylistic patterns, into copyrighted text, allowing copyright holders to track and verify training data ownership. |
| Approach: | They propose a data watermarking approach that injects coherent and plausible yet fictitious knowledge into training data using generated passages describing a fictious entity and its associated attributes. |
| Outcome: | The proposed method is designed to be memorized by the LLM, and that increasing their density, length, and diversity of attributes strengthens their memorization. |
LLM-Enhanced Query Generation and Retrieval Preservation for Task-Oriented Dialogue (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing knowledge retrieval methods for task-oriented dialogues are limited by data scarcity and lack of data to annotate. |
| Approach: | They propose an LLM-enhanced model of query-guided knowledge retrieval for task-oriented dialogue . they propose to select the most relevant knowledge from retrieved top-K records and incorporate them as prompts to guide a generator in response generation. |
| Outcome: | The proposed model outperforms state-of-the-art in three benchmarks on three standard benchmarks. |
ClozeMath: Improving Mathematical Reasoning in Language Models by Learning to Fill Equations (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods to train large language models do not capture how humans learn to think. |
| Approach: | They propose a method to fine-tune large language models for mathematical reasoning by using a text-infilling task that predicts masked equations from a given solution. |
| Outcome: | Experiments on GSM8K, MATH, and GSM-Symbolic show that ClozeMath surpasses baseline Masked Thought in performance and robustness with two test-time scaling decoding algorithms, Beam Search and Chain-of-Thought decoding. |
Low-Entropy Watermark Detection via Bayes’ Rule Derived Detector (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for text watermarking ignore strong evidences embedded in low-entropy tokens, causing statistical measures to falsely indicate the absence of a watermark. |
| Approach: | They propose a Bayes' Rule derived watermark Detector which exploits watermark information from every token by leveraging the posterior probability of watermark’s presence. |
| Outcome: | The proposed method achieves 50% and 70% relative improvements over baselines in code generation and math problem-solving tasks. |
CoD, Towards an Interpretable Medical Agent using Chain of Diagnosis (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing large language models (LLMs) are proving to be effective in medical automatic diagnosis, but their interpretability remains unaddressed. |
| Approach: | They propose to use a "Chain-of-Diagnosis" approach to enhance the interpretability of medical automatic diagnosis by outputting the disease confidence distribution. |
| Outcome: | The proposed model outperforms other LLMs on automatic diagnostic tasks across three real-world benchmarks and provides interpretability while ensuring controllability in diagnostic rigor. |
DaNet: Dual-Aware Enhanced Alignment Network for Multimodal Aspect-Based Sentiment Analysis (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods assume a direct alignment between images and aspects, matching the entire image with a corresponding aspect. Existing algorithms assume 'direct alignment' between images, introducing noise. |
| Approach: | They propose a Dual-Aware Enhanced Alignment Network (DaNet) that can enhance fine-grained multimodal aspect-image alignment and denoising. |
| Outcome: | The proposed system outperforms existing methods in three subtasks and is available on https://github.com/***/DaNet. |
Exploring Multimodal Challenges in Toxic Chinese Detection: Taxonomy, Benchmark, and Findings (2025.findings-acl)
Copied to clipboard
Shujian Yang, Shiyao Cui, Chuanrui Hu, Haicheng Wang, Tianwei Zhang, Minlie Huang, Jialiang Lu, Han Qiu
| Challenge: | Recent studies show that character substitutions in toxic Chinese text can confuse state-of-the-art LLMs. |
| Approach: | They propose a taxonomy of 3 perturbation strategies and 8 specific approaches in Chinese text to assess if they can detect perturbed Chinese toxic contents. |
| Outcome: | The proposed model can detect perturbed Chinese text with 8 different approaches . the proposed model is compared with 9 other LLMs from the US and China . |
LDIR: Low-Dimensional Dense and Interpretable Text Embeddings with Relative Representations (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing text embeddings with high dimensions are difficult to trace and interpret. |
| Approach: | They propose low-dimensional and interpretable text embeddings with relative representations that encode semantic meanings in a vector space where similar texts are close together in the representation space. |
| Outcome: | The proposed embeddings outperform existing models on multiple tasks with fewer dimensions and are lowdimensional and dense while maintaining interpretability. |
Ranked Voting based Self-Consistency of Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing majority voting methods generate only a single answer in each trial, ignoring the possibility of other possible answers. |
| Approach: | They propose to generate ranked answers in each reasoning process and conduct ranked voting among multiple ranked responses from different responses. |
| Outcome: | Extensive experiments show that the proposed method outperforms baselines on multiple-choice and open-ended questions. |
SemanticCamo: Jailbreaking Large Language Models through Semantic Camouflage (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have made safety issues of LLMs more prominent and critical. |
| Approach: | They propose a framework which attacks LLMs through semantic camouflage and replaces unsafe content with semantic features to conceal malicious intent . |
| Outcome: | The proposed framework outperforms existing models in over 80% of cases and is highly effective against various defenses. |
Assigning Distinct Roles to Quantized and Low-Rank Matrices Toward Optimal Weight Decomposition (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing joint optimization methods prioritize one component at the expense of the other, resulting in suboptimal decompositions that fail to leverage each component’s unique strengths. |
| Approach: | They introduce Outlier-Driven Low-Rank Initialization (ODLRI) which assigns low-rank components the specific role of capturing activation-sensitive weights. |
| Outcome: | Experiments on Llama2 (7B, 13B, 70B, and Mistral-7B) and Llma3-8B show that incorporating ODLRI into the joint optimization framework reduces activation-aware error, minimizes quantization scale, and improves perplexity and zero-shot accuracy in low-bit settings. |
Better Process Supervision with Bi-directional Rewarding Signals (2025.findings-acl)
Copied to clipboard
Wenxiang Chen, Wei He, Zhiheng Xi, Honglin Guo, Boyang Hong, Jiazheng Zhang, Nijun Li, Tao Gui, Yun Li, Qi Zhang, Xuanjing Huang
| Challenge: | Existing processes that reward for each step are one-directional and lack a mechanism to model the distance to the final target. |
| Approach: | They propose a process supervision model that evaluates the correctness of previous steps and the probability of future success. |
| Outcome: | The proposed model outperforms existing supervision models like ORM and PRM on reasoning tasks and improves solution re-design. |
KnowCoder-X: Boosting Multilingual Information Extraction via Code (2025.findings-acl)
Copied to clipboard
Yuxin Zuo, Wenxuan Jiang, Wenxuan Liu, Zixuan Li, Long Bai, Hanbin Wang, Yutao Zeng, Xiaolong Jin, Jiafeng Guo, Xueqi Cheng
| Challenge: | Empirical evidence indicates that Large Language Models exhibit spontaneous cross-lingual alignment in Information Extraction (IE) however, a significant imbalance across languages persists, highlighting an underlying deficiency. |
| Approach: | They propose a code LLM with advanced cross-lingual and multilingual capabilities for universal IE that standardizes the representation of multilingual schemas using Python classes and conducts IE alignment instruction tuning on translated instance prediction task. |
| Outcome: | The proposed model surpasses ChatGPT and SoTA by 30.17% without training in 29 unseen languages and significantly improves cross-lingual IE transferability. |
MEIT: Multimodal Electrocardiogram Instruction Tuning on Large Language Models for Report Generation (2025.findings-acl)
Copied to clipboard
Zhongwei Wan, Che Liu, Xin Wang, Chaofan Tao, Hui Shen, Jing Xiong, Rossella Arcucci, Huaxiu Yao, Mi Zhang
| Challenge: | Recent studies have focused on classifying cardiac conditions using ECG data but have overlooked ECG report generation, which is time-consuming and requires clinical expertise. |
| Approach: | They propose a Multimodal ECG Instruction Tuning framework that extends the capability of large language models (LLMs) for the task. |
| Outcome: | The proposed framework outperforms open-source LLMs and LLM backbones across two large-scale ECG datasets. |
Harnessing Large Language Models for Disaster Management: A Survey (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have demonstrated remarkable capabilities across various domains, including their emerging role in mitigating threats to human life, infrastructure, and the environment during natural disasters. |
| Approach: | They propose a taxonomy that categorizes existing LLMs based on disaster phases and application scenarios to provide valuable insights for the research community and practitioners . |
| Outcome: | The proposed taxonomy categorizes existing LLMs based on disaster phases and application scenarios. |
Towards Medical Complex Reasoning with LLMs through Medical Verifiable Problems (2025.findings-acl)
Copied to clipboard
| Challenge: | OpenAI o1 has been a significant milestone in large language model development . however, most research in reasoning has focused on mathematical tasks . medical domains require robust reasoning to provide reliable answers . |
| Approach: | They propose a method to verify medical reasoning using a medical verifier . they also propose RL and reinforcement learning to enhance reasoning . |
| Outcome: | The proposed method outperforms general and medical-specific baselines using only 40K verifiable problems. |
Monitoring Decoding: Mitigating Hallucination via Evaluating the Factuality of Partial Response during Generation (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods to mitiga hallucinations rely on sampling multiple full-length generations, which introduces significant response latency and becomes ineffective when the model consistently produces hallucines. |
| Approach: | They propose a framework that dynamically monitors the generation process and selectively applies in-process interventions to revise hallucination-prone tokens. |
| Outcome: | The proposed framework outperforms self-consistency-based approaches in both effectiveness and efficiency, achieving higher factual accuracy while significantly reducing computational overhead. |
LLM Critics Help Catch Bugs in Mathematics: Towards a Better Mathematical Verifier with Natural Language Feedback (2025.findings-acl)
Copied to clipboard
Bofei Gao, Zefan Cai, Runxin Xu, Peiyi Wang, Ce Zheng, Runji Lin, Keming Lu, Dayiheng Liu, Chang Zhou, Wen Xiao, Tianyu Liu, Baobao Chang
| Challenge: | Existing mathematical verifiers are trained with binary classification labels, which are not informative enough for the model to accurately assess the solutions. |
| Approach: | They propose a natural language feedback-enhanced verifier that can validate the correctness of response generated by policy models by constructing automatically generated training data and a two-stage training paradigm. |
| Outcome: | The proposed verifier significantly improves in verification and reinforcement learning and alleviates data-demanding problems of the reward model. |
EvoBench: Towards Real-world LLM-Generated Text Detection Benchmarking for Evolving Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods to detect LLM-generated texts rely on static benchmarks that neglect the evolving nature of LLMs. |
| Approach: | They propose a benchmark to evaluate the generalization of LLM-generated text detection methods. |
| Outcome: | The proposed benchmark measures generalization of 14 detection methods across LLMs. |
MMSciBench: Benchmarking Language Models on Chinese Multimodal Scientific Problems (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing scientific benchmarks lack human-annotated difficulty levels and structured taxonomies of scientific concepts. |
| Approach: | They propose a benchmark for evaluating mathematical and physical reasoning through text-only and text-image formats with human-annotated difficulty levels and detailed explanations. |
| Outcome: | The proposed model achieves only 63.77% accuracy and struggles with visual reasoning tasks. |
Lightweight Query Checkpoint: Classifying Faulty User Queries to Mitigate Hallucinations in Large Language Model Question Answering (2025.findings-acl)
Copied to clipboard
| Challenge: | Question Answering (QA) with large language models has shown impressive performance, yet hallucinations persist . incorporating LQC into QA pipelines reduces hallucinosities while preserving strong answer quality. |
| Approach: | They propose a small classification model that detects verification-required queries before the LLM generates a potentially faulty answer. |
| Outcome: | The proposed model reduces hallucinations while maintaining strong answer quality. |
Exploring LLM Annotation for Adaptation of Clinical Information Extraction Models under Data-sharing Restrictions (2025.findings-acl)
Copied to clipboard
| Challenge: | In-hospital text data often contains valuable clinical information, yet fine-tuned small language models (SLMs) for information extraction remain challenging due to differences in formatting and vocabulary across institutions. |
| Approach: | They leverage large language models to annotate the target domain data for adaptation . they use in-hospital text data to extract clinical information . |
| Outcome: | The proposed model outperforms manual annotation on four clinical information extraction tasks with a larger number of annotated data. |
Enhancing the Comprehensibility of Text Explanations via Unsupervised Concept Discovery (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing concepts-based explainable approaches do not discover unseen concepts . a recent approach to solve this problem is concept-based explanations . |
| Approach: | They propose a framework that extracts comprehensible concepts automatically with no annotations . ECO-Concept uses an object-centric architecture to extract task-specific semantic concepts . |
| Outcome: | a new framework extracts comprehensible concepts with no concept annotations . the proposed framework outperforms existing methods in computability tests on diverse tasks . |
RecordTwin: Towards Creating Safe Synthetic Clinical Corpora (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods to generate high-quality synthetic corpus from clinical documents require learning from the original clinical documents. |
| Approach: | They propose a method to generate synthetic corpus from clinical documents using a large language model. |
| Outcome: | The proposed method generates synthetic documents from in-hospital clinical documents. |
Beyond Surface-Level Patterns: An Essence-Driven Defense Framework Against Jailbreak Attacks in LLMs (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods focus on surface-level patterns, overlooking the deeper attack essences. |
| Approach: | They propose an Essence-Driven Defense Framework Against Jailbreak Attacks in Aligned Large Language Models that extracts the "attack essence" from a diverse set of known attack instances and stores it in an offline vector database. |
| Outcome: | The proposed framework outperforms existing methods by reducing the Attack Success Rate by at least 20%, underscoring its superior robustness against jailbreak attacks. |
Multimodal Invariant Sentiment Representation Learning (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for multimodal sensing ignore significant sentiment distribution imbalances and cross-modal sentiment conflicts, hindering performance improvement. |
| Approach: | They propose a method to learn stable multimodal invariant sentiment representations by incorporating distributional discrepancies and sentiment conflicts into the model training. |
| Outcome: | The proposed method improves MSA performance and achieves new state-of-the-art. |
ChuLo: Chunk-Level Key Information Representation for Long Document Understanding (2025.findings-acl)
Copied to clipboard
| Challenge: | Traditional approaches to truncate inputs, sparse self-attention, and chunking often lead to information loss and hinder the model’s ability to capture long-range dependencies. |
| Approach: | They propose a novel chunk representation method that uses unsupervised keyphrase extraction to group input tokens to retain core document content while reducing input length. |
| Outcome: | The proposed method minimizes information loss and improves the efficiency of Transformer-based models. |
REVS: Unlearning Sensitive Information in Language Models via Rank Editing in the Vocabulary Space (2025.findings-acl)
Copied to clipboard
| Challenge: | Current approaches to address this issue involve costly dataset scrubbing or model filtering through unlearning and model editing. |
| Approach: | They propose a method for unlearning sensitive information from language models . they curate email and URL datasets and a social security number dataset . |
| Outcome: | The proposed method shows superior performance and robustness to extraction attacks on real-world datasets. |
Is External Information Useful for Stance Detection with LLMs? (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing studies suggest that external information, such as Wikipedia excerpts, can improve stance detection performance, but whether such information can benefit large language models remains an unanswered question. |
| Approach: | They conduct a systematic evaluation on how Wikipedia and web search external information can affect stance detection across eight LLMs and in three datasets with 12 targets. |
| Outcome: | The proposed model degrades performance in most cases, with macro F1 scores dropping by up to 27.9%, while fine-tuning mitigates but does not eliminate it. |
Benchmarking Query-Conditioned Natural Language Inference (2025.findings-acl)
Copied to clipboard
| Challenge: | Query-conditioned natural language inference (QC-NLI) is a new approach to detect inconsistencies in large language models. |
| Approach: | They propose a task of Query-Conditioned Natural Language Inference to determine the semantic relationship between two documents conditioned on a query. |
| Outcome: | The proposed task is based on a query-conditioned natural language inference (QC-NLI) it is used to determine the relationship between the premise and hypothesis given a given query. |
Flowchart-Based Decision Making with Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models face significant challenges in interpretability of dialogue flow and reproducibility of expert knowledge. |
| Approach: | They propose a method that extracts flowcharts from dialogue data and incorporates them into large language models to improve interpretability and reproducibility. |
| Outcome: | The proposed method reconstructs expert decision-making paths with high precision and recall scores on dialogue datasets. |
NarGINA: Towards Accurate and Interpretable Children’s Narrative Ability Assessment via Narrative Graphs (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for assessing children's narrative ability are limited to evaluating completeness of narrative content and the coherence of expression, as well as interpretability of assessment results. |
| Approach: | They propose a computational framework for assessing narrative ability using a narrative graph to provide a concise and structured summary representation of narrative text. |
| Outcome: | The proposed framework achieves significant performance improvement over baselines while possessing good interpretability. |
Improving Efficiency in Large Language Models via Extendable Block Floating Point Representation (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) are becoming more and more resource-intensive as their size increases. |
| Approach: | They propose a block floating-point (BFP) arithmetic representation that extends the exponent bit width to capture a wider dynamic range. |
| Outcome: | Extendable Exponent Sharing (EES) outperforms representative baselines in accuracy and computational efficiency. |
EpiCoDe: Boosting Model Performance Beyond Training with Extrapolation and Contrastive Decoding (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods to enhance performance of Large language models are limited due to the cost of training data and privacy concerns. |
| Approach: | They propose a method that enhances a finetuned model with its inferior version and adopts contrastive decoding to reduce predicted errors. |
| Outcome: | The proposed method outperforms existing methods in data-scarcity scenarios across three domains and shows that it is more robust and robust. |
NativQA: Multilingual Culturally-Aligned Natural Query for LLMs (2025.findings-acl)
Copied to clipboard
Md. Arid Hasan, Maram Hasanain, Fatema Ahmad, Sahinur Rahman Laskar, Sunaya Upadhyay, Vrunda N Sukhadia, Mucahid Kutlu, Shammur Absar Chowdhury, Firoj Alam
| Challenge: | Existing frameworks for QA datasets lack regional specificity and cultural specificity. |
| Approach: | They propose a framework to quench native language QA datasets in native languages for LLM evaluation and tuning. |
| Outcome: | The proposed framework is scalable, language-independent and can be used to build culturally and regionally aligned QA datasets in native languages. |
DoCIA: An Online Document-Level Context Incorporation Agent for Speech Translation (2025.findings-acl)
Copied to clipboard
Xinglin Lyu, Wei Tang, Yuang Li, Xiaofeng Zhao, Ming Zhu, Junhui Li, Yunfei Lu, Min Zhang, Daimeng Wei, Hao Yang, Min Zhang
| Challenge: | Document-level context is crucial for speech translation due to noise from ASR . incorporating document-level contextual information into ST remains a challenge . |
| Approach: | They develop an online framework that integrates document-level context into machine translation . they use document-based modules to integrate document- level context into ST . |
| Outcome: | The proposed framework outperforms baselines in sentence and discourse metrics . it can correct ASR transcription errors and improve translation performance . |
RISE: Reasoning Enhancement via Iterative Self-Exploration in Multi-hop Question Answering (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) excel in many areas but face challenges with complex reasoning tasks, such as Multi-Hop Question Answering (MHQA). |
| Approach: | They propose a framework to enhance models’ reasoning capability through iterative self-exploration that addresses key errors in MHQA tasks such as Evidence Aggregation and Reasoning Decomposition. |
| Outcome: | Extensive experiments on multiple MHQA benchmarks show that the proposed framework significantly improves reasoning accuracy and task performance. |
VADE: Visual Attention Guided Hallucination Detection and Elimination (2025.findings-acl)
Copied to clipboard
| Challenge: | Vision Language Models (VLMs) are prone to hallucinations, generating outputs that lack grounding in the actual visual data. |
| Approach: | They propose a sequence modelling approach to learn complex sequential patterns from transformer attention maps. |
| Outcome: | The proposed approach achieves an average PR-AUC of 80% in hallucination detection on M-HalDetect and an 5% improvement in hallucinosis mitigation on MSCOCO. |
PGPO: Enhancing Agent Reasoning via Pseudocode-style Planning Guided Preference Optimization (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing LLM agents generate verbose and inefficient natural language plans to guide reasoning, which restricts agents’ ability to generalize across similar tasks. |
| Approach: | They propose a pseudocode-style planning guide optimization method that captures the structural logic of reasoning and uses two planning-oriented rewards to enhance agent learning. |
| Outcome: | The proposed method outperforms existing LLM agents on representative agent benchmarks and outperformed the current leading baselines. |
The Effectiveness of Uncased Tokeniziaion for Clinical Notes (2025.findings-acl)
Copied to clipboard
| Challenge: | a recent study shows that uncased models exceed the performance of cased models on clinical notes . case-sensitive tokenization is often seen as useful for resolving ambiguities, but for some types of text, the observed case is merely an interpretation of spoken language . |
| Approach: | They show that uncased models exceed the performance of cased models on clinical notes . authors propose independent case encoding to better balance model performance on cases . |
| Outcome: | a new study shows that uncased models outperform cased models on clinical notes . the authors say the new model improves on case-sensitive and improperly-cased tasks . |
AMXFP4: Taming Activation Outliers with Asymmetric Microscaling Floating-Point for 4-bit LLM Inference (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing research shows unsatisfactory empirical results for microscaling (MX) floating-point (FP) formats. |
| Approach: | They propose a 4-bit asymmetric FP format that handles activation outliers without calibration. |
| Outcome: | The proposed format outperforms MXFP4 by 3% on VQA and rotation-based methods by 1.6% on CSQA. |
Improving Continual Pre-training Through Seamless Data Packing (2025.findings-acl)
Copied to clipboard
| Challenge: | Empirical evaluations across various model architectures and corpus domains demonstrate the effectiveness of our method, outperforming baselines in 99% of all settings. |
| Approach: | They propose a method that uses a sliding window technique to pack data before continual pre-training to preserve contextual information and enhance model performance. |
| Outcome: | Empirical evaluations across various model architectures and corpus domains demonstrate the effectiveness of the proposed method outperforming baselines in 99% of settings. |
The Impact of Name Age Perception on Job Recommendations in LLMs (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing studies have examined implicit age-related name bias in LLMs . older-sounding names are favored for senior roles, while younger-sounders are linked to youth-dominant jobs . |
| Approach: | They analyze six LLMs and 117 American names categorized by perceived age across 30 occupations . older-sounding names are favored for senior roles, while younger-sounders are linked to youth-dominant jobs . |
| Outcome: | The proposed model based on six LLMs and 117 American names shows that older-sounding names are favored for senior roles, while younger-sounders are linked to youth-dominant jobs. |
DAPI: Domain Adaptive Toxicity Probe Vector Intervention, for Fine-Grained Detoxification (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing studies rely on a single toxicity probe vector to reduce toxicity, but toxicity can be fine-grained into subcategories, making it difficult to remove certain types of toxicity. |
| Approach: | They propose a category-specific toxicity probe vector approach that uses multiple toxicity vectors to generate a model and dynamically scale it to remove toxicity from categories that the single probe vector failed to detoxify. |
| Outcome: | The proposed method reduces toxicity from categories that the single probe vector approach failed to detoxify. |
Task Knowledge Injection via Interpolations and Reinstatement for Large Language Model Generalization (2025.findings-acl)
Copied to clipboard
| Challenge: | Pre-trained large language models have been widely adopted to elicit their superior performance on downstream tasks, but instruction tuning may overfit them to specific task formats, compromising their generalization on unseen tasks. |
| Approach: | They propose to inject latent task adaptation and knowledge reinstatement into large language models to mitigate spurious correlations between inputs and targets. |
| Outcome: | The proposed method improves generalization on in-domain and out-of-domain unseen tasks. |
STARS: A Unified Framework for Singing Transcription, Alignment, and Refined Style Annotation (2025.findings-acl)
Copied to clipboard
Wenxiang Guo, Yu Zhang, Changhao Pan, Zhiyuan Zhu, Ruiqi Li, ZheTao Chen, Wenhao Xu, Fei Wu, Zhou Zhao
| Challenge: | Existing automated singing annotation (ASA) methods tackle isolated aspects of the annotation pipeline. |
| Approach: | They propose a framework that addresses transcription, alignment, and refined style annotations. |
| Outcome: | The proposed framework delivers comprehensive multi-level annotations encompassing: (1) precise phoneme-audio alignment, (2) robust note transcription and temporal localization, (3) expressive vocal technique identification, and (4) global stylistic characterization including emotion and pace. |
Unveiling the Key Factors for Distilling Chain-of-Thought Reasoning (2025.findings-acl)
Copied to clipboard
Xinghao Chen, Zhijing Sun, Guo Wenjin, Miaoran Zhang, Yanjun Chen, Yirong Sun, Hui Su, Yijie Pan, Dietrich Klakow, Wenjie Li, Xiaoyu Shen
| Challenge: | Large Language Models (LLMs) excel in reasoning tasks through Chain-of-Thought prompting. |
| Approach: | They examine the factors influencing CoT distillation including granularity, format and teacher model. |
| Outcome: | The proposed model is based on four teacher models and seven student models across seven mathematical and commonsense reasoning datasets. |
INT: Establishing Information Transfer for Multilingual Intent Detection and Slot Filling (2025.findings-acl)
Copied to clipboard
Di Wu, Liting Jiang, Bohui Mao, Hongyan Xie, Haoxiang Su, Zhongjiang He, Ruiyu Fang, Shuangyong Song, Hao Huang, Xuelong Li
| Challenge: | Existing studies struggle to achieve performance comparable to that on high-resource languages due to inherent linguistic diversity of multilingual SLU tasks. |
| Approach: | They propose a multilingual information transfer network to solve these challenges . they propose to reformulate SF as a span prediction problem and introduce a slot-matching attention mechanism to achieve slot alignment across languages. |
| Outcome: | The proposed model outperforms baseline models on the MASSIVE and MASSIV-UG datasets in overall accuracy across all languages. |
Enhancing LLM Agent Safety via Causal Influence Prompting (2025.findings-acl)
Copied to clipboard
| Challenge: | Experimental results demonstrate that our method effectively enhances safety in code execution and mobile device control tasks. |
| Approach: | They propose a technique that leverages causal influence diagrams to identify and mitigate risks arising from agent decision-making. |
| Outcome: | The proposed method enhances safety in code execution and mobile device control tasks. |
Position Paper: MeMo: Towards Language Models with Associative Memory Mechanisms (2025.findings-acl)
Copied to clipboard
Fabio Massimo Zanzotto, Elena Sofia Ruzzetti, Giancarlo A. Xompero, Leonardo Ranaldi, Davide Venditti, Federico Ranaldi, Cristina Giannone, Andrea Favalli, Raniero Romagnoli
| Challenge: | Memorization is a fundamental ability of Transformer-based Large Language Models, achieved through learning. |
| Approach: | They propose an architecture that explicitly memorizes sequences of tokens in layered associative memories. |
| Outcome: | The proposed architecture shows that memorization is a fundamental ability of large language models, achieved through learning. |
DeRAGEC: Denoising Named Entity Candidates with Synthetic Rationale for ASR Error Correction (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent studies have demonstrated that postprocessing speech recognition transcriptions with large language models can significantly enhance the accuracy of Automatic Speech Recognition (ASR). |
| Approach: | They propose a method to improve Named Entity (NE) correction in Automatic Speech Recognition systems by leveraging phonetic similarity and augmented definitions. |
| Outcome: | The proposed method outperforms baseline methods on common voice and STOP datasets and achieves a 28% reduction in WER and NE hit ratio. |
Rehearse With User: Personalized Opinion Summarization via Role-Playing based on Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent studies show that large language models can achieve stateof-the-art performance on standard summarization benchmarks without the need for large-scale training data. |
| Approach: | They propose a personalized opinion summarization framework via LLM-based role-playing to better understand the user's personalized needs. |
| Outcome: | The proposed framework can improve the level of personalization in large model-generated summaries by taking into account user characteristics and interests while summarizing multiple product reviews. |
AdParaphrase v2.0: Generating Attractive Ad Texts Using a Preference-Annotated Paraphrase Dataset (2025.findings-acl)
Copied to clipboard
| Challenge: | Identifying factors that make ad text attractive is essential for advertising success . identifying the linguistic factors presents a significant challenge because of the intricate interplay between the semantic content and its linguistic expression. |
| Approach: | They propose to use a dataset for ad text paraphrasing that contains human preference data to enable analysis of linguistic factors. |
| Outcome: | The proposed dataset is 20 times larger than v1.0 and contains 16,460 pairs of ad text paraphrase pairs . it shows that human preference and ade- t attractiveness are related . |
Beyond the Average Reader: the Reader Embedding Approach (2025.findings-acl)
Copied to clipboard
| Challenge: | a new approach to predict reading times is proposed to use eye-tracking data to collect data from all subjects rather than from the most similar ones. |
| Approach: | They propose a method to collect eye-tracking data that are averaged and used to train learning models. |
| Outcome: | The proposed approach outperforms existing methods by combining eye-tracking data with averaged data. |
PredictaBoard: Benchmarking LLM Score Predictability (2025.findings-acl)
Copied to clipboard
Lorenzo Pacchiardi, Konstantinos Voudouris, Ben Slater, Fernando Martínez-Plumed, Jose Hernandez-Orallo, Lexin Zhou, Wout Schellaert
| Challenge: | Large Language Models (LLMs) fail unpredictably, demonstrating inconsistent success in even basic common sense reasoning tasks. |
| Approach: | They propose a framework to evaluate the ability of score predictors to anticipate LLM errors on specific task instances from existing datasets. |
| Outcome: | The proposed framework evaluates the ability of score predictors to anticipate LLM errors on specific task instances from existing datasets. |
FedDQC: Data Quality Control in Federated Instruction-tuning of Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Federated Learning (FL) enables privacy-preserving collaborative instruction tuning of large language models. |
| Approach: | They propose a federated instruction tuning framework with dynamic data quality control to solve this problem. |
| Outcome: | The proposed framework improves performance on mixed-quality datasets on synthetic and real-world datasets. |
Weed Out, Then Harvest: Dual Low-Rank Adaptation is an Effective Noisy Label Detector for Noise-Robust Learning (2025.findings-acl)
Copied to clipboard
| Challenge: | Experimental results show that PEFT can fine-tune language models without relying on perfectly labeled datasets. |
| Approach: | They propose a framework that decouples sample selection from model training by introducing clean and noisy LoRA. |
| Outcome: | The proposed framework decouples sample selection from model training. |
“I understand your perspective”: LLM Persuasion through the Lens of Communicative Action Theory (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) can generate high-quality arguments, yet their ability to engage in nuanced and persuasive communicative actions remains largely unexplored. |
| Approach: | They examine whether Large Language Models express illocutionary intent in ways comparable to human communication by simulated online discussions . |
| Outcome: | The proposed models express illocutionary intents in ways comparable to human communication, and crowd-sourced workers prefer them over human-written ones. |
Nunchi-Bench: Benchmarking Language Models on Cultural Reasoning with a Focus on Korean Superstition (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing research has evaluated large language models' cultural knowledge and contextual understanding, reducing their effectiveness in multicultural settings. |
| Approach: | They propose a benchmark to evaluate LLMs' cultural understanding with a focus on Korean superstitions. |
| Outcome: | The proposed benchmark assesses multilingual LLMs in Korean and English to analyze their ability to reason about Korean cultural contexts and how language variations affect performance. |
Let’s Be Self-generated via Step by Step: A Curriculum Learning Approach to Automated Reasoning with Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing efforts to improve CoT prompting have limitations that require extensive human effort or performance needs to be improved. |
| Approach: | They propose a prompt approach for automatic reasoning called LBS3 inspired by curriculum learning which better reflects human learning habits. |
| Outcome: | The proposed approach achieves strongly competitive performance compared to baselines in reasoning-intensive tasks with varying open- and closed-source LLMs. |
daDPO: Distribution-Aware DPO for Distilling Conversational Abilities (2025.findings-acl)
Copied to clipboard
| Challenge: | Knowledge distillation (KD) with Direct Preference Optimization (DPO) has emerged as a promising approach to enhance the conversational abilities of smaller models using a larger teacher model. |
| Approach: | They propose a framework that integrates the teacher's distributional information into DPO distillation while preserving theoretical guarantees. |
| Outcome: | The proposed framework outperforms existing methods in restoring performance for pruned models and enhancing smaller models within the same LLM family. |
Consultant Decoding: Yet Another Synergistic Mechanism (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) have attracted widespread attention and adoption across diverse domains due to their exceptional performance and robust generalization abilities. |
| Approach: | They propose a synergetic mechanism for Consultant Decoding (CD) that achieves a 2.5-fold increase in inference speed compared to the target model while maintaining comparable generation quality. |
| Outcome: | The proposed mechanism achieves 2.5-fold increase in inference speed while maintaining comparable generation quality (100% of the target model’s performance). |
IntelliCockpitBench: A Comprehensive Benchmark to Evaluate VLMs for Intelligent Cockpit (2025.findings-acl)
Copied to clipboard
| Challenge: | Visual Question Answering (VQA) is a key task in vehicular systems. |
| Approach: | They propose a benchmark that encompasses diverse automotive scenarios . they use images from front, side, and rear cameras, various road types, weather conditions, and interior views . |
| Outcome: | The proposed benchmark includes images from front, side, and rear cameras, various road types, weather conditions, and interior views. |
Analyzing Political Bias in LLMs via Target-Oriented Sentiment Classification (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods to analyze political biases rely on small-size intermediate tasks and the LLMs themselves. |
| Approach: | They propose an entropy-based inconsistency metric to encode political biases . they insert 1319 demographically and politically diverse politician names in 450 political sentences . |
| Outcome: | The proposed method combines high accuracy with a correct understanding of the candidate candidate. |
PISCO: Pretty Simple Compression for Retrieval-Augmented Generation (2025.findings-acl)
Copied to clipboard
| Challenge: | Document compression methods suffer from accuracy losses and limited context size. |
| Approach: | They propose a method that achieves a 16x compression rate with minimal accuracy loss . they show that PISCO outperforms existing compression models by 8% in accuracy . |
| Outcome: | The proposed method outperforms existing compression models by 8% in accuracy. |
AnchorCoT: Anchors Pave the Way for Multi-hop Reasoning (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) have demonstrated potential reasoning capabilities through prompt design, such as the Chain of Thought (CoT). |
| Approach: | They propose a new reasoning approach that predicts key entities which work as important “anchors” and employs a ranking algorithm to ensure the logical sequence of the predicted answers. |
| Outcome: | The proposed approach outperforms existing methods in multi-hop question reasoning and provides more accurate reasoning results in multihop question answering tasks. |
Token Pruning in Multimodal Large Language Models: Are We Solving the Right Problem? (2025.findings-acl)
Copied to clipboard
| Challenge: | Multimodal large language models have shown remarkable performance for cross-modal understanding and generation, yet suffer from severe inference costs. |
| Approach: | They propose to prune redundant tokens in MLLMs to reduce computation and storage costs. |
| Outcome: | The proposed method reduces the computational and storage costs of MLLMs by identifying redundant tokens and pruning them. |
Federated Data-Efficient Instruction Tuning for Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing federated learning (FL) uses all local data, causing excessive computational overhead and overfitting to local data. |
| Approach: | They propose a federated data-efficient instruction tuning approach which utilizes a representative subset of edge-side data to tune LLMs. |
| Outcome: | The proposed method improves Rouge-L on unseen tasks by 10.72% over the SOTA full-data instruction tuning methods while using less than 1.5% of the data samples. |
They want to pretend not to understand: The Limits of Current LLMs in Interpreting Implicit Content of Political Discourse (2025.findings-acl)
Copied to clipboard
| Challenge: | a recent study shows that large language models lack the pragmatic capabilities needed to interpret highly implicit content. |
| Approach: | They propose to use transcribed italian political speeches to test their ability to interpret implicit content. |
| Outcome: | The proposed model provides a fully correct explanation in only one-fourth of cases in the open-ended generation setup. |
ZeroNER: Fueling Zero-Shot Named Entity Recognition via Entity Type Descriptions (2025.findings-acl)
Copied to clipboard
Alessio Cocchieri, Marcos Martínez Galindo, Giacomo Frisoni, Gianluca Moro, Claudio Sartori, Giuseppe Tagliavini
| Challenge: | Existing zero-shot learning methods rely on entity type names for generalization . current solutions require large datasets and prioritize a handful of commonly occurring types . |
| Approach: | They propose a description-driven framework that enhances hard zero-shot NER in low-resource settings. |
| Outcome: | The proposed framework outperforms existing models by up to 16% in the F1 score . it also surpasses baseline models that use type names alone . |
Do Large Language Models Have “Emotion Neurons”? Investigating the Existence and Role (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing evaluations of LLMs' emotional capabilities have been criticized for not illuminating how emotion information is processed and represented within an LLM. |
| Approach: | They examine whether there are “emotion neurons” within large language models that selectively process and express certain emotions and what functional role they play. |
| Outcome: | The proposed model is based on the representative emotion theory of the six basic emotions and demonstrates that it is functionally significant to examine whether the prediction accuracy for a specific emotion decreases when the neurons are removed. |
Grammar-Based Code Representation: Is It a Worthy Pursuit for LLMs? (2025.findings-acl)
Copied to clipboard
Qingyuan Liang, Zhao Zhang, Zeyu Sun, Zheng Lin, Qi Luo, Xiao Yueyi, Yizhou Chen, Yuqun Zhang, Haotian Zhang, Lu Zhang, Chenbin Chenbin, Yingfei Xiong
| Challenge: | Existing research demonstrates the effectiveness of grammar-based code representations in small-scale models, showing their ability to reduce syntax errors and enhance performance. |
| Approach: | They develop a series of billion-scale grammar-based code representations that incorporate grammar rules into the code generation process. |
| Outcome: | Experiments on HumanEval and MBPP show that grammar-based representations reduce syntax errors and improve performance even in billion-scale models. |
Investigating Inference-time Scaling for Chain of Multi-modal Thought: A Preliminary Study (2025.findings-acl)
Copied to clipboard
| Challenge: | Inference-time scaling of chain-of-thought (CoT) has been demonstrated as a promising approach for addressing multi-modal reasoning tasks. |
| Approach: | They propose to integrate visual and textual modalities within the reasoning process . they adopt a consistency-enhanced verifier to ensure effective guidance for both methods across different thought paradigms. |
| Outcome: | The proposed method outperforms text-only reasoning on 10 tasks spanning diverse domains and requires higher token consumption for processing richer visual inputs. |
UI-E2I-Synth: Advancing GUI Grounding with Large-Scale Instruction Synthesis (2025.findings-acl)
Copied to clipboard
| Challenge: | Graphical User Interface (GUI) agents that utilize human-like vision perception capabilities are gaining a wider applicability compared to GUI metadata-based approaches. |
| Approach: | They propose a large-scale data synthesis pipeline for generating varying complex instruction datasets using GPT-4o instead of human annotators. |
| Outcome: | The proposed model achieves superior performance in GUI instruction grounding, demonstrating the advancements of proposed data synthesis pipeline. |
A Study into Investigating Temporal Robustness of LLMs (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) are limited in their ability to process temporal information and perform tasks requiring temporal reasoning and factual knowledge. |
| Approach: | They propose to use eight time-sensitiverobustness tests to test the model's temporal robustness for user questions in the zero-shot setting. |
| Outcome: | The proposed tests improve the temporal QA performance by up to 55%. |
ToolExpNet: Optimizing Multi-Tool Selection in LLMs with Similarity and Dependency-Aware Experience Networks (2025.findings-acl)
Copied to clipboard
| Challenge: | Current empirical methods that focus on isolated tools learning struggle with accurate multi-tool selection due to issues like confusing similar tools and neglecting dependencies. |
| Approach: | They propose a tool-learning paradigm which integrates tools and trial-and-error experiences into a network characterized by semantic similarity and dependency relationships. |
| Outcome: | The proposed model outperforms existing methods on multiple real-world API datasets and significantly outperformed baselines. |
SPILL: Domain-Adaptive Intent Clustering based on Selection and Pooling with Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for intent clustering rely on labeled examples or unsupervised fine-tuning to optimize results for each new dataset. |
| Approach: | They propose a method that uses an embedder to derive an embedding for each utterance and then pool them with the seed to improve the embeddable results. |
| Outcome: | The proposed method outperforms embedding methods and is comparable to state-of-the-art methods. |
How Far are LLMs from Being Our Digital Twins? A Benchmark for Persona-Based Behavior Chain Simulation (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent studies have focused on dialogue simulation while overlooking human behavior simulation, which is crucial for digital twins. |
| Approach: | They propose to integrate persona metadata into LLMs and use it to iteratively infer contextually appropriate behaviors within dynamic scenarios. |
| Outcome: | The proposed model is based on 15,846 distinct behaviors across 1,001 unique personas and incorporates persona metadata to iteratively infer appropriate behaviors within dynamic scenarios. |
GRI-QA: a Comprehensive Benchmark for Table Question Answering over Environmental Data (2025.findings-acl)
Copied to clipboard
Michele Luca Contalbo, Sara Pederzoli, Francesco Del Buono, Venturelli Valeria, Francesco Guerra, Matteo Paganelli
| Challenge: | Existing Table Question Answering systems for environmental reporting are complex and require specialized terminology and complex questions. |
| Approach: | They propose a test benchmark to evaluate Table QA approaches in the environmental domain. |
| Outcome: | The proposed benchmark reveals a significant gap between human and model performance in multi-step reasoning. |
WebUIBench: A Comprehensive Benchmark for Evaluating Multimodal Large Language Models in WebUI-to-Code (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing benchmarks for large language models focus on webpage generation outcomes. |
| Approach: | They propose a multi-view evaluation framework to evaluate MLLMs in four key areas: WebUI Perception, HTML Programming, WebUI-HTML Understanding, and WebUI to code. |
| Outcome: | The proposed framework evaluates MLLMs in four key areas: WebUI Perception, HTML Programming, WebUI-HTML Understanding, and WebUI to code. |
Optimizing Multi-Hop Document Retrieval Through Intermediate Representations (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing approaches to addressing multi-hop queries are computationally expensive . despite their success, large language models often generate factually incorrect answers . |
| Approach: | They propose a layer-by-layer reasoning approach that leverages intermediate representations from the middle layers to retrieve external knowledge. |
| Outcome: | The proposed method outperforms existing RAG methods on open-domain multi-hop question-answering datasets while maintaining inference overhead similar to that of standard RAG. |
Towards Better Understanding of Program-of-Thought Reasoning in Cross-Lingual and Multilingual Environments (2025.findings-acl)
Copied to clipboard
Patomporn Payoungkhamdee, Pume Tuchinda, Jinheon Baek, Samuel Cahyawijaya, Can Udomcharoenchaikit, Potsawee Manakul, Peerat Limkonchotiwat, Ekapol Chuangsuwanich, Sarana Nutanong
| Challenge: | Multi-step reasoning is essential for large language models, yet multilingual performance remains challenging. |
| Approach: | They propose a framework to evaluate Program-of-Thought (PoT) prompting by separating multilingual reasoning from code execution to examine impact of fine-tuning on question-reasoning alignment and reasoning quality. |
| Outcome: | The proposed framework outperforms CoT fine-tuned models in multilingual settings and shows strong correlation between reasoning quality and answer accuracy. |
A Fully Automated Pipeline for Conversational Discourse Annotation: Tree Scheme Generation and Labeling with Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent advances in Large Language Models (LLMs) have shown promise in automating discourse annotation for conversations. |
| Approach: | They propose a pipeline that uses large language models to construct and perform annotations using speech functions and the Switchboard-DAMSL taxonomies. |
| Outcome: | The proposed pipeline outperforms existing tree annotation schemes and can match or surpass human annotations while significantly reducing time required for annotation. |
Can Language Models Serve as Analogy Annotators? (2025.findings-acl)
Copied to clipboard
| Challenge: | Conceptual abstraction and analogy-making are crucial for human learning, reasoning, and adapting to unfamiliar domains. |
| Approach: | They propose a multi-stage progressive reasoning prompt framework A3E which is based on the structure mapping theory from cognitive psychology and efficiently annotates candidate story pairs across six fine-grained categories. |
| Outcome: | The proposed framework achieves an average performance gain of + 73% across a range of prompting baselines and base LLMs. |
Reward Generalization in RLHF: A Topological Perspective (2025.findings-acl)
Copied to clipboard
Tianyi Alex Qiu, Fanzhi Zeng, Jiaming Ji, Dong Yan, Kaile Wang, Jiayi Zhou, Yang Han, Josef Dai, Xuehai Pan, Yaodong Yang
| Challenge: | Existing alignment methods share a common topology of information flow, but their alternatives have not been thoroughly explored. |
| Approach: | They propose a theory of reward generalization in reinforcement learning from human feedback . they propose induced Bayesian networks to model the impact of dataset topologies on reward generalisation . |
| Outcome: | The proposed method achieves an average win rate of 65% on three NLP tasks. |
Enhanced Data Synthesis for LLM through Reasoning Structures Generated by Hierarchical GFlowNet (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods to optimize instruction-response pairs lack a systematic design for the underlying reasoning structure. |
| Approach: | They propose a Reasoning Structure driven data Synthesis method that leverages a coarse-to-fine directed acyclic graph to construct reasoning structures efficiently. |
| Outcome: | The proposed method outperforms existing methods in 48.50%, 84.00%, 79.90% of the synthetic datasets trained on the proposed model. |
Capturing Nuanced Preferences: Preference-Aligned Distillation for Small Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for aligning small language models with human values model preference knowledge from large language models (LLMs) however, this limitation hinders student SLMs from capturing nuanced preferences for multiple responses. |
| Approach: | They propose a framework which models teacher's preference knowledge as a probability distribution over all potential preferences, thereby providing more nuanced supervisory signals. |
| Outcome: | The proposed framework outperforms existing methods on four benchmark tasks and achieves 20% improvement on AlpacaEval 2 and Arena-Hard. |
Token-level Preference Self-Alignment Optimization for Multi-style Outline Controllable Generation (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing attempts to outline generation are limited by response pair requirements and substantial computation costs. |
| Approach: | They propose a token-level preference self-alignment optimization for outline controllable generation that extends the Bradley-Terry model from pair-wise to list-wise comparison. |
| Outcome: | The proposed method outperforms existing methods by 19.28% in performance while requiring only 56.25% training time. |
HatePRISM: Policies, Platforms, and Research Integration. Advancing NLP for Hate Speech Proactive Mitigation (2025.findings-acl)
Copied to clipboard
Naquee Rizwan, Seid Muhie Yimam, Daryna Dementieva, Dr. Florian Skupin, Tim Fischer, Daniil Moskovskiy, Aarushi Ajay Borkar, Robert Geislinger, Punyajoy Saha, Sarthak Roy, Martin Semmann, Alexander Panchenko, Chris Biemann, Animesh Mukherjee
| Challenge: | Existing approaches to manage hate speech rely on reactive measures such as blocking or suspending offensive messages . despite regulations imposed by nations and social media platforms, hateful content remains a challenge . |
| Approach: | They propose a framework for automated hate speech moderation based on different strategies . they examine hate speech regulations and strategies from three perspectives . |
| Outcome: | The proposed framework could be based on a combination of country regulations, social platform policies, and NLP research datasets. |
Local Look-Ahead Guidance via Verifier-in-the-Loop for Automated Theorem Proving (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent methods for AI reasoning require applying variants of reinforcement learning (RL) on rolled out trajectories, even for step-wise rewards, or large quantities of human-annotated trajectory data. |
| Approach: | They propose a verifier-in-the-loop design that uses an automated verifier to give intermediate feedback at each step of the reasoning process. |
| Outcome: | The proposed model improves on the Automatic Theorem Proving task using Lean as the verifier. |
Generalizable Cross-Lingual Cognitive Distortion Detection with Standardized Annotations and Multi-Task Learning (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing studies on cognitive distortion have limited generalizability and performance of models in large-scale and cross-linguistic contexts. |
| Approach: | They propose a multi-task learning model based on teacher student architecture solution which improves generalization performance. |
| Outcome: | The proposed model improves generalizability and interpretability of the proposed model. |
How Do Multilingual Language Models Remember Facts? (2025.findings-acl)
Copied to clipboard
| Challenge: | Prior research has focused on English monolingual models, but how these mechanisms generalize to non-English languages remains unexplored. |
| Approach: | They analyze three multilingual LLMs to find out how they can generalize recall mechanisms . they find that subject enrichment is language-independent, object extraction is language dependent . |
| Outcome: | The proposed model performs better in multilingual contexts than in English models . the model is more efficient in multi-lingual context, but it is more complex in multilinguistic models compared to English models. |
SeqPO-SiMT: Sequential Policy Optimization for Simultaneous Machine Translation (2025.findings-acl)
Copied to clipboard
| Challenge: | SeqPO-SiMT is a new policy optimization framework for simultaneous machine translation that combines a tailored reward with a single step task. |
| Approach: | They propose a new policy optimization framework that defines the simultaneous machine translation task as a sequential decision making problem with a tailored reward. |
| Outcome: | The proposed framework outperforms the supervised fine-tuning model by 1.13 points while reducing the Average Lagging by 6.17 in the NEWSTEST2021 En Zh dataset. |
Do Large Language Models Know Folktales? A Case Study of Yokai in Japanese Folktales (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have strong language understanding and generation abilities, but cultural knowledge is often limited to English-speaking communities. |
| Approach: | They propose a benchmark dataset to evaluate the cultural awareness of large language models . they use Japanese folktales to evaluate their knowledge of Yokai . |
| Outcome: | The proposed model trains with Japanese language resources and achieves higher accuracy than English-centric models. |
BOSE: A Systematic Evaluation Method Optimized for Base Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing evaluation methods for large language models (LLMs) are inadequate to provide solid conclusions for key experiments such as data ablation and scaling law. |
| Approach: | They propose a method specifically designed to optimize the evaluation of base models by incorporating two innovations: In-Context Light-instruction Prompt and Blank-ppl for multi-choice tasks with candidate options. |
| Outcome: | The proposed method significantly improves stability and consistency of evaluations during pre-training and consistency between base and instruct models. |
DPGA-TextSyn: Differentially Private Genetic Algorithm for Synthetic Text Generation (2025.findings-acl)
Copied to clipboard
Zhonghao Sun, Zhiliang Tian, Yiping Song, Yuyi Si, Juhua Zhang, Minlie Huang, Kai Lu, Zeyu Xiong, Xinwang Liu, Dongsheng Li
| Challenge: | Existing methods to fine-tune large language models pose privacy risks . researchers have synthesized data with strong generation capabilities closed-source LLMs to alleviate this problem . |
| Approach: | They propose to combine general LLMs with genetic algorithm to produce relevant and diverse synthetic text under differential privacy constraints. |
| Outcome: | The proposed method significantly improves the performance of the model in downstream tasks while ensuring privacy. |
Semantic Aware Linear Transfer by Recycling Pre-trained Language Models for Cross-lingual Transfer (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) are increasingly incorporating multilingual capabilities, fueling the demand to transfer them into target language-specific models. |
| Approach: | They propose a novel cross-lingual transfer technique that recycles embeddings from target language Pre-trained Language Models to transmit deep representational strengths to LLMs. |
| Outcome: | The proposed technique outperforms existing methods in cross-lingual understanding setups and achieves faster convergence and lower loss during language adaptation. |
Boost, Disentangle, and Customize: A Robust System2-to-System1 Pipeline for Code Generation (2025.findings-acl)
Copied to clipboard
Kounianhua Du, Hanjing Wang, Jianxing Liu, Jizheng Chen, Xinyi Dai, Yasheng Wang, Ruiming Tang, Yong Yu, Jun Wang, Weinan Zhang
| Challenge: | Existing systems 2 methods for code generation are difficult to implement due to the complex hidden reasoning process and heterogeneous data distribution. |
| Approach: | They propose a framework that Boosts reasoning exploration via multi-agent collaboration and Disentangles heterogeneous data into specialized experts. |
| Outcome: | The proposed framework outperforms state-of-the-art methods on APPS and CodeContest benchmarks and achieves 73.8% accuracy on hard problems. |
On the Consistency of Commonsense in Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing evaluations of commonsense for large language models focus on downstream knowledge tasks, failing to probe whether LLMs truly understand and utilize knowledge or merely memorize it. |
| Approach: | They propose to automatically construct a large benchmark named CoCo which measures LLMs’ knowledge memorization, comprehension, and application and examines the consistency between these tasks. |
| Outcome: | The proposed benchmark systematically assesses LLMs’ knowledge memorization, comprehension, and application and examines the consistency between these tasks. |
Statement-Tuning Enables Efficient Cross-lingual Generalization in Encoder-only Models (2025.findings-acl)
Copied to clipboard
Ahmed Elshabrawy, Thanh-Nhi Nguyen, Yeeun Kang, Lihan Feng, Annant Jain, Faadil Abdullah Shaikh, Jonibek Mansurov, Mohamed Fazli Mohamed Imam, Jesus-German Ortiz-Barajas, Rendi Chevi, Alham Fikri Aji
| Challenge: | Large Language Models excel in zero-shot and few-shot tasks, but their architecture makes them difficult to use. |
| Approach: | They adapt Large Language Models (LLMs) for zero-shot generalization using Statement Tuning . they find encoders can achieve zero- shot cross-lingual generalization . |
| Outcome: | The proposed model generalizes well across languages while being more efficient. |
Evaluating Large Language Models for Confidence-based Check Set Selection (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models have shown promise in automating high-labor data tasks, but their tendency to answer despite uncertainty and their difficulty handling long input contexts robustly are key challenges for adoption. |
| Approach: | They propose to use LLMs to prioritize information needing human judgment to identify low-confidence outputs for human review through "check set selection" using social media monitoring, they define the "check sets" as a list of tweets escalated to the disaster manager when the LLM has the least confidence. |
| Outcome: | The proposed approach outperforms random-sample check set selection in disaster tweet classification. |
Training Multi-Modal LLMs through Dialogue Planning for HRI (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing approaches to enhance Multi-Modal Large Language Models (MLLMs) with explicit dialogue planning improves response accuracy and quality, and allows models trained in one language to transfer effectively to another. |
| Approach: | They propose an approach that enhances Multi-Modal Large Language Models with a novel explicit dialogue planning phase that allows agents to refine their understanding of ambiguous commands. |
| Outcome: | The proposed approach reduces hallucinations and improves task feasibility by fine-tuning and assessing Multi-Modal models in human-robot interaction scenarios. |
MVL-SIB: A Massively Multilingual Vision-Language Benchmark for Cross-Modal Topical Matching (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing multilingual vision-language (VL) benchmarks typically only cover a handful of languages, underscoring the need for evaluation data for low-resource languages. |
| Approach: | They propose a multilingual vision-language benchmark that evaluates cross-modal and text-only topical matching across 205 languages. |
| Outcome: | The proposed model performs better in cross-modal and text-only topical matching in lower-resource languages than the most multilingual benchmarks. |
The Rise of Darkness: Safety-Utility Trade-Offs in Role-Playing Dialogue Agents (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) demonstrate their utility in character simulations, but they pose a risk of generating unsafe content. |
| Approach: | They propose a method which dynamically adjusts safety-utility preferences based on the degree of risk coupling and guides the model to generate responses biased toward utility or safety. |
| Outcome: | The proposed method improves safety metrics while maintaining utility. |
SynGraph: A Dynamic Graph-LLM Synthesis Framework for Sparse Streaming User Sentiment Modeling (2025.findings-acl)
Copied to clipboard
| Challenge: | Traditional sentiment analysis methods focus on static reviews, failing to capture temporal relationship between user sentiment rating and textual content. |
| Approach: | They propose a dynamic graph-based framework that addresses data sparsity in streaming reviews. |
| Outcome: | The proposed framework reduces data sparsity by categorizing users into mid-tail, long-tail and extreme scenarios and incorporating LLM enhancements within a dynamic graph-based structure. |
Enhancing Tool Learning in Large Language Models with Hierarchical Error Checklists (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) have advanced natural language processing, but their effectiveness is often hampered by parameter mis-filling during tool calling. |
| Approach: | They propose a hierarchical tool error checklist framework to diagnose and mitigate tool-calling errors without relying on extensive real-world interactions. |
| Outcome: | The proposed framework improves parameter-filling accuracy and tool-calling success rates compared to baseline methods. |
A Large and Balanced Corpus for Fine-grained Arabic Readability Assessment (2025.findings-acl)
Copied to clipboard
| Challenge: | Texts above a student's readability level can lead to disengagement and disengagement . Developing readability models is crucial for improving literacy, language learning, and academic performance. |
| Approach: | They introduce the Balanced Arabic Readability Evaluation Corpus (BAREC) a large-scale, fine-grained dataset for Arabic readability assessment. |
| Outcome: | The proposed model outperforms existing methods in Arabic readability assessment. |
Can Medical Vision-Language Pre-training Succeed with Purely Synthetic Data? (2025.findings-acl)
Copied to clipboard
Che Liu, Zhongwei Wan, Haozhe Wang, Yinda Chen, Talha Qaiser, Chen Jin, Nikolay Burlutskiy, Fariba Yousefi, Rossella Arcucci
| Challenge: | Medical Vision-Language Pretraining (MedVLP) models typically require large-scale datasets with paired, high-quality image-text data. |
| Approach: | They propose to generate large-scale synthetic image-text pairs using off-the-shelf generative models . they propose to isolate model and training settings, focusing entirely from the data perspective. |
| Outcome: | The proposed pipeline outperforms models trained on real data by 3.8% on averaged AUC on zero-shot classification tasks. |
See the World, Discover Knowledge: A Chinese Factuality Evaluation for Large Vision Language Models (2025.findings-acl)
Copied to clipboard
Jihao Gu, Yingyao Wang, Pi Bu, Chen Wang, Ziming Wang, Tengtao Song, Donglai Wei, Jiale Yuan, Yingxiu Zhao, Yancheng He, Shilong Li, Jiaheng Liu, Meng Cao, Jun Song, Yingshui Tan, Xiang Li, Wenbo Su, Xiaoyong Zhu, Bo Zheng
| Challenge: | Existing models for large vision language models do not fully reflect their knowledge capacity and reliability, resulting in erroneous outputs that do not align with the image content or provide answers lacking knowledge evidence. |
| Approach: | They propose a Chinese-based benchmark for visual factuality across 8 major topics and 56 subtopics and a multi-hop question construction. |
| Outcome: | The proposed model decouples visual factuality into two parts: seeing the world and discovering knowledge. |
Argus: Benchmarking and Enhancing Vision-Language Models for 3D Radiology Report Generation (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing work on 3D radiograph report generation focuses on 2D images, but 3D medical images provide more comprehensive diagnostic information. |
| Approach: | They propose a comprehensive training recipe for building high-performing VLMs for 3DRRG using a publicly available 3D CT-report dataset. |
| Outcome: | The proposed model achieves superior performance across different model sizes and input 3D medical image resolutions. |
Resource-Friendly Dynamic Enhancement Chain for Multi-Hop Question Answering (2025.findings-acl)
Copied to clipboard
Binquan Ji, Haibo Luo, YifeiLu YifeiLu, Lei Hei, Jiaqi Wang, Tingjing Liao, Wang Lingyu, Shichao Wang, Feiliang Ren
| Challenge: | Existing approaches to solve multi-hop question answering challenges require multiple rounds of retrieval and iterative generation. |
| Approach: | They propose a framework that decomposes complex questions into coherent subquestions . it then iteratively refines these subquests through context-aware rewriting to generate effective query formulations. |
| Outcome: | The proposed framework performs on par with or surpasses state-of-the-art benchmarks while significantly reducing token consumption. |
Evaluating LLMs’ Assessment of Mixed-Context Hallucination Through the Lens of Summarization (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models generate coherent text and follow instructions across diverse tasks, but a critical challenge in scaling LLM applications is hallucination, where the generated content lacks factual grounding or deviates from the intended discourse context. |
| Approach: | They use summarization as a representative task to evaluate LLMs' capability in detecting mixed-context hallucinations, specifically distinguishing between factual and non-factual hallucinos. |
| Outcome: | The proposed model distinguishes between factual and non-factual hallucinations, and their performance bottlenecks. |
TUBA: Cross-Lingual Transferability of Backdoor Attacks in LLMs with Instruction Tuning (2025.findings-acl)
Copied to clipboard
Xuanli He, Jun Wang, Qiongkai Xu, Pasquale Minervini, Pontus Stenetorp, Benjamin I. P. Rubinstein, Trevor Cohn
| Challenge: | Despite the increasing support for multilingual capabilities, the impact of backdoor attacks on LLMs remains under-explored. |
| Approach: | They propose to use poisoned instructiontuning data to attack multilingual LLMs . their results show that more powerful models show increased susceptibility to transferable cross-lingual backdoor attacks . |
| Outcome: | The proposed attack is effective in models like BLOOM and GPT-4o with high success rates in more than 7 out of 12 languages. |
Eliciting Textual Descriptions from Representations of Continuous Prompts (2025.findings-acl)
Copied to clipboard
| Challenge: | Prior attempts to interpret continuous prompts relied on projecting individual tokens onto the vocabulary space, but this approach yielded arbitrary or contradictory text. |
| Approach: | They propose to project individual prompt tokens onto the model's vocabulary space and interpret each prompt token individually. |
| Outcome: | The proposed method yields accurate task descriptions and is more faithful as task performance increases. |
Mitigating Hallucination in Multimodal Large Language Model via Hallucination-targeted Direct Preference Optimization (2025.findings-acl)
Copied to clipboard
| Challenge: | Multimodal Large Language Models (MLLMs) are known to hallucinate, which limits their practical applications. |
| Approach: | They propose a method that uses three types of preference pairs to target hallucinations from their diverse forms and causes. |
| Outcome: | The proposed method surpasses most state-of-the-art methods and shows potential for further improvements. |
Review-Instruct: A Review-Driven Multi-Turn Conversations Generation Method for Large Language Models (2025.findings-acl)
Copied to clipboard
Jiangxu Wu, Cong Wang, Tianhuang Su, Lin Haozhi, JunYang JunYang, Zhangchao Zhangchao, Binqiang Pan, SongpanYang SongpanYang, Mingpeng Mingpeng, Kai Shi, Zixian Li
| Challenge: | Existing methods for generating multi-turn dialogue data struggle to ensure both diversity and quality in instructions. |
| Approach: | They propose a framework that synthesizes multi-turn conversations through an iterative "Ask-Respond-Review" process involving three agent roles: a Candidate, multiple Reviewers, and a Chairman. |
| Outcome: | The proposed framework synthesizes multi-turn conversations through an iterative "Ask-Respond-Review" process involving three agent roles: a Candidate, multiple Reviewers, and a Chairman. |
Why Uncertainty Estimation Methods Fall Short in RAG: An Axiomatic Analysis (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing UE methods cannot reliably estimate the correctness of LLM responses in Retrieval-Augmented Generation (RAG) . Existing methods generate low uncertainty values without considering relevance of context to query . |
| Approach: | They propose an axiomatic framework to identify deficiencies in existing UE methods and introduce five constraints that an effective UE method should meet after incorporating retrieved documents into the LLM’s prompt. |
| Outcome: | The proposed framework satisfies all the axioms and improves correlation between uncertainty estimates and correctness. |
EuroVerdict: A Multilingual Dataset for Verdict Generation Against Misinformation (2025.findings-acl)
Copied to clipboard
| Challenge: | a global issue that shapes public discourse shapes opinion and decision-making . many multilingual work has focused on claim verification rather than generating explanatory verdicts . |
| Approach: | They propose a multilingual dataset designed for verdict generation covering eight European languages. |
| Outcome: | The EuroVerdict dataset covers claims, manual verdicts, and supporting evidence . it is compared with other datasets in eight European languages . |
LoFTI: Localization and Factuality Transfer to Indian Locales (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models encode vast amounts of world knowledge acquired via training on large web-scale datasets crawled from the internet. |
| Approach: | They propose a benchmark that evaluates an LLM's contextual localization and factual text transfer capabilities by comparing it with a mixtral, Llama3.3-70B and GPT-4 model. |
| Outcome: | The proposed benchmark evaluates Mixtral, Llama3.3-70B, GPT-4 and two other Mixtral-based approaches. |
Hierarchical Retrieval with Evidence Curation for Open-Domain Financial Question Answering on Standardized Documents (2025.findings-acl)
Copied to clipboard
| Challenge: | standardized documents share similar formats and table structures . this similarity forces traditional RAG methods to misidentify near-duplicate text . |
| Approach: | They propose a hierarchical retrieval framework that performs hierarchically to reduce confusion among similar texts. |
| Outcome: | The proposed framework reduces confusion among similar documents by removing irrelevant passages . it generates complementary queries to collect missing information . |
GNN-RAG: Graph Neural Retrieval for Efficient Large Language Model Reasoning on Knowledge Graphs (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing approaches to retrieval-augmented generation (RAG) rely on costly LLM calls to generate relation paths or traverse the KG. |
| Approach: | They propose a framework that uses lightweight Graph Neural Networks to enhance retrieval. |
| Outcome: | The proposed framework outperforms existing methods on multi-hop and multi-entity questions. |
ASTRID - An Automated and Scalable TRIaD for the Evaluation of RAG-based Clinical Question Answering Systems (2025.findings-acl)
Copied to clipboard
| Challenge: | Current automated RAG metrics perform poorly in clinical and conversational use cases. |
| Approach: | They propose an automated and scaleable TRIaD for evaluating clinical QA systems leveraging Retrieval Augmented Generation (RAG) metric consisting of three metrics: Context Relevance (CR), Refusal Accuracy (RA), and Conversational Faithfulness (CF). |
| Outcome: | The proposed metric captures the faithfulness of a model’s response without penalising conversational elements and captures refusal to address questions outside of the system’s scope of practice. |
On Entity Identification in Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing work has shed light on the internal mechanisms of language models that can recall factual knowledge composed of entities and relations. |
| Approach: | They propose a framework analogous to clustering quality metrics to analyze the correspondence between entities and their mentions. |
| Outcome: | The proposed framework is analogous to clustering quality metrics. |
RAPID: Efficient Retrieval-Augmented Long Text Generation with Writing Planning and Information Discovery (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for knowledge-intensive long texts struggle with issues like hallucinations, topic incoherence, and significant latency. |
| Approach: | They propose a retrieval-augmented long text generation framework with writing P**lanning and I**nformation to address these challenges. |
| Outcome: | The proposed framework outperforms state-of-the-art methods on a freshWiki-2024 dataset. |
CHARPEVAL: Benchmarking Large Language Models’ Contextual Reasoning in Knowledge-Grounded Dialogue (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing benchmarks that evaluate the ability of Large Language Models (LLMs) to perform contextualized reasoning in knowledge-grounded dialogue scenarios are lacking. |
| Approach: | They propose a benchmark to evaluate the ability of Large Language Models to perform contextualized reasoning in knowledge-grounded dialogue scenarios. |
| Outcome: | The proposed benchmark shows that open-weight LLMs are ineffective at reasoning over discontinuous chunks of text across the input. |
Ask in Any Modality: A Comprehensive Survey on Multimodal Retrieval-Augmented Generation (2025.findings-acl)
Copied to clipboard
Mohammad Mahdi Abootorabi, Amirhosein Zobeiri, Mahdi Dehghani, Mohammadali Mohammadkhani, Bardia Mohammadi, Omid Ghahroodi, Mahdieh Soleymani Baghshah, Ehsaneddin Asgari
| Challenge: | Large Language Models (LLMs) suffer from hallucinations and outdated knowledge due to their reliance on static training data. |
| Approach: | They review training strategies, robustness enhancements, loss functions, and agent-based approaches and outline open challenges and future directions to guide research in this evolving field. |
| Outcome: | The proposed model improves accuracy and accuracy while integrating external dynamic information for improved factual grounding. |
Debate4MATH: Multi-Agent Debate for Fine-Grained Reasoning in Math (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing data annotation methods suffer from high annotation cost and lack of effective automatic validation. |
| Approach: | They propose a Fine-grained Multi-Agent Debate framework and a dataset that prompts multiple agents to debate and then a Multi-agent Debates Reward Model (MRM) to improve its mathematical reasoning capabilities. |
| Outcome: | The proposed model outperforms the state-of-the-art methods by 1.2% and 3.5% on a GSM8K dataset and 45.1% on the MATH dataset. |
Disambiguate First, Parse Later: Generating Interpretations for Ambiguity Resolution in Semantic Parsing (2025.findings-acl)
Copied to clipboard
| Challenge: | Natural language interfaces are often ambiguous, vague, or underspecified, giving rise to multiple valid interpretations. |
| Approach: | They propose a modular approach that resolves ambiguity using natural language interpretations before mapping them to logical forms. |
| Outcome: | The proposed approach improves interpretation coverage and generalizes across datasets with different annotation styles, database structures, and ambiguity types. |
The Anatomy of Evidence: An Investigation Into Explainable ICD Coding (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing systems that use explainability methods have limited their effectiveness due to a lack of annotated data. |
| Approach: | They propose match measures and highlight success and failure cases to improve automatic medical coding. |
| Outcome: | The proposed models show that ground truth evidence aligns with code descriptions to a certain degree. |
AVG-LLaVA: An Efficient Large Multimodal Model with Adaptive Visual Granularity (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing large multimodal models typically divide high-resolution images into multiple local images and a global image, leading to a large number of visual tokens. |
| Approach: | They propose an LMM that can adaptively select the appropriate visual granularity based on the input image and instruction. |
| Outcome: | The proposed model significantly reduces visual tokens and speeds up inference on 11 benchmarks. |
Word Form Matters: LLMs’ Semantic Reconstruction under Typoglycemia (2025.findings-acl)
Copied to clipboard
| Challenge: | Typoglycemia is a phenomenon where people can read words even when the middle letters of the words are scrambled. |
| Approach: | They propose a reliable metric to quantify the degree of semantic reconstruction and validate its effectiveness. |
| Outcome: | The proposed metric quantifies the degree of semantic reconstruction and validates its effectiveness. |
LLM-based Translation Inference with Iterative Bilingual Understanding (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing studies show that the ability of large language models to generate contextual understanding of the sentence can degrade translation quality. |
| Approach: | They propose a method that generates contextual understanding for both source and target languages separately. |
| Outcome: | The proposed method outperforms strong comparison methods in multiple domains. |
Vulnerability of Text-to-Image Models to Prompt Template Stealing: A Differential Evolution Approach (2025.findings-acl)
Copied to clipboard
Yurong Wu, Fangwen Mu, Qiuhong Zhang, Jinjing Zhao, Xinrun Xu, Lingrui Mei, Yang Wu, Lin Shi, Junjie Wang, Zhiming Ding, Yiwei Wang
| Challenge: | Prompt trading has emerged as a significant intellectual property concern in recent years, where vendors entice users by showcasing sample images before selling prompt templates that can generate similar images. |
| Approach: | They propose a prompt-stealing benchmark consisting of 50 templates and 450 images organized into Easy and Hard difficulty levels. |
| Outcome: | The proposed method outperforms baseline methods with an average improvement of over 10%. |
mStyleDistance: Multilingual Style Embeddings and their Evaluation (2025.findings-acl)
Copied to clipboard
| Challenge: | Multilingual StyleDistance embeddings are useful for stylistic analysis and style transfer, but they only exist for English. |
| Approach: | They propose a method that can generate style embeddings in new languages using synthetic data and a contrastive loss. |
| Outcome: | The proposed method outperforms existing style embeddings on these benchmarks and generalizes well to unseen features and languages. |
SeqMMR: Sequential Model Merging and LLM Routing for Enhanced Batched Sequential Knowledge Editing (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing research has demonstrated strong performance in single-instance or few-instantial sequential editing and one-time massive editing scenarios, but the batched sequential editing paradigm remains a significant challenge. |
| Approach: | They propose a framework for batched sequential knowledge editing that leverages **SeqMMR** and a model router to merge parameters from current batch-edited models with those of their predecessors. |
| Outcome: | The proposed framework iteratively merges parameters from current batch-edited models with those of their predecessors, ensuring that newly emerging knowledge is integrated while mitigating the forgetting of previously edited knowledge. |
ReflectEvo: Improving Meta Introspection of Small LLMs by Learning Self-Reflection (2025.findings-acl)
Copied to clipboard
Jiaqi Li, Xinyi Dong, Yang Liu, Zhizhuo Yang, Quansen Wang, Xiaobo Wang, Song-Chun Zhu, Zixia Jia, Zilong Zheng
| Challenge: | ReflectEvo-460k is a large-scale, comprehensive, self-generated reflection dataset with broadened instructions and diverse multi-domain tasks. |
| Approach: | They propose a pipeline that iteratively generates self-reflection for self-training and a large-scale reflection dataset with broadened instructions and diverse multi-domain tasks. |
| Outcome: | The proposed pipeline improves Llama-3 reasoning ability by up to 71.2% and Mistral by upto 44.4%. |
MAGIC-VQA: Multimodal And Grounded Inference with Commonsense Knowledge for Visual Question Answering (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing Large Vision-Language Models (LVLMs) lack integrated commonsense knowledge . lack of integrated common knowledge limits their robustness and accuracy in VQA . |
| Approach: | They propose a framework to enhance multimodal inference by integrating commonsense reasoning. |
| Outcome: | MAGIC-VQA improves comprehensive benchmark datasets, surpassing existing models in tasks requiring advanced commonsense reasoning. |
Automatic Transmission for LLM Tiers: Optimizing Cost and Accuracy in Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) are powerful tools for a wide range of natural language tasks. |
| Approach: | They propose an LLM automatic transmission framework that automatically selects LLM tiers without training. |
| Outcome: | The proposed framework achieves superior performance while reducing costs. |
Low-Rank Interconnected Adaptation across Layers (2025.findings-acl)
Copied to clipboard
| Challenge: | Low-rank adaptation (LoRA) is a parameter-efficient fine-tuning (PEFT) method that learns weight updates W = AB for pretrained weights W through low-rank adapters A and B. |
| Approach: | They propose a low-rank interconnected adaptation across layers method that introduces an interconnected framework with locally shared A and globally shared B experts. |
| Outcome: | The proposed method improves expressiveness across domains and modalities and enables higher-rank W with equal or fewer parameters. |
GaRAGe: A Benchmark with Grounding Annotations for RAG Evaluation (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have shown consistent improvements across many tasks requiring natural language understanding, coding, mathematical or logical reasoning . |
| Approach: | They propose to use GaRAGe to evaluate whether LLMs can identify relevant grounding when generating RAG answers. |
| Outcome: | The proposed model over-summarises rather than ground answers strictly on annotated relevant passages, or deflects when no relevant grounding is available. |
Change Entity-guided Heterogeneous Representation Disentangling for Change Captioning (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing approaches to describe differences between two images are highly challenging due to distractors such as illumination and viewpoint changes. |
| Approach: | They propose a change-entity-guided disentanglement network that explicitly learns difference representations while mitigating the impact of distractors. |
| Outcome: | The proposed method outperforms existing methods on CLEVR-Change, CLE VR-DC and Spot-the-Diff datasets and achieves state-of-the art performance. |
RAG-RewardBench: Benchmarking Reward Models in Retrieval Augmented Generation for Preference Alignment (2025.findings-acl)
Copied to clipboard
Zhuoran Jin, Hongbang Yuan, Tianyi Men, Pengfei Cao, Yubo Chen, Jiexin Xu, Huaijun Li, Xiaojian Jiang, Kang Liu, Jun Zhao
| Challenge: | Existing retrieval augmented language models often overlook effective alignment with human preferences. |
| Approach: | They propose a benchmark to evaluate RMs in retrieval augmented language models . they incorporate 18 RAG subsets, six retrievers, and 24 RALMs to increase diversity . |
| Outcome: | The proposed benchmark combines 18 RAG subsets, six retrievers, and 24 RALMs to increase diversity of data sources. |
Generate, Discriminate, Evolve: Enhancing Context Faithfulness via Fine-Grained Sentence-Level Self-Evolution (2025.findings-acl)
Copied to clipboard
Kun Li, Tianhua Zhang, Yunxiang Li, Hongyin Luo, Abdalla Mohamed Salama Sayed Moustafa, Xixin Wu, James R. Glass, Helen M. Meng
| Challenge: | Existing methods to improve context faithfulness in large language models are either inadequate or overlook the potential for self-improvement. |
| Approach: | They propose a framework that enhances context faithfulness through fine-grained sentence-level optimization. |
| Outcome: | Experiments on ASQA and ConFiQA datasets show that GenDiE surpasses baselines in faithfulness and correctness and exhibits robust performance for domain adaptation. |
PAM: Paraphrase AMR-Centric Evaluation Metric (2025.findings-acl)
Copied to clipboard
| Challenge: | Current evaluation metrics for paraphrase generation are based on borrowed metrics from text-to-text tasks . this is not ideal for paraphrasing as we typically want variation in the lexicon while persisting semantics. |
| Approach: | They propose a Paraphrase AMR-Centric Evaluation Metric that uses AMR graphs extracted from the input text to evaluate paraphrases. |
| Outcome: | The proposed evaluation metric improves on different semantic textual similarity datasets on paraphrases with human semantic scores. |
VP-MEL: Visual Prompts Guided Multimodal Entity Linking (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for multimodal entity linking rely on mention words as retrieval cues, which limits their ability to effectively utilize information from both images and text. |
| Approach: | They propose a visual prompt-guided multimodal entity linking task for a text-image pair . they propose VPWiki to facilitate this task and a framework to capture latent information. |
| Outcome: | The proposed framework outperforms baseline methods on a VPWiki dataset. |
FADE: Why Bad Descriptions Happen to Good Features (2025.findings-acl)
Copied to clipboard
Bruno Puri, Aakriti Jain, Elena Golimblevskaia, Patrick Kahardipraja, Thomas Wiegand, Wojciech Samek, Sebastian Lapuschkin
| Challenge: | Recent advances in mechanistic interpretability have highlighted the potential of automating interpretability pipelines in analyzing the latent representations within LLMs. |
| Approach: | They propose a framework for automatically evaluating feature-to-description alignment that measures alignment across four key metrics and quantifies the causes of misalignment. |
| Outcome: | The proposed framework evaluates alignment across four key metrics and quantifies the causes of misalignment between features and descriptions. |
In the LLM era, Word Sense Induction remains unsolved (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing word sense induction methods rely on predefined sense inventories of arbitrary granularity and require labor intensive sense-annotated data. |
| Approach: | They propose to evaluate an LLM-based word sense induction method for English . they use data augmentation sources, corpus and lexicon, and semi-supervised scenarios to augment the data. |
| Outcome: | The proposed method surpasses the previous SOTA system on a English set by 3.3%. |
Navigating the Political Compass: Evaluating Multilingual LLMs across Languages and Nationalities (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) are ubiquitous in today’s technological landscape, boasting a plethora of applications, and even endangering human jobs in complex and creative fields. |
| Approach: | They evaluate the political bias of 15 multilingual LLMs using the Political Compass Test and assign a nationality to each model. |
| Outcome: | The models on the 50 most populous countries and their official languages exhibit political bias. |
Who Can Withstand Chat-Audio Attacks? An Evaluation Benchmark for Large Audio-Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing research focused on model-specific adversarial methods, but real-world applications demand a more generalizable approach to audio adversarials. |
| Approach: | They propose a Chat-Audio Attacks benchmark to evaluate LALMs' robustness . they propose standard evaluation, GPT-4o-based evaluation and human evaluation . |
| Outcome: | The proposed benchmark aims to explore the robustness of six state-of-the-art LALMs with voice interaction capabilities. |
Beyond the Tip of Efficiency: Uncovering the Submerged Threats of Jailbreak Attacks in Small Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Small language models (SLMs) have become increasingly prominent in the deployment on edge devices due to their high efficiency and low computational cost. |
| Approach: | They evaluate the security performance of 13 state-of-the-art small language models under various jailbreak attacks. |
| Outcome: | The proposed methods demonstrate that SLMs are quite susceptible to jailbreak attacks and some are even vulnerable to harmful prompts. |
EMRs2CSP : Mining Clinical Status Pathway from Electronic Medical Records (2025.findings-acl)
Copied to clipboard
| Challenge: | Current studies focus on extracting tests or treatments when constructing clinical pathways, neglecting the patient's symptoms and diagnosis. |
| Approach: | They propose a novel clinical pathway representation: the clinical status pathway and a pipeline framework for extracting clinical status from electronic medical records. |
| Outcome: | The proposed framework improves extraction accuracy by modeling diagnostic and treatment processes and demonstrates significant improvements on medical question-answering and decision-support tasks. |
A Law Reasoning Benchmark for LLM with Tree-Organized Structures including Factum Probandum, Evidence and Experiences (2025.findings-acl)
Copied to clipboard
Jiaxin Shen, Jinan Xu, Huiqi Hu, Luyi Lin, Guoyang Ma, Fei Zheng, Fandong Meng, Jie Zhou, Wenjuan Han
| Challenge: | a recent study focuses on generating impartial and interpretable judicial judgments based on established criminal fact. |
| Approach: | They propose a law reasoning schema enriched with hierarchical factum probandum, evidence, and implicit experience that enables public scrutiny and preventing bias. |
| Outcome: | The proposed schema enables public scrutiny and prevents bias in the "Intelligent Court" it employs a suite of legal analysis tools to address the challenge task. |
Libra: Leveraging Temporal Images for Biomedical Radiology Analysis (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for radiology report generation rely on single-image analysis or rule-based heuristics to process multiple images. |
| Approach: | They propose a temporal-aware MLLM tailored for chest X-ray report generation that combines a radiology-specific image encoder with a novel Temporal Alignment Connector. |
| Outcome: | The proposed model sets new standards in clinical relevance and lexical accuracy on the MIMIC-CXR dataset. |
Stereotype Detection as a Catalyst for Enhanced Bias Detection: A Multi-Task Learning Approach (2025.findings-acl)
Copied to clipboard
| Challenge: | a new study addresses bias and stereotypes in language models by exploring how learning them together improves performance. |
| Approach: | They propose a dataset for bias and stereotype detection that integrates religion, gender, socio-economic status, race, profession, and others. |
| Outcome: | The proposed dataset compares encoder-only models and fine-tuned decoder- only models . the results show that learning stereotypes together improves bias detection . |
Filling the Temporal Void: Recovering Missing Publication Years in the Project Gutenberg Corpus Using LLMs (2025.findings-acl)
Copied to clipboard
| Challenge: | Currently, there is no publicly available corpus for diachronic text analysis due to the lack of accurate temporal metadata. |
| Approach: | They propose to add missing temporal metadata to the Gutenberg corpus by using open web, Wikipedia, and Open Library API sources. |
| Outcome: | The proposed corpus includes 53,774 books with a total of 3.8 billion tokens in 11 languages, produced between 1600 and 2000. |
ExpliCa: Evaluating Explicit Causal Reasoning in Large Language Models (2025.findings-acl)
Copied to clipboard
Martina Miliani, Serena Auriemma, Alessandro Bondielli, Emmanuele Chersoni, Lucia Passaro, Irene Sucameli, Alessandro Lenci
| Challenge: | Large Language Models (LLMs) are increasingly used in tasks requiring interpretive and inferential accuracy. |
| Approach: | They propose a dataset for evaluating Large Language Models in explicit causal reasoning that integrates causal and temporal relations presented in different linguistic orders and explicitly expressed by linguistic connectives. |
| Outcome: | The proposed model performs better than existing models in the domain of causal reasoning. |
Are Dialects Better Prompters? A Case Study on Arabic Subjective Text Classification (2025.findings-acl)
Copied to clipboard
| Challenge: | specialized fine-tuned models with Arabic and Arabizi scripts achieve the best results . transfer learning demonstrated limited effectiveness, despite ability to adapt to other regionally specific varieties . |
| Approach: | They evaluate the performance of 12 open source LLMs for Arabic and Arabizi scripts . they highlight the impact of Arabic-centric LLM fine-tuning and prompt design on models . |
| Outcome: | The results show that specialized models with Arabic and Arabizi scripts achieve the best results . |
Natural Logic at the Core: Dynamic Rewards for Entailment Tree Generation (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing approaches to generating entailment trees lack logical consistency . static reward structures or intricate dependencies within multi-step reasoning are often ignored . |
| Approach: | They propose a method that integrates natural logic principles into reinforcement learning to guide entailment tree generation. |
| Outcome: | Experiments on EntailmentBank show that the proposed method improves interpretability and generalization. |
R.R.: Unveiling LLM Training Privacy through Recollection and Ranking (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing privacy attacks focus on membership inference or data extraction, but reconstructing specific personally identifiable information (PII) in training data remains challenging. |
| Approach: | They propose a two-step privacy stealing attack that enables attackers to reconstruct PII entities from scrubbed training data where the PI I entities have been masked. |
| Outcome: | The proposed attack can reconstruct PII entities from scrubbed training data where the PI I entities have been masked. |
Nested-Refinement Metamorphosis: Reflective Evolution for Efficient Optimization of Networking Problems (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) excel in network algorithm design but suffer from inefficient iterative coding and high computational costs. |
| Approach: | They propose a method to iteratively refine task descriptions and metamorphosis on algorithms to generate more effective solutions. |
| Outcome: | Experimental results show that Nested-Refinement Metamorphosis outperforms state-of-the-art approaches in performance and efficiency. |
MC-MKE: A Fine-Grained Multimodal Knowledge Editing Benchmark Emphasizing Modality Consistency (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing benchmarks for knowledge editing in multimodal large language models focus on limited scenarios due to the lack of rigorous definition of multimodal knowledge. |
| Approach: | They propose a decomposed definition of multimodal knowledge and a benchmark to evaluate it. |
| Outcome: | The proposed method reveals that it is difficult to define multimodal knowledge editing in LLMs. |
Visualising Policy-Reward Interplay to Inform Zeroth-Order Preference Optimisation of Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | ZOPrO is a novel algorithm designed for *Preference Optimisation* in large language models. |
| Approach: | They propose a ZO algorithm designed for *Preference Optimisation* in LLMs that uses function evaluations instead of gradients to reduce memory usage. |
| Outcome: | The proposed method improves reward signals while achieving convergence times comparable to first-order methods. |
Metaphor and Large Language Models: When Surface Features Matter More than Deep Understanding (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing studies on metaphor processing have focused on single datasets and specific task settings, often using artificially constructed data through lexical replacement. |
| Approach: | They propose to evaluate the capabilities of Large Language Models (LLMs) in metaphor interpretation across multiple datasets, tasks, and prompt configurations. |
| Outcome: | The proposed frameworks are more realistic and efficient than current models and are more efficient than existing models. |
AskQE: Question Answering as Automatic Evaluation for Machine Translation (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing MT error detection and quality estimation (QE) techniques do not address this practical scenario. |
| Approach: | They propose a question generation and answering framework that detects critical MT errors and provides actionable feedback to help users decide whether to accept or reject MT outputs even without the knowledge of the target language. |
| Outcome: | The proposed framework has higher Kendall’s Tau correlation and decision accuracy with human ratings compared to other QE metrics. |
ExPerT: Effective and Explainable Evaluation of Personalized Long-Form Text Generation (2025.findings-acl)
Copied to clipboard
| Challenge: | Evaluating personalized text generated by large language models is challenging, as only the LLM user, i.e. prompt author, can reliably assess the output. |
| Approach: | They propose an explainable reference-based evaluation framework that leverages an LLM to extract atomic aspects and their evidences from the generated and reference texts, match the aspects, and evaluate their alignment based on content and writing style. |
| Outcome: | The proposed framework achieves a 7.2% improvement in alignment with human judgments compared to the state-of-the-art evaluation methods. |
Bridging Intuitive Associations and Deliberate Recall: Empowering LLM Personal Assistant with Graph-Structured Long-term Memory (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs)-based personal assistants struggle to capture entity relationships and handle multiple intents effectively. |
| Approach: | They propose a graph-structured memory framework that mimics human cognitive processes and an event-centric memory graph. |
| Outcome: | The proposed framework outperforms retrieval and QA methods across long-term dialogue benchmarks and enables more human-like memory systems. |
Each graph is a new language: Graph Learning with LLMs (2025.findings-acl)
Copied to clipboard
| Challenge: | Natural language is used to describe graphs, but graph descriptions become verbose and only relying on attribute embeddings limits LLM’s ability to capture adequate graph structural information. |
| Approach: | They propose a graph-defined language for large language model that translates the graph into a corpus instead of graph descriptions and pre-trains LLMs on this corpus to adequately understand the graph. |
| Outcome: | Experiments on five datasets show that the proposed framework outperforms description-based and embedding-based baselines by efficiently modeling different orders of neighbors. |
100-LongBench: Are de facto Long-Context Benchmarks Literally Evaluating Long-Context Ability? (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing benchmarks for long-context capability are too synthetic and do not represent the real world usage of LLMs. |
| Approach: | They propose a length-controllable, real-life reflective benchmark that disentangles baseline knowledge from long-context capabilities. |
| Outcome: | Experiments show that the proposed benchmarks disentangle baseline knowledge from long-context capabilities. |
Multimodal Fusion and Coherence Modeling for Video Topic Segmentation (2025.findings-acl)
Copied to clipboard
| Challenge: | Traditional video topic segmentation methods struggle to discern topical transitions . supervised approaches have improved performance on video action or scene segmentation . |
| Approach: | They propose a new task for video topic segmentation that enhances multimodality alignment and fusion by exploring different architectures using Cross-Attention and Mixture of Experts. |
| Outcome: | The proposed model improves on educational videos, in the form of lectures . it combines cross-attention and mixture of experts to strengthen multimodality alignment and fusion . |
Are Your LLMs Capable of Stable Reasoning? (2025.findings-acl)
Copied to clipboard
Junnan Liu, Hongwei Liu, Linchen Xiao, Ziyi Wang, Kuikun Liu, Songyang Gao, Wenwei Zhang, Songyang Zhang, Kai Chen
| Challenge: | Existing evaluation protocols and metrics do not capture the full spectrum of LLM capabilities, especially in complex reasoning tasks. |
| Approach: | They propose a new evaluation metric that continuously assesses model performance across multiple sampling attempts, quantifying both the model’s potential capabilities and operational consistency. |
| Outcome: | The proposed evaluation metric measures model performance across multiple sampling attempts and provides comprehensive insights into their potential capabilities and operational consistency. |
FANNO: Augmenting High-Quality Instruction Data with Open-Sourced LLMs Only (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent studies explore approaches to synthesize instruction data with open-sourced LLMs but require high-quality human-crafted seed data. |
| Approach: | They propose an end-to-end framework to synthesize high-quality instruction data with open-sourced LLMs and sampled unlabeled documents, eliminating the need for seed data. |
| Outcome: | The proposed framework synthesizes high-quality instruction data with open-sourced LLMs and sampled unlabeled documents, eliminating the need for seed data. |
JEBS: A Fine-grained Biomedical Lexical Simplification Task (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing systems for simplification of complex medical terms are limited in the scope of their topics and require massive cost and effort to keep up with the latest research. |
| Approach: | They propose a fine-grained lexical simplification task and dataset to enable more targeted development and evaluation of systems for replacing or explaining complex biomedical terms. |
| Outcome: | The proposed task and dataset pave the way for development and evaluation of systems for replacing or explaining complex biomedical terms. |
Multi-Hop Reasoning for Question Answering with Hyperbolic Representations (2025.findings-acl)
Copied to clipboard
| Challenge: | a rigorous and detailed comparison of the two spaces for multi-hop reasoning is lacking. |
| Approach: | They compare the capacity of hyperbolic space versus Euclidean space in multi-hop reasoning . they use an encoder-decoder model to integrate hyperbolical representations with a knowledge graph . |
| Outcome: | The proposed model outperforms the Euclidean space in multi-hop reasoning. |
Look & Mark: Leveraging Radiologist Eye Fixations and Bounding boxes in Multimodal Large Language Models for Chest X-ray Report Generation (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent advances in multimodal Large Language Models (LLMs) have significantly enhanced the automation of medical image analysis, but still suffer from hallucinations and clinically significant errors. |
| Approach: | They propose a grounding fixation strategy that integrates radiologist eye fixations and bounding box annotations into the LLM prompting framework. |
| Outcome: | The proposed model improves performance without retraining across domain-specific and general-purpose models and achieves an 87.3% clinical average performance. |
Hatevolution: What Static Benchmarks Don’t Tell Us (2025.findings-acl)
Copied to clipboard
| Challenge: | Language evolves rapidly following social dynamics and cultural shifts. |
| Approach: | They empirically evaluate the robustness of 20 language models across two evolving hate speech experiments and propose time-sensitive benchmarks for their work. |
| Outcome: | The proposed model evaluations show that the language models are misaligned between static and time-sensitive evaluations. |
Tag-Instruct: Controlled Instruction Complexity Enhancement through Structure-based Augmentation (2025.findings-acl)
Copied to clipboard
| Challenge: | High-quality instruction data is crucial for developing large language models (LLMs), yet existing approaches struggle to effectively control instruction complexity. |
| Approach: | They propose a framework that compresses instructions into a compact tag space and enhances complexity through RL-guided tag expansion. |
| Outcome: | The proposed framework outperforms existing methods in the evaluation of instruction complexity augmentation and semantic compression of text into a compact tag space. |
Code-SPA: Style Preference Alignment to Large Language Models for Effective and Robust Code Debugging (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) have demonstrated impressive capabilities in coding tasks like code generation and debugging. |
| Approach: | They propose a method which aligns noisy code with the well-structured style familiar to LLMs, mitigating the impact of stylistic inconsistencies. |
| Outcome: | The proposed method improves debugging performance on poorly styled code across the HumanEval, MBPP and EvalPlus datasets. |
Open-World Authorship Attribution (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing benchmarks for large language models do not evaluate their performance in academic research . authors aim to identify authors from anonymous text without additional information . |
| Approach: | They propose a benchmark to quantitatively assess LLMs' ability to infer author from text . they propose 'open-world' authorship attribute' to be a two-stage framework . |
| Outcome: | The proposed approach achieves 60.7% accuracy and 44.3% accuracy in two stages. |
What is in a name? Mitigating Name Bias in Text Embedding Similarity via Anonymization (2025.findings-acl)
Copied to clipboard
| Challenge: | Text-embedding models often exhibit name bias due to data on which they are trained. |
| Approach: | They propose a method to mitigate name bias in text-embedding models by removing references to names from the text. |
| Outcome: | The proposed approach achieves significant performance gains on three downstream NLP tasks involving embedding similarities. |
BenNumEval: A Benchmark to Assess LLMs’ Numerical Reasoning Capabilities in Bengali (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) excel in general-purpose tasks but struggle with numerical reasoning, especially in low-resource languages like Bengali. |
| Approach: | They propose a benchmark to assess LLMs on numerical reasoning tasks in Bengali. |
| Outcome: | The proposed benchmark assesses LLMs on numerical reasoning tasks in Bengali. |
LLM Agents for Coordinating Multi-User Information Gathering (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent large language models (LLMs) are becoming a crucial building block in developing automated agents that can assist human users with complex tasks. |
| Approach: | They introduce PeopleJoin, a benchmark for evaluating LM-mediated collaborative problem solving. |
| Outcome: | The proposed benchmarks are adapted from existing benchmarks for database question answering and multi-document summarization. |
C2KD: Cross-layer and Cross-head Knowledge Distillation for Small Language Model-based Recommendation (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) show promise but their size and high inference costs limit deployment on resource-constrained devices. |
| Approach: | They propose a framework to transfer task-relevant knowledge from two complementary dimensions to Large Language Models (LLMs) Large Language models (LLMS) have demonstrated great potential in sequential recommendation tasks . |
| Outcome: | Extensive experiments across diverse model families show that the proposed framework achieves competitive performance compared to LLMs. |
Sign2Vis: Automated Data Visualization from Sign Language (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods to translate natural language descriptions into visualization queries focus on spoken languages, not sign languages. |
| Approach: | They propose a sign language interface that enables the DHH community to engage more fully with data analysis. |
| Outcome: | The proposed interface can be used by the deaf and hard-of-hearing community. |
Transparentize the Internal and External Knowledge Utilization in LLMs with Trustworthy Citation (2025.findings-acl)
Copied to clipboard
| Challenge: | citation generation and retrieval-augmented generation are still lacking in large language models due to hallucinations. |
| Approach: | They propose a retrieval-augmented citation generation task that requires models to generate citations considering both external and internal knowledge while providing trustworthy references. |
| Outcome: | The proposed method achieves better performance across scenarios compared to baselines . retrieval quality, question types, and model knowledge influence trustworthiness . |
JARVIS-VLA: Post-Training Large-Scale Vision Language Models to Play Visual Games with Keyboards and Mouse (2025.findings-acl)
Copied to clipboard
| Challenge: | Visual Language Action models have shown promise in decision-making tasks, but have been neglected in previous work . |
| Approach: | They propose a new paradigm for visual language action models that enhances the foundation model prior to action-specific tuning by first post-training it on a curated set of visual and linguistic tasks using self-supervised learning. |
| Outcome: | The proposed model outperforms the best agent baseline on a diverse set of atomic tasks and surpasses imitation learning-based policies in Minecraft. |
Generative Frame Sampler for Long Video Understanding (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing video large language models (LMMs) employ an impedance of thousands of frames to understand long videos. |
| Approach: | They propose a plug-and-play module integrated with VideoLLMs to facilitate efficient lengthy video perception. |
| Outcome: | The proposed module boosts the performance of open-source VideoLLMs and proprietary assistants on long-form video benchmarks. |
Annotating the Annotators: Analysis, Insights and Modelling from an Annotation Campaign on Persuasion Techniques Detection (2025.findings-acl)
Copied to clipboard
Davide Bassi, Dimitar Iliyanov Dimitrov, Bernardo D’Auria, Firoj Alam, Maram Hasanain, Christian Moro, Luisa Orrù, Gian Piero Turchi, Preslav Nakov, Giovanni Da San Martino
| Challenge: | Existing annotation campaigns based on heuristic guidelines have not been thoroughly discussed. |
| Approach: | They propose a probabilistic model for optimizing intervention scheduling to reduce the cost of an expert oversight in annotation tasks. |
| Outcome: | The proposed model advocates for an expert oversight in annotation tasks and periodic quality audits to reduce costs. |
On the Generalization vs Fidelity Paradox in Knowledge Distillation (2025.findings-acl)
Copied to clipboard
| Challenge: | Knowledge distillation (KD) is a key technique for compressing large language models into smaller ones while preserving performance. |
| Approach: | They propose to use knowledge distillation to compress large language models into smaller ones while preserving performance. |
| Outcome: | The proposed technique improves the performance of smaller models by 10% while providing only marginal benefits for larger models. |
BEDAA: Bayesian Enhanced DeBERTa for Uncertainty-Aware Authorship Attribution (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for authorship attribution struggle with trustworthiness and interpretability across domains, languages, and stylistic variations. |
| Approach: | They propose a Bayesian-Enhanced DeBERTa framework that integrates Bayes' reasoning with transformer-based language models to enable uncertainty-aware authorship attribution. |
| Outcome: | The proposed framework achieves 19.69% improvement in F1-score across multiple authorship attribution tasks, including binary, multiclass, and dynamic authorship detection. |
Benchmarking the Benchmarks: Reproducing Climate-Related NLP Tasks (2025.findings-acl)
Copied to clipboard
| Challenge: | a recent study examines the use of climate-related natural language processing (NLP) for climate-relevant tasks. |
| Approach: | They perform a reproducibility study on 8 tasks and 29 datasets, testing 6 models. |
| Outcome: | The proposed models are based on 8 tasks and 29 datasets. |
Exploring Supervised Approaches to the Detection of Anthropomorphic Language in the Reporting of NLP Venues (2025.findings-acl)
Copied to clipboard
| Challenge: | anthropomorphisms are used to describe technical contributions to AI . however, they also give potential for incorrect assumptions about LLMs' capacities. |
| Approach: | They undertake a corpus annotation of one year of ACL abstracts and news articles from the same period and train a regression classifier based on BERT to identify anthropomorphic language. |
| Outcome: | The proposed method can automatically label abstracts for their degree of anthropomorphism based on their corpus and reporting on diachronic and inter-venue findings. |
PersonaLens: A Benchmark for Personalization Evaluation in Conversational AI Assistants (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing personalization benchmarks focus on chit-chat, non-conversational tasks, or narrow domains, failing to capture complexities of personalized task-oriented assistance. |
| Approach: | They propose a benchmark to evaluate personalization in task-oriented AI assistants . the benchmark features user profiles equipped with rich preferences and interaction histories . |
| Outcome: | The proposed benchmark features user profiles equipped with rich preferences and interaction histories . it also features a judge agent and user agent that employs the LLM-as-a-Judge paradigm . |
iAgent: LLM Agent as a Shield between User and Recommender Systems (2025.findings-acl)
Copied to clipboard
Wujiang Xu, Yunxiao Shi, Zujie Liang, Xuying Ning, Kai Mei, Kun Wang, Xi Zhu, Min Xu, Yongfeng Zhang
| Challenge: | Traditional recommender systems focus on the user-platform paradigm, where users are directly exposed under the control of the platform's recommendation algorithms. |
| Approach: | They propose a user-agent-platform paradigm where agent serves as the protective shield between user and recommender system that enables indirect exposure. |
| Outcome: | The proposed model improves 16.6% over baselines on four datasets and mitigates echo chamber effects and reduces model bias in disadvantaged users. |
FactLens: Benchmarking Fine-Grained Fact Verification (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have shown impressive capability in language generation and understanding, but their tendency to hallucinate and produce factually incorrect information remains a key limitation. |
| Approach: | They propose a benchmark to evaluate fine-grained fact verification where claims are broken down into smaller sub-claims for individual verification. |
| Outcome: | The proposed model enables more precise identification of inaccuracies, improved transparency, and reduced ambiguity in evidence retrieval. |
Process-based Self-Rewarding Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods to reward LLMs' outputs are not effective in mathematical reasoning scenarios and may lead to a decline in performance. |
| Approach: | They propose a process-based self-rewarding pipeline that integrates long-thought reasoning, step-wise LLM-as-a-Judge, and step- wise preference optimization within the existing paradigm. |
| Outcome: | The proposed model improves the performance of Large Language Models on multiple mathematical reasoning benchmarks and shows that it can surpass human capabilities. |
The Devil Is in the Word Alignment Details: On Translation-Based Cross-Lingual Transfer for Token Classification Tasks (2025.findings-acl)
Copied to clipboard
| Challenge: | Translation-based strategies for cross-lingual transfer XLT include label projection . word aligners (WAs) are commonly used for label projection, but low-level design decisions for using them have not been investigated . |
| Approach: | They revisit word aligners (WAs) for label projection and propose a new projection strategy that outperforms WAs. |
| Outcome: | The proposed projection strategy outperforms marker-based methods in token classification tasks. |
ShieldHead: Decoding-time Safeguard for Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent advances in LLM-based moderation methods have demonstrated remarkable promise in identifying safety risks associated with both inputs and outputs in human-AI interactions. |
| Approach: | They propose to learn a classification head on the last-layer hidden states of a dialogue model and use it to detect harmful content. |
| Outcome: | The proposed framework is 300 faster (**1ms**) than previous LLM-based moderation models with 99% less parameters than LlamaGuard. |
A Survey on Proactive Defense Strategies Against Misinformation in Large Language Models (2025.findings-acl)
Copied to clipboard
Shuliang Liu, Hongyi Liu, Aiwei Liu, Duan Bingchen, Zheng Qi, Yibo Yan, He Geng, Peijie Jiang, Jia Liu, Xuming Hu
| Challenge: | Existing methods for detection of misinformation generated by large language models fail to mitigate societal risks . authors propose a paradigm shift from passive detection to anticipatory mitigation strategies . existing defenses remain reactionary in an era demanding proactive defense, authors say . |
| Approach: | They propose a three-pillar approach to prevent misinformation by fortifying integrity of training data and inference reliability by embedding self-corrective mechanisms during reasoning. |
| Outcome: | The proposed framework improves existing methods in misinformation prevention by 63% . it demonstrates that existing methods exhibit false negative rates against misinformation . |
Smotrom tvoja på ander drogoj verden! Resurrecting Dead Pidgin with Generative Models: Russenorsk Case Study (2025.findings-acl)
Copied to clipboard
| Challenge: | Russenorsk is a pidgin language used in trade interactions between Russian and Norwegian speakers. |
| Approach: | They propose to analyze Russenorsk's lexicon using modern large language models . they construct a structured dictionary of the language, grouped by synonyms and word origins . |
| Outcome: | The proposed model can generate linguistic hypotheses about Russenorsk, the authors show . their model can be used to generate renderings of contemporary Russian and Norwegian texts . |
PromptCoT: Synthesizing Olympiad-level Problems for Mathematical Reasoning in Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | PromptCoT is a novel approach for synthesizing Olympiad-level math problems . it integrates rationale generation and mathematical concepts to generate complex problems based on concepts and rationale behind problem construction. |
| Approach: | They propose a method for automatically generating high-quality Olympiad-level math problems . they use mathematical concepts and the rationale behind problem construction to synthesize complex problems based on mathematical concepts . |
| Outcome: | The proposed method outperforms existing problem generation methods on benchmarks including GSM8K, MATH-500, and AIME2024. |
Speculative Sampling via Exponential Races (2025.findings-acl)
Copied to clipboard
| Challenge: | Speculative decoding accelerates large language model inference using a smaller draft model. |
| Approach: | They propose a speculative decoding method that generates multiple draft tokens for each model evaluation using a more efficient draft model. |
| Outcome: | The proposed method matches state-of-the-art performance and is based on exponential races. |
Going Beyond Your Expectations in Latency Metrics for Simultaneous Speech Translation (2025.findings-acl)
Copied to clipboard
| Challenge: | Current evaluation practices in Simultaneous Speech Translation systems involve segmenting the input audio and its translations, calculating quality and latency metrics for each segment, and averaging the results. |
| Approach: | They propose to use the mean to estimate latency for Simultaneous Speech Translation systems to provide a better understanding of their results. |
| Outcome: | The proposed methods can provide a better understanding of SimulST systems’ latency. |
Towards a Design Guideline for RPA Evaluation: A Survey of Large Language Model-Based Role-Playing Agents (2025.findings-acl)
Copied to clipboard
| Challenge: | Role-Playing Agents (RPAs) are increasingly popular due to diverse task requirements and agent designs. |
| Approach: | They propose an evidence-based evaluation design guideline for LLM-based RPAs based on agent attributes, task attributes, and evaluation metrics. |
| Outcome: | The proposed evaluation design guideline is based on a systematic review of 1,676 papers published between Jan. 2021 and Dec. 2024. |
Recursive Question Understanding for Complex Question Answering over Heterogeneous Personal Data (2025.findings-acl)
Copied to clipboard
| Challenge: | a novel method for question answering over mixed sources, like text and tables, has been developed for question-answering . personal information is a prominent case of such heterogeneous data, such as calendar entries, workout statistics, shopping records, streaming history, and more. |
| Approach: | They propose a method that creates an executable operator tree for a given question . they use recursive decomposition to decompose a question into an operator tree . |
| Outcome: | The proposed method outperforms methods based on verbalization or translation . it can be executed on user devices and yields a traceable answer . |
PreSumm: Predicting Summarization Performance Without Summarizing (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent advances in summarization models do not produce all documents in the same way, despite their inherent design principles and operational mechanisms. |
| Approach: | They propose a task where a system predicts summarization performance based solely on the source document. |
| Outcome: | The proposed task identifies documents that require manual summarization and improves dataset quality by filtering outliers and noisy documents. |
Mixture of Structural-and-Textual Retrieval over Text-rich Graph Knowledge Bases (2025.findings-acl)
Copied to clipboard
Yongjia Lei, Haoyu Han, Ryan A. Rossi, Franck Dernoncourt, Nedim Lipka, Mahantesh M Halappanavar, Jiliang Tang, Yu Wang
| Challenge: | Existing methods for textual and structural retrieval ignore mutual reinforcement and only use structural retrievals for text-rich Graph Knowledge Bases (TG-KBs). |
| Approach: | They propose a Mixture of Structural-and-Textual Retrieval to retrieve textual and structural knowledge via a Planning-Reasoning-Organizing framework. |
| Outcome: | Experiments show that the proposed framework performs better than existing methods in analyzing TG-KBs and integrating structural trajectories for candidate reranking. |
Fact Recall, Heuristics or Pure Guesswork? Precise Interpretations of Language Models for Fact Completion (2025.findings-acl)
Copied to clipboard
| Challenge: | Language models (LMs) can make a correct prediction based on many possible signals in a prompt, but not all corresponding to recall of factual associations. |
| Approach: | They propose a model-specific recipe for constructing datasets with examples of four different prediction scenarios: generic language modeling, guesswork, heuristics recall and exact fact recall. |
| Outcome: | The proposed model-specific recipe yields distinct results for each scenario. |
FPE2M2: Approaching Lossless and Efficient Quantization with Native Floating Point (2025.findings-acl)
Copied to clipboard
Ke Yi, Jianwei Zhang, Zhiying Xu, Xinlong Yang, Yang Zhou, Minmin Sun, Zengke Liu, Tong Zhang, Junyang Lin, Jingren Zhou
| Challenge: | Auto-regressive decoding is a memory-bound job, meaning decoding performance is limited by the bandwidth rather than the computational capabilities of the GPU. |
| Approach: | They propose a framework that supports lossless weight-only quantization inference and validate it on Qwen and LLaMA Models. |
| Outcome: | The proposed framework achieves the highest efficiency with lossless accuracy on Qwen and LLaMA Models across various modalities. |
Asymmetric Conflict and Synergy in Post-training for LLM-based Multilingual Machine Translation (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing work in LLM-based MMT typically mitigates the Curse of Multilinguality . asymmetric phenomenon in linguistic conflicts and synergy varies in different translation directions . |
| Approach: | They propose a direction-aware training approach to address asymmetry in linguistic conflicts and synergy . they propose X-ALMA-13B-Pretrain with multilingual pre-training to achieve comparable performance . |
| Outcome: | The proposed method achieves comparable performance to X-ALMA-13B-Pretrain (only SFT) with fewer pretraining tokens and 17B parameters. |
VISIAR: Empower MLLM for Visual Story Ideation (2025.findings-acl)
Copied to clipboard
Zhaoyang Xia, Somdeb Sarkhel, Mehrab Tanjim, Stefano Petrangeli, Ishita Dasgupta, Yuxiao Chen, Jinxuan Xu, Di Liu, Saayan Mitra, Dimitris N. Metaxas
| Challenge: | Existing literature on visual storytelling has not explored the ideation process fully. |
| Approach: | They propose a visual story ideation task that automates the selection and arrangement of visual assets into coherent sequences that convey expressive storylines. |
| Outcome: | The proposed framework surpasses baseline by 33.5% and 18.5%, respectively, on three metrics. |
Same Company, Same Signal: The Role of Identity in Earnings Call Transcripts (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing studies rely on earnings call transcripts to predict volatility, but current models focus on capturing ticker identity rather than providing meaningful insights specific to each earnings. |
| Approach: | They propose a dataset that provides 20 earnings records per ticker to help predict volatility . they propose two training-free baselines to capture ticker-specific patterns . |
| Outcome: | The proposed dataset provides 20 earnings records per ticker, with a priorAfterMarket attribute and dense ticker coverage. |
Understanding and Meeting Practitioner Needs When Measuring Representational Harms Caused by LLM-Based Systems (2025.findings-acl)
Copied to clipboard
Emma Harvey, Emily Sheng, Su Lin Blodgett, Alexandra Chouldechova, Jean Garcia-Gathright, Alexandra Olteanu, Hanna Wallach
| Challenge: | Existing tools for measuring representational harms caused by large language model systems are not useful for practitioners. |
| Approach: | They examine the extent to which public instruments are used to measure representational harms caused by large language model-based systems. |
| Outcome: | The proposed instruments do not meet the needs of practitioners evaluating large language model-based systems. |
Mind the (Belief) Gap: Group Identity in the World of LLMs (2025.findings-acl)
Copied to clipboard
| Challenge: | Social biases and belief-driven behaviors can significantly impact Large Language Models’ (LLMs) decisions on several tasks. |
| Approach: | They propose a multi-agent framework that simulates belief congruence, a group psychology theory that plays a crucial role in shaping societal interactions and preferences. |
| Outcome: | The proposed framework reduces misinformation dissemination and improves learning by 11% while reducing misinformation dissemination by up to 37%. |
A General Framework to Enhance Fine-tuning-based LLM Unlearning (2025.findings-acl)
Copied to clipboard
Jie Ren, Zhenwei Dai, Xianfeng Tang, Hui Liu, Jingying Zeng, Zhen Li, Rahul Goutam, Suhang Wang, Yue Xing, Qi He, Hui Liu
| Challenge: | Existing approaches to remove copyrighted and privacy-sensitive data from Large Language Models (LLMs) have been proposed to remove specific data from LLMs without requiring full retraining. |
| Approach: | They propose a general framework that enhances the utility of fine-tuning-based methods by distinguishing target data and suppressing related generations. |
| Outcome: | The proposed framework improves the unlearning and utility of fine-tuning-based methods by distinguishing the target data and suppressing related generations. |
Right Answer, Wrong Score: Uncovering the Inconsistencies of LLM Evaluation in Multiple-Choice Question Answering (2025.findings-acl)
Copied to clipboard
| Challenge: | Multiple-choice question answering tasks are one of the most commonly used tasks for evaluating Large Language Models (LLMs). |
| Approach: | They analyze whether existing answer extraction methods are aligned with human judgment and how they are influenced by answer constraints in the prompt across different domains. |
| Outcome: | The proposed evaluation strategies can be inconsistent with human judgment, and can lead to inaccurate and misleading comparisons. |
Machine Theory of Mind Needs Machine Validation (2025.findings-acl)
Copied to clipboard
| Challenge: | In recent years there has been an explosion of interest in studying the extent to which language models (LMs) display a theory of mind (ToM) despite the growth of evaluation tools, the extent of evidence for ToM remains unclear. |
| Approach: | They conduct a survey of 16 recent studies aimed at measuring ToM in language models and found that only half do so for patterns only a machine might exploit. |
| Outcome: | The results show that the datasets that show high LM performance on ToM tasks are easier than their peers, likely due to the presence of spurious patterns in the data. |
MiniKV: Pushing the Limits of 2-Bit KV Cache via Compression and System Co-Design for Efficient Long Context Inference (2025.findings-acl)
Copied to clipboard
| Challenge: | State-of-the-art 2-bit KV cache quantization methods achieve excellent results in accelerating LLM inference while retaining accuracy on long context tasks. |
| Approach: | They propose a method based on 2-bit KV cache quantization with adaptive KV policies that retain LLM accuracy with only a subset of KV states. |
| Outcome: | The proposed method outperforms state-of-the-art methods on a wide range of long context tasks while retaining accuracy. |
Sci-LoRA: Mixture of Scientific LoRAs for Cross-Domain Lay Paraphrasing (2025.findings-acl)
Copied to clipboard
| Challenge: | Lay paraphrasing aims to make scientific information accessible to non-experts . existing studies focus on a single domain, such as biomedicine . |
| Approach: | a new lay paraphrasing model leverages a mixture of LoRAs fine-tuned on multiple scientific domains. |
| Outcome: | a new model outperforms state-of-the-art large language models in lay paraphrasing . the model can adjust the impact of different domains without explicit labels . |
Trick or Neat: Adversarial Ambiguity and Language Model Evaluation (2025.findings-acl)
Copied to clipboard
| Challenge: | Direct prompting fails to detect ambiguity while linear probes can decode ambiguities with high accuracy, sometimes exceeding 90%. |
| Approach: | They introduce an adversarial ambiguity dataset that includes syntactic, lexical, and phonological ambiguities along with adversarials. |
| Outcome: | The proposed dataset includes syntactic, lexical, and phonological ambiguities along with adversarial variations. |
Biases Propagate in Encoder-based Vision-Language Models: A Systematic Analysis From Intrinsic Measures to Zero-shot Retrieval Outcomes (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing encoder-based vision-language models (VLMs) contain intrinsic biases that manifest in biased outputs. |
| Approach: | They propose a framework to measure intrinsic bias propagation by correlating intrinsic bias with extrinsic bias in zero-shot text-to-image and image-totext retrieval. |
| Outcome: | The proposed framework shows that larger/better-performing models exhibit greater bias propagation, raising concerns given the trend towards increasingly complex AI models. |
Stepwise Perplexity-Guided Refinement for Efficient Chain-of-Thought Reasoning in Large Language Models (2025.findings-acl)
Copied to clipboard
Yingqian Cui, Pengfei He, Jingying Zeng, Hui Liu, Xianfeng Tang, Zhenwei Dai, Yan Han, Chen Luo, Jing Huang, Zhen Li, Suhang Wang, Yue Xing, Jiliang Tang, Qi He
| Challenge: | Chain-of-Thought (CoT) reasoning has improved the performance of large language models (LLMs) however, the detailed reasoning process in CoT often incurs long generation times and high computational costs due to the inclusion of unnecessary steps. |
| Approach: | They propose a method to identify critical reasoning steps using perplexity as a measure of their importance. |
| Outcome: | The proposed method achieves a better balance between reasoning accuracy and efficiency of CoT. |
Can Multimodal Foundation Models Understand Schematic Diagrams? An Empirical Study on Information-Seeking QA over Scientific Papers (2025.findings-acl)
Copied to clipboard
| Challenge: | MISS-QA is the first benchmark specifically designed to evaluate the ability of models to interpret schematic diagrams within scientific literature. |
| Approach: | They propose an automated evaluation protocol powered by open-source LLMs trained on human-scored data to ensure reliable evaluation. |
| Outcome: | The proposed protocol is powered by open-source LLMs trained on human-scored data. |
MultiChallenge: A Realistic Multi-Turn Conversation Evaluation Benchmark Challenging to Frontier LLMs (2025.findings-acl)
Copied to clipboard
Kaustubh Deshpande, Ved Sirdeshmukh, Johannes Baptist Mols, Lifeng Jin, Ed-Yeremai Hernandez-Cardona, Dean Lee, Jeremy Kritz, Willow E. Primack, Summer Yue, Chen Xing
| Challenge: | Existing evaluation frameworks for large language models have limited coverage for multi-turn conversations . multi-turned conversations require accurate instruction following, context allocation, and in-context reasoning at the same time. |
| Approach: | They propose a benchmark to evaluate large language models' ability to conduct multi-turn conversations with humans. |
| Outcome: | The proposed benchmarks achieve near perfect scores on existing benchmarks but only a 41.4% accuracy on the frontier models. |
Privacy Ripple Effects from Adding or Removing Personal Information in Language Model Training (2025.findings-acl)
Copied to clipboard
Jaydeep Borkar, Matthew Jagielski, Katherine Lee, Niloofar Mireshghallah, David A. Smith, Christopher A. Choquette-Choo
| Challenge: | PII is a sensitive information that can be removed from large-language model training due to evolving curation techniques, or because it was recently scraped for retraining. |
| Approach: | They characterize a phenomenon where PII that appeared earlier in training becomes extractable at a later step after fine-tuning on other PI I. |
| Outcome: | The authors show that PII memorization is a dynamic property of a model that evolves throughout training pipelines and depends on commonly altered design choices. |
Safety is Not Only About Refusal: Reasoning-Enhanced Fine-tuning for Interpretable LLM Safety (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) are vulnerable to jailbreak attacks that exploit weaknesses in traditional safety alignment. |
| Approach: | They propose a framework that trains models to engage in explicit safe reasoning before response . they propose RATIONAL, which allows models to reject harmful prompts while providing meaningful and context-aware responses. |
| Outcome: | The proposed framework fine-tunes models to reason about query intent, ethics, and potential harm. |
Is a cute puyfred cute? Context-dependent form-meaning systematicity in LLMs (2025.findings-acl)
Copied to clipboard
| Challenge: | valence is encoded in meaningful ways in large language models and in some LLMs, pseudowords affect the representation of whole sentences similarly to words. |
| Approach: | They investigate how LLMs represent valence, a key semantic attribute, and how they deal with contextualisation of pseudowords in sentences. |
| Outcome: | The results show that the models represent valence, a key semantic attribute, in sentences and in context, and that they handle the contextualisation of pseudowords differently. |
MetaSynth: Meta-Prompting-Driven Agentic Scaffolds for Diverse Synthetic Data Generation (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent smaller language models rely on synthetic data generated using larger Language models. |
| Approach: | They propose a method for generating synthetic data that enhances diversity through meta-prompting . they use 25 million tokens of synthetic data generated by a language model orchestrated by multiple “expert” LLM agents to collaboratively generate data. |
| Outcome: | The proposed method outperforms the base LLM in Finance and Biomedicine with 25 million tokens of synthetic data. |
MVTamperBench: Evaluating Robustness of Vision-Language Models (2025.findings-acl)
Copied to clipboard
Amit Agarwal, Srikant Panda, Angeline Charles, Hitesh Laxmichand Patel, Bhargava Kumar, Priyaranjan Pattnayak, Taki Hasan Rafi, Tejaswini Kumar, Hansa Meghwani, Karan Gupta, Dong-Kyu Chae
| Challenge: | Multimodal Large Language Models (MLLMs) have been a key advance in video understanding but their vulnerability to adversarial tampering remains underexplored. |
| Approach: | They evaluate MLLMs against five prevalent tampering techniques to assess their robustness . they use a tampered video format to examine the vulnerability of ML models . |
| Outcome: | The benchmark evaluates MLLMs against five prevalent tampering techniques based on 19 video manipulation tasks. |
Multimodal Inconsistency Reasoning (MMIR): A New Benchmark for Multimodal Reasoning Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing Multimodal Large Language Models (MLLMs) are predominantly trained on consistent visual-textual inputs, leaving open the question of whether they can handle semantic mismatches in layout-rich content. |
| Approach: | They propose to use multimodal inconsistency reasoning to assess MLLMs' ability to reason about semantic mismatches in webpages, presentation slides, and posters. |
| Outcome: | The proposed model outperforms open-source models in detecting inconsistencies in webpages, presentation slides, and posters while remaining vulnerable to inconsistent errors. |
Vision-Language Models Struggle to Align Entities across Modalities (2025.findings-acl)
Copied to clipboard
| Challenge: | Several real-world applications require the ability to perform cross-modal entity linking . cross-functional entity linking is a skill needed for multimodal code generation and scene understanding . |
| Approach: | They propose a task and benchmark to evaluate cross-modal entity linking performance . they use visual scenes aligned with their textual representations to evaluate performance a question-answering task . |
| Outcome: | The proposed task and benchmark aims to improve cross-modal entity linking performance . it evaluates state-of-the-art vision-language models and humans on the task . |
A Multi-Labeled Dataset for Indonesian Discourse: Examining Toxicity, Polarization, and Demographics Information (2025.findings-acl)
Copied to clipboard
Lucky Susanto, Musa Izzanardi Wijanarko, Prasetia Anugrah Pratama, Zilu Tang, Fariz Akyas, Traci Hong, Ika Karlina Idris, Alham Fikri Aji, Derry Tanti Wijaya
| Challenge: | Prior research has focused on toxicity and polarization as separate problems . extreme polarizing deepens divisions, often leading to hostility and fragmentation . |
| Approach: | They propose to use a multi-label Indonesian dataset annotated for toxicity, polarization, and annotator demographic information to study polarizing language and toxicity. |
| Outcome: | The proposed dataset shows that polarization cues improve toxicity classification and vice versa. |
MedCite: Can Language Models Generate Verifiable Text for Medicine? (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing LLM-based medical question answering systems lack citation generation and evaluation capabilities, raising concerns about their adoption in practice. |
| Approach: | They propose a framework that facilitates the design and evaluation of LLM citations for medical tasks and a retrieval-citation method that generates high-quality citation. |
| Outcome: | The proposed method achieves superior citation precision and recall improvements compared to strong baseline methods and correlates well with annotation results from professional experts. |
Let The Jury Decide: Fair Demonstration Selection for In-Context Learning through Incremental Greedy Evaluation (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing demonstration selection strategies focus on optimizing performance metrics such as accuracy. |
| Approach: | They propose a framework for selecting fair and representative demonstrations that improve group fairness in In-Context Learning. |
| Outcome: | The proposed framework improves fairness metrics without compromising accuracy. |
The Lies Characters Tell: Utilizing Large Language Models to Normalize Adversarial Unicode Perturbations (2025.findings-acl)
Copied to clipboard
| Challenge: | Homoglyphs are visually homogeneous to Latin letters and are used to mask offensive content. |
| Approach: | They propose two methods to normalize homoglyphs by replacing non-Latin characters with a delimiter and using large language models to determine which characters should be replaced with Latin letters. |
| Outcome: | The proposed methods normalize homoglyphs by replacing non-Latin characters with a delimiter and prompting large language models to "fill in the blanks" the authors found that the proposed methods produced normalized text with an average cosine similarity score of 0.91 to the original tweets and 0.96 to the tweets using the direct method. |
Speech Act Patterns for Improving Generalizability of Explainable Politeness Detection Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Politeness detection is a task that requires explainability but lacks generalizability . recent approaches for improving explainable models rely on discovering domain-specific word-level features. |
| Approach: | They propose a method for improving the generalizability of explainable politeness models by relying on speech act patterns instead of words. |
| Outcome: | The proposed method improves generalizability of explainable politeness models by relying on speech act patterns instead of words. |
Systematic Evaluation of Auto-Encoding and Large Language Model Representations for Capturing Author States and Traits (2025.findings-acl)
Copied to clipboard
Khushboo Singh, Vasudha Varadarajan, Adithya V Ganesan, August Håkan Nilsson, Nikita Soni, Syeda Mahwish, Pranav Chitale, Ryan L. Boyd, Lyle Ungar, Richard N Rosenthal, H. Schwartz
| Challenge: | Large Language Models (LLMs) are increasingly used in human-centered applications, yet their ability to model diverse psychological constructs is not well understood. |
| Approach: | They evaluated a range of Transformer-LMs to predict psychological variables across five major dimensions: affect, substance use, mental health, sociodemographics, and personality. |
| Outcome: | The models predict affect, substance use, mental health, sociodemographics, and personality across five major dimensions. |
TReMu: Towards Neuro-Symbolic Temporal Reasoning for LLM-Agents with Memory in Multi-Session Dialogues (2025.findings-acl)
Copied to clipboard
| Challenge: | Temporal reasoning in multi-session dialogues presents a significant challenge which has been under-studied in previous temporal reasoning benchmarks. |
| Approach: | They propose to augment LoCoMo dialogues and create multi-choice QAs to construct a temporal reasoning evaluation task and a framework to enhance temporal thinking capabilities of LLM-agents. |
| Outcome: | The proposed framework significantly improves temporal reasoning performance compared to baseline methods, raising from 29.83 on GPT-4o via standard prompting to 77.67 via the proposed framework. |
Conservative Bias in Large Language Models: Measuring Relation Predictions (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) exhibit pronounced conservative bias in relation extraction tasks, often defaulting to no_relation label when an appropriate option is unavailable. |
| Approach: | They systematically evaluate the trade-off between conservative bias and hallucination in relation extraction tasks by using SBERT and LLM prompts to quantify this effect. |
| Outcome: | The proposed model defaults to no_relation label twice as often as hallucination, resulting in significant information loss when reasoning is not explicitly included in the output. |
Mitigating Bias in RAG: Controlling the Embedder (2025.findings-acl)
Copied to clipboard
| Challenge: | a promising modular AI system enhances factuality and privacy in large language models . however, each component introduces its own biases into the RAG system, which could cause representational harm and unsafe user interactions. |
| Approach: | They study the conflict between biases of each component and their relationship to the overall bias of the retrieval augmented generation system. |
| Outcome: | The proposed model can be controlled by the embedder while maintaining utility . the proposed model is more fair than existing models, the authors show . |
V-ALPHASOCIAL: Benchmark and Self-Reflective Chain-of-Thought Generation for Visual Social Commonsense Reasoning (2025.findings-acl)
Copied to clipboard
Zongyu Lin, Zhikun Xu, Xiaohan Song, Yixin Wan, Xingcheng Yao, Tsung-Han Lin, Selina Song, Pranav Subbaraman, Ben Zhou, Kai-Wei Chang, Yizhou Sun
| Challenge: | Social commonsense reasoning is a multimodal task that requires both textual and visual cues. |
| Approach: | They propose a method that integrates visual cues into social commonsense reasoning tasks. |
| Outcome: | The proposed method improves social commonsense reasoning on a multimodal foundation model. |
AfroBench: How Good are Large Language Models on African Languages? (2025.findings-acl)
Copied to clipboard
Jessica Ojo, Odunayo Ogundepo, Akintunde Oladipo, Kelechi Ogueji, Jimmy Lin, Pontus Stenetorp, David Ifeoluwa Adelani
| Challenge: | Large-scale multilingual evaluations often include only a handful of African languages due to the scarcity of high-quality data and the limited discoverability of existing datasets. |
| Approach: | They propose a multi-task benchmark to evaluate the performance of LLMs across 64 African languages, 15 tasks and 22 datasets. |
| Outcome: | The proposed benchmark compares LLMs across 64 African languages, 15 tasks and 22 datasets. |
Training Bilingual LMs with Data Constraints in the Targeted Language (2025.findings-acl)
Copied to clipboard
| Challenge: | a large number of languages have insufficient data for pretraining, but most non-English models are trained on scrapes of the web. |
| Approach: | They propose to use data from an auxiliary language to boost model performance . they quantify the performance gap between training with data in a data-rich auxiliary and training in the target language . |
| Outcome: | The proposed method boosts model performance in a target language with insufficient data . it also explores the benefits of translation systems and the limitations of model scaling when data is limited. |
ChartQAPro: A More Diverse and Challenging Benchmark for Chart Question Answering (2025.findings-acl)
Copied to clipboard
Ahmed Masry, Mohammed Saidul Islam, Mahir Ahmed, Aayush Bajaj, Firoz Kabir, Aaryaman Kartha, Md Tahmid Rahman Laskar, Mizanur Rahman, Shadikur Rahman, Mehrad Shahmohammadi, Megh Thakkar, Md Rizwan Parvez, Enamul Hoque, Shafiq Joty
| Challenge: | Chart Question Answering systems are limited in their ability to interpret data visually and reason with visual representations. |
| Approach: | They propose a chart-based chart question-answering system that includes 1,341 charts from 99 diverse sources and 1,948 questions in various types. |
| Outcome: | The new benchmark includes 1,341 charts from 99 diverse sources and 1,948 questions in various types. |
From Observation to Understanding: Front-Door Adjustments with Uncertainty Calibration for Enhancing Egocentric Reasoning in LVLMs (2025.findings-acl)
Copied to clipboard
Shenshen Li, Wenxin Meng, Lei Wang, Hao Yang, Chong Peng, Peng Yan, Fumin Shen, Jingkuan Song, Heng Tao Shen, Xing Xu
| Challenge: | Existing methods that adapt LVLMs to egocentric tasks overlook critical agent-environment interactions, limiting their ability to perform egoic reasoning. |
| Approach: | They propose a zero-shot paradigm to enhance egocentric reasoning by simulating human causal reasoning by formalizing ego-centric reasoning using a structural causal model. |
| Outcome: | The proposed method improves egocentric reasoning abilities on six tasks. |
Hypothetical Documents or Knowledge Leakage? Rethinking LLM-based Query Expansion (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent studies have demonstrated effectiveness in zero-shot retrieval tasks using large language models. |
| Approach: | They challenge this assumption by analyzing whether knowledge leakage in benchmarks contributes to performance gains. |
| Outcome: | The proposed methods have demonstrated significant performance gains across multiple benchmarks. |
Worse than Random? An Embarrassingly Simple Probing Evaluation of Large Multimodal Models in Medical VQA (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Multimodal Models (LMMs) have demonstrated impressive performance on existing medical visual question answering benchmarks. |
| Approach: | They evaluate large multimodal models that perform worse than random guessing on medical questions . authors suggest more robust evaluation methods to ensure reliability of LMMs . |
| Outcome: | a new study shows that large multimodal models perform worse than random guessing on medical visual question answering benchmarks. |
Optimizing Reasoning for Text-to-SQL with Execution Feedback (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models excel in many reasoning tasks, but their ability to leverage Chain-of-Thought (CoT) reasoning remains underexplored. |
| Approach: | They propose a framework that iteratively optimizes open-source LLMs by combining CoT reasoning with off-policy and on-poly DPO, relying solely on execution accuracy as feedback. |
| Outcome: | The proposed framework improves execution accuracy on BIRD and Spider datasets. |
Disentangling Logic: The Role of Context in Large Language Model Reasoning Capabilities (2025.findings-acl)
Copied to clipboard
Wenyue Hua, Kaijie Zhu, Lingyao Li, Lizhou Fan, Mingyu Jin, Shuhang Lin, Haochen Xue, Zelong Li, Jindong Wang, Yongfeng Zhang
| Challenge: | Using large language models, large language model models can be used to evaluate reasoning abilities in context-rich scenarios. |
| Approach: | They construct datasets for both propositional logic and abductive logic reasoning with four difficulty levels across 12 distinct domains based on Wikipedia categorization and those with purely abstract variables. |
| Outcome: | The proposed model can be used to benchmark LLMs in real-world scenarios, but not in context-rich scenarios. |
Sens-Merging: Sensitivity-Guided Parameter Balancing for Merging Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing task vector-based model merging methods apply uniform coefficients across all parameters, overlooking varying parameter importance both within and across tasks. |
| Approach: | They propose a sensitivity-guided coefficient adjustment method that optimizes existing model merging techniques by operating at both task-specific and cross-task levels. |
| Outcome: | The proposed method outperforms existing model merging techniques on mistral 7B and LLaMA2 7B/13B models and enables them to outperformed specialized models. |
EgoNormia: Benchmarking Physical-Social Norm Understanding (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing VLMs lack robust grounded norm understanding, a new study finds . current VLM models lack robust grounding, despite a high score for safety and privacy . |
| Approach: | They propose a pipeline to generate grounded MCQs from ego-centric videos of human interactions. |
| Outcome: | The proposed pipeline can generate grounded MCQs from egocentric video . it shows that current VLMs lack robust grounded norm understanding . |
Large Language Models as Neurolinguistic Subjects: Discrepancy between Performance and Competence (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing evaluations of Large Language Models (LLMs) reflect statistical rules that may not accurately represent LLMs’ true linguistic competence. |
| Approach: | They propose a method that combines minimal pair and diagnostic probing to analyze activation patterns across model layers. |
| Outcome: | The proposed method combines minimal pair and diagnostic probing to analyze activation patterns across model layers. |
The Impact of Large Language Models in Academia: from Writing to Speaking (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) are impacting human society, especially in textual information. |
| Approach: | They propose to build an automated monitoring platform to track the impact of large language models on human expression. |
| Outcome: | The results show that LLM-style words such as significant are used more frequently in abstracts and oral presentations. |
X-WebAgentBench: A Multilingual Interactive Web Benchmark for Evaluating Global Agentic System (2025.findings-acl)
Copied to clipboard
| Challenge: | X-WebAgentBench evaluates the planning and interaction performance of language agents across multiple languages. |
| Approach: | They propose a multilingual agent benchmark that evaluates the interaction performance of language agents across multiple languages. |
| Outcome: | The proposed benchmark evaluates the planning and interaction performance of language agents across multiple languages. |
MemBench: Towards More Comprehensive Evaluation on the Memory of LLM-based Agents (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent studies have highlighted the significance of memory mechanisms in LLM-based agents, which enable them to store observed information and adapt to dynamic environments. |
| Approach: | They propose a dataset and benchmark to evaluate the memory capability of LLM-based agents from multiple aspects including their effectiveness, efficiency, and capacity. |
| Outcome: | The proposed benchmark incorporates factual memory and reflective memory as different levels, and proposes participation and observation as various interactive scenarios. |
Adaptive LoRA Merge with Parameter Pruning for Low-Resource Generation (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for adapting LLMs to low-resource tasks keep LoRA parameters frozen and the low-level problem out of their scope. |
| Approach: | They propose a LoRA merge method that updates and prunes LoRA parameters through fine-tuning with minimal target task data. |
| Outcome: | The proposed method improves performance on a low-resource language generation task and improves on previous methods. |
LongAttn: Selecting Long-context Training Data via Token-level Attention (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods to select long-context data often rely on sentence-level analysis, which can be greatly optimized in both performance and efficiency. |
| Approach: | They propose a token-level framework which quantifies long-range dependencies for LLMs by calculating token-based dependency strength and distribution uniformity of token scores. |
| Outcome: | The proposed framework quantifies long-range dependencies, enabling more accurate and efficient data selection. |
CoRE: Condition-based Reasoning for Identifying Outcome Variance in Complex Events (2025.findings-acl)
Copied to clipboard
| Challenge: | Identifying implied conditions and examining their influence on an outcome is challenging. |
| Approach: | They combine annotations from goals and states to examine the influence of conditions . they examine open and closed LLMs of varying sizes and intent-alignment on reasoning tasks . |
| Outcome: | The proposed models are more cautious in less constrained situations when conditions are used to replace missing context. |
FaVe: Factored and Verified Search Rationale for Long-form Answer (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing solutions for long-form question-answering (LFQA) use chain-of-thought (CoT) with retrieval-augmented generation (RAG). |
| Approach: | They propose to integrate chain-of-thought (CoQ) with retrieval-augmented generation to improve answer comprehensiveness and verifiability. |
| Outcome: | The proposed approach outperforms ChatGPT baselines while maintaining efficiency. |
UnrealLLM: Towards Highly Controllable and Interactable 3D Scene Generation by LLM-powered Procedural Content Generation (2025.findings-acl)
Copied to clipboard
SongTang SongTang, Kaiyong Zhao, Lei Wang, Yuliang Li, Xuebo Liu, Junyi Zou, Qiang Wang, Xiaowen Chu
| Challenge: | UnrealLLM is a novel framework that connects natural language descriptions with the professional PCG system (Unreal Engine 5) to automate scene generation. |
| Approach: | They propose a novel multi-agent framework that connects natural language descriptions with the professional PCG system (Unreal Engine 5) to automate scene generation. |
| Outcome: | The proposed framework achieves competitive performance in technical metrics and aesthetic quality, offering unique advantages in generation scale and interactivity. |
Tree-of-Prompts: Abstracting Control-Flow for Prompt Optimization (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing prompt optimization methods struggle with disjoint cases in complex tasks. |
| Approach: | They propose a tree-of-prompts structure which expands child prompts from parent prompts . they propose to use a nested if-else structure to address varying similarities and complexities . |
| Outcome: | The proposed tree-of-prompts outperforms PromptAgent and MoP on Gorilla, MATH and subset of BBH benchmarks. |
Outlier-weighed Layerwise Sampling for LLM Fine-tuning (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) are a powerful tool for processing complex natural language processing tasks. |
| Approach: | They propose an approach to fine-tune LLMs with outliers and a gradient low-rank projection to increase the number of fine-sampled layers without a proportional increase in memory costs. |
| Outcome: | The proposed approach outperforms baseline approaches while being more memory efficient. |
KVPR: Efficient LLM Inference with I/O-Aware KV Cache Partial Recomputation (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods to inference large language models are limited by CPU capabilities and memory constraints. |
| Approach: | They propose an efficient I/O-aware LLM inference method that overlaps GPU computation with KV cache transfer to minimize idle GPU time. |
| Outcome: | The proposed method reduces the cost of auto-regressive decoding by 35.8% . it also achieves 46.2% higher throughput during decoding compared to state-of-the-art methods. |
Direct Behavior Optimization: Unlocking the Potential of Lightweight LLMs (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing prompt optimization methods rely on extensive manual effort or meta-cognitive abilities, making them less effective for LwLLMs. |
| Approach: | They propose a direct behavior optimization parameter that transforms the optimization of complex prompts into discrete, quantifiable execution sequences using a gradient-free Monte Carlo Tree Search. |
| Outcome: | The proposed method outperforms current prompt optimization methods on seven challenging tasks where state-of-the-art LLMs excel but LwLLMs generally underperform. |
Whether LLMs Know If They Know: Identifying Knowledge Boundaries via Debiased Historical In-Context Learning (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for active retrieval (AR) rely on training classification models or using the confidence of the model’s answer to determine knowledge boundaries. |
| Approach: | They propose a method to identify knowledge boundaries in active retrieval by retrieving historical queries as high-confidence in-context examples. |
| Outcome: | Experiments on four QA benchmarks show that DH-ICL achieves performance comparable to full retrieval on LLaMA with only half the number of retrievals, without any additional training. |
How do LLMs’ Preferences Affect Event Argument Extraction? CAT: Addressing Preference Traps in Unsupervised EAE (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing approaches to supervised EAE suffer from preference traps due to misalignments between prior knowledge, instructions, or output constraints and LLMs’ preferences. |
| Approach: | They propose an unsupervised EAE framework that handles LLMs' preference traps by targeting their prior knowledge and instructions. |
| Outcome: | The proposed framework matches the best DeepSeek-R1 API model with a significantly lower time cost. |
Out-of-Distribution Detection via LLM-Guided Outlier Generation for Text-attributed Graph (2025.findings-acl)
Copied to clipboard
| Challenge: | Text-Attributed Graphs (TAGs) are widely used in the real world. |
| Approach: | They propose to use Large Language Models to generate OOD-nodes with high quality . they also use LLMs to integrate existing nodes with LLM-generated edges . |
| Outcome: | The proposed method performs well on samples outside the In-Distribution (ID) data, but it is difficult to obtain high-quality OOD samples in the real world. |
Document-Level Relation Extraction with Global Relations and Entity Pair Reasoning (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing document-level relation extraction models focus on individual entity pairs, limiting their ability to handle complex reasoning tasks. |
| Approach: | They propose a document-level relation extraction framework based on global relations and entity pair reasoning that captures fine-grained interactions between entity pairs. |
| Outcome: | The proposed framework outperforms existing models on widely-used datasets. |
Towards Storage-Efficient Visual Document Retrieval: An Empirical Study on Reducing Patch-Level Embeddings (2025.findings-acl)
Copied to clipboard
Yubo Ma, Jinsong Li, Yuhang Zang, Xiaobao Wu, Xiaoyi Dong, Pan Zhang, Yuhang Cao, Haodong Duan, Jiaqi Wang, Yixin Cao, Aixin Sun
| Challenge: | Visualized Document Retrieval (VDR) uses large vision-language models to encode document pages into embeddings. |
| Approach: | They evaluate methods to reduce patch embeddings per page while minimizing performance degradation. |
| Outcome: | The proposed method maintains 98.2% of retrieval performance with only 11.8% of original memory usage and preserves 94.6% effectiveness at 2% memory footprint. |
Step-by-Step Mastery: Enhancing Soft Constraint Following Ability of Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | In real-world scenarios, user instructions often contain soft constraints, which are semantically related and cannot be rule-based verified, posing challenges for large language models. |
| Approach: | They propose a pipeline to construct datasets with high-quality outputs for instructions containing soft constraints automatically and use Direct Preference Optimization (DPO) as the training method. |
| Outcome: | The proposed model improves the LLMs' soft constraint following ability by using direct preference optimization (DPO) and constraint quantity. |
ZeroDL: Zero-shot Distribution Learning for Text Clustering via Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) have shown impressive performance on downstream tasks, but if they cannot be fully described in prompts, they could fail to perform the task. |
| Approach: | They propose a method to contextualize a task toward a large language model (LLM) they use open-ended zero-shot inference from the entire dataset to aggregate the inference results and incorporate the aggregated meta-information for the actual task. |
| Outcome: | The proposed method improves text clustering tasks and improves on several datasets. |
Patterns Over Principles: The Fragility of Inductive Reasoning in LLMs under Noisy Observations (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent advances in large language models (LLMs) have demonstrated remarkable performance across various reasoning tasks. |
| Approach: | They propose a task that evaluates LLMs’ capability in inferring rules from data fused with noisy examples. |
| Outcome: | The proposed method outperforms other methods with minimal performance degradation under noise and counterfactual task gaps highlight LLMs’ reliance on memorized patterns over genuine abstraction. |
LLMTaxo: Leveraging Large Language Models for Constructing Taxonomy of Factual Claims from Social Media (2025.findings-acl)
Copied to clipboard
| Challenge: | Social media's global reach and ease of use have transformed how millions of users exchange opinions, news, and factual claims in real-time, making it fertile ground for misinformation. |
| Approach: | They propose a framework that leverages large language models to construct taxonomies of factual claims from social media by generating topics at multiple levels of granularity. |
| Outcome: | The proposed framework produces clear, coherent, and comprehensive taxonomies on three diverse datasets and outperforms other frameworks in most metrics. |
AnCast++: Document-Level Evaluation of Graph-based Meaning Representations (2025.findings-acl)
Copied to clipboard
| Challenge: | Abstract Meaning Representation (UMR) is a cross-lingual document-level graph-based representation that extends it to document- level semantic annotations. |
| Approach: | They propose an evaluation metric that unifies evaluation of four distinct sub-structures of UMR. |
| Outcome: | The proposed metric is made available on Github. |
MMEvol: Empowering Multimodal Large Language Models with Evol-Instruct (2025.findings-acl)
Copied to clipboard
Run Luo, Haonan Zhang, Longze Chen, Ting-En Lin, Xiong Liu, Yuchuan Wu, Min Yang, Yongbin Li, Minzheng Wang, Pengpeng Zeng, Lianli Gao, Heng Tao Shen, Yunshui Li, Hamid Alinejad-Rokny, Xiaobo Xia, Jingkuan Song, Fei Huang
| Challenge: | a new framework for image-text instruction data evolution improves MLLM performance . lack of high-quality instruction data remains a major bottleneck in ML modeling . |
| Approach: | They propose a multimodal instruction data evolution framework that iteratively enhances data quality through fine-grained perception, cognitive reasoning, and interaction evolution. |
| Outcome: | The proposed approach improves MLLM performance in nine vision-language tasks while using significantly less data. |
SciVerse: Unveiling the Knowledge Comprehension and Visual Reasoning of LMMs on Multi-modal Scientific Problems (2025.findings-acl)
Copied to clipboard
| Challenge: | SciVerse is a multi-modal scientific evaluation benchmark to assess large multi-models . it examines the scientific knowledge comprehension, multi-mod content interpretation and Chain-of-Thought reasoning . authors examine the scientific proficiency of LMMs in scientific domains based on their work . |
| Approach: | They propose a multi-modal scientific evaluation benchmark to thoroughly assess Large Multi-modal Models across 5,735 test instances in five different versions. |
| Outcome: | The proposed evaluation reveals critical limitations in LMMs' scientific proficiency and provides new insights into future developments. |
Exploring Layer-wise Representations of English and Chinese Homonymy in Pre-trained Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | lexical ambiguity can arise due to the misunderstanding of its multiple senses. |
| Approach: | They propose to use part of speech to examine homonyms in Chinese and English . they find no universal layer depth excels in differentiating homnomial representations . |
| Outcome: | The proposed model improves contextualization of homonym representations in Chinese . the results challenge the simplistic understanding of their inner workings, the authors say . |
DocMEdit: Towards Document-Level Model Editing (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing models only output short phrases or sentences, raising doubts about their practical usability. |
| Approach: | They propose a dataset focused on document-level model editing that aims to correct errors and outdated knowledge in Large language models (LLMs) they propose to use document-based model editing to improve model capabilities in real-world scenarios. |
| Outcome: | The proposed model editing task improves model capabilities in real-world scenarios and reduces the cost of retraining. |
Adaptive Detoxification: Safeguarding General Capabilities of LLMs through Toxicity-Aware Knowledge Editing (2025.findings-acl)
Copied to clipboard
Yifan Lu, Jing Li, Yigeng Zhou, Yihui Zhang, Wenya Wang, Xiucheng Li, Meishan Zhang, Fangming Liu, Jun Yu, Min Zhang
| Challenge: | Existing knowledge editing methods for large language models (LLMs) suffer from over-editing, where detoxified models reject legitimate queries, compromising overall performance. |
| Approach: | They propose a toxicity-aware knowledge editing approach that dynamically detects toxic activation patterns during forward propagation and then routes computations through adaptive inter-layer pathways to mitigate toxicity effectively. |
| Outcome: | The proposed method outperforms existing methods on large language models and enhances the SafeEdit benchmark. |
Evaluating the Long-Term Memory of Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent studies have not thoroughly investigated the memory performance of large language models in long-term tasks. |
| Approach: | They propose a dataset to evaluate the long-term memory capabilities of large language models. |
| Outcome: | The proposed model exhibits memory preferences across different categories of information. |
Explain-then-Process: Using Grammar Prompting to Enhance Grammatical Acceptability Judgments (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) can explain grammatical rules, but fail to apply those rules when judging sentence acceptability. |
| Approach: | They propose a grammar prompting paradigm that feeds an LLM's metalinguistic explanation back to the target model before deciding which sentence of a minimal pair is grammatical. |
| Outcome: | The proposed model improves on the English BLiMP, Chinese SLING, and Russian RuBLimp benchmarks. |
Data Interpreter: An LLM Agent for Data Science (2025.findings-acl)
Copied to clipboard
Sirui Hong, Yizhang Lin, Bang Liu, Bangbang Liu, Binhao Wu, Ceyao Zhang, Danyang Li, Jiaqi Chen, Jiayi Zhang, Jinlin Wang, Li Zhang, Lingyao Zhang, Min Yang, Mingchen Zhuge, Taicheng Guo, Tuo Zhou, Wei Tao, Robert Tang, Xiangtao Lu, Xiawu Zheng, Xinbing Liang, Yaying Fei, Yuheng Cheng, Yongxin Ni, Zhibin Gou, Zongze Xu, Yuyu Luo, Chenglin Wu
| Challenge: | Large Language Models (LLMs) excel in various domains but face challenges when applied to data science workflows due to their complex, multi-stage nature. |
| Approach: | They propose a hierarchical graph-based agent that represents complexity and a progressive strategy for step-by-step verification, refinement, and consistent context management. |
| Outcome: | The proposed agent surpasses state-of-the-art baselines on the MATH dataset and performs better on InfiAgent-DABench. |
DReSD: Dense Retrieval for Speculative Decoding (2025.findings-acl)
Copied to clipboard
| Challenge: | Speculative decoding (SD) uses an efficient draft model to propose the next few tokens, which are verified by the LLM in a single forward call, reducing latency while preserving its outputs. |
| Approach: | They propose a draft model that proposes the next few tokens from a non-parametric datastore and uses a framework that uses approximate nearest neighbour search with contextualised token embeddings to retrieve the most semantically relevant sequences for SD. |
| Outcome: | The proposed framework achieves (on average) 87% higher acceptance rates, 65% longer accepted tokens and 19% faster generation speeds compared to sparse retrieval (REST). |
Core: Robust Factual Precision with Informative Sub-Claim Identification (2025.findings-acl)
Copied to clipboard
Zhengping Jiang, Jingyu Zhang, Nathaniel Weir, Seth Ebner, Miriam Wanner, Kate Sanders, Daniel Khashabi, Anqi Liu, Benjamin Van Durme
| Challenge: | Using the Decompose-Then-Verify framework, such as FActScore, can be manipulated by adding obvious or repetitive subclaims to artificially inflate scores. |
| Approach: | They propose a decomposition-based tool called Core to filter subclaims based on their uniqueness and informativeness. |
| Outcome: | The proposed evaluation framework supports easy and modular use of Core and various decomposition strategies. |
Rethinking Diverse Human Preference Learning through Principal Component Analysis (2025.findings-acl)
Copied to clipboard
| Challenge: | Decomposed Reward Models extract diverse human preferences from binary comparisons without fine-grained annotations. |
| Approach: | They propose a decomposed reward model that extracts diverse human preferences from binary comparisons without fine-grained annotations. |
| Outcome: | The proposed approach extracts diverse human preferences from binary comparisons without fine-grained annotations. |
Improving Word Alignment Using Semi-Supervised Learning (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing word alignment methods rely on labeled data, but augmenting training with pseudo-labeled data improves performance. |
| Approach: | They propose a semi-supervised framework to improve word alignment methods . they use pseudo-labeled data from multilingual encoder models as word aligners . |
| Outcome: | The proposed framework outperforms the current state-of-the-art binary alignment method on word alignment datasets. |
How Do LLMs Acquire New Knowledge? A Knowledge Circuits Perspective on Continual Pre-Training (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have exceptional capabilities in knowledge-intensive tasks . however, they struggle with knowledge updates due to dynamic nature of world knowledge . |
| Approach: | They propose to identify computational subgraphs that facilitate knowledge storage and processing . they also identify a phase shift from formation to optimization in LLMs . |
| Outcome: | The proposed model can capture factual knowledge from pre-training corpus and encapsulate it as extensive parametric knowledge. |
LLM-Symbolic Integration for Robust Temporal Tabular Reasoning (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for temporal tabular question answering are inconsistent and fail to provide the variability needed to thoroughly evaluate models. |
| Approach: | TEMPTABQA-C uses a synthetic dataset and symbolic representation to generate and execute SQL queries. |
| Outcome: | TEMPTABQA-C improves on previous methods for temporal tabular question answering . incorporating adaptive fewshot prompting with tailored examples improves performance . lack of robustness, scalability, and interpretable solutions is key obstacle . |
Multimodal Large Language Models for Text-rich Image Understanding: A Comprehensive Review (2025.findings-acl)
Copied to clipboard
Pei Fu, Tongkun Guan, Zining Wang, Zhentao Guo, Chen Duan, Hao Sun, Boming Chen, Qianyi Jiang, Jiayao Ma, Kai Zhou, Junfeng Luo
| Challenge: | Recent advances in vision-language models have unified perception and understanding tasks within Visual Question Answering paradigms. |
| Approach: | They propose to outline timeline, architecture, and pipeline of nearly all TIU MLLMs and review their performance on mainstream benchmarks. |
| Outcome: | The proposed models perform well on mainstream benchmarks and are compared with other models. |
PruneVid: Visual Token Pruning for Efficient Video Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing approaches to video token pruning face significant computational challenges due to the redundancy inherent in video data. |
| Approach: | They propose a training-free visual token pruning method that reduces the redundancy inherent in video data and leverages LLMs’ inherent ability to selectively prune visual tokens irrelevant to specific queries. |
| Outcome: | The proposed method can prune over 80% of tokens while maintaining competitive performance when combined with different video LLMs. |
PromptWizard: Optimizing Prompts via Task-Aware, Feedback-Driven Self-Evolution (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) have transformed AI across diverse domains, with prompting being central to their success in guiding model outputs. |
| Approach: | They propose a framework for discrete prompt optimization that generates human-readable prompts using feedback-driven critique and synthesis process. |
| Outcome: | The proposed framework improves prompt quality across 45 tasks and reduces API calls, token usage and overall cost. |
Exposing Numeracy Gaps: A Benchmark to Evaluate Fundamental Numerical Abilities in Large Language Models (2025.findings-acl)
Copied to clipboard
Haoyang Li, Xuejia Chen, Zhanchao Xu, Darian Li, Nicole Hu, Fei Teng, Yiming Li, Luyu Qiu, Chen Jason Zhang, Li Qing, Lei Chen
| Challenge: | Existing benchmarks focus on linguistic competence or structured mathematical problem-solving, neglecting fundamental numerical reasoning required in real-world scenarios. |
| Approach: | They propose a benchmark to evaluate numerical capabilities for large language models . they use a dataset to assess number recognition, arithmetic operations, contextual retrieval, comparison, summary, and multi-step reasoning. |
| Outcome: | The proposed benchmark evaluates six fundamental numerical capabilities: number recognition, arithmetic operations, contextual retrieval, comparison, summary, and multi-step reasoning. |
TABGEN-ICL: Residual-Aware In-Context Example Selection for Tabular Data Generation (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing approaches to tabular data generation require fine-tuning, which is computationally expensive. |
| Approach: | They propose a new in-context learning framework to prompt a fixed LLM with in-constitut examples to enhance the in-text learning ability of LLMs for tabular data generation. |
| Outcome: | The proposed framework outperforms random selection strategies on five real-world tabular datasets and reduces error rate by 42.2% on fidelity metric. |
Benchmarking Multi-National Value Alignment for Large Language Models (2025.findings-acl)
Copied to clipboard
Chengyi Ju, Weijie Shi, Chengzhong Liu, Jiaming Ji, Jipeng Zhang, Ruiyuan Zhang, Jiajie Xu, Yaodong Yang, Sirui Han, Yike Guo
| Challenge: | Existing studies on large language models focus on ethical reviews, failing to capture the diversity of national values. |
| Approach: | They propose a national value extraction pipeline to efficiently construct value assessment datasets and a model-based model with instruction tagging to process raw data sources. |
| Outcome: | The proposed benchmark evaluates the alignment of LLMs with the values of five major nations: China, the United States, the UK, France, and Germany. |
MotiveBench: How Far Are We From Human-Like Motivational Reasoning in Large Language Models? (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing benchmarks for large language models lack information asymmetry with real-world situations. |
| Approach: | They propose a benchmark to evaluate the human-like motivational and behavioral reasoning ability of LLMs with detailed, realistic situations. |
| Outcome: | The proposed benchmark compared LLMs with real-world scenarios on seven model families and found that the most advanced models struggle with understanding "love & belonging" needs. |
Confidence Improves Self-Consistency in LLMs (2025.findings-acl)
Copied to clipboard
| Challenge: | Modern large language models (LLMs) demonstrate strong reasoning capabilities, driven in part by their capacity to generate a sequence of intermediate reasoning steps that lead them toward a final answer. |
| Approach: | They propose a method that performs a weighted majority vote based on confidence scores obtained directly from the model. |
| Outcome: | The proposed method outperforms self-consistency on nine models and four datasets, reducing the required number of reasoning paths by over 40% on average. |
None of the Above, Less of the Right Parallel Patterns in Human and LLM Performance on Multi-Choice Questions Answering (2025.findings-acl)
Copied to clipboard
| Challenge: | Multiple-choice exam questions with “None of the above” (NA) options have been extensively studied in educational testing . however, their impact on Large Language Models (LLMs) evaluation remains underexplored . |
| Approach: | They conduct systematic experiments with 28 LLMs on the MMLU benchmark to examine how NA options affect model performance and confidence calibration. |
| Outcome: | The results highlight important implications for benchmark design and raise questions about LLMs’ ability to handle uncertainty in real-world applications. |
In Search of the Lost Arch in Dialogue: A Dependency Dialogue Acts Corpus for Multi-Party Dialogues (2025.findings-acl)
Copied to clipboard
Jon Cai, Brendan King, Peyton Cameron, Susan Windisch Brown, Miriam Eckert, Dananjay Srinivas, George Arthur Baker, V Kate Everson, Martha Palmer, James Martin, Jeffrey Flanigan
| Challenge: | Understanding speaker intentions remains a challenge in NLP . a number of corpora annotated using theoretical frameworks of dialogue focus on utterance-level labeling of speaker intent, missing wider context, or the rhetorical structure of a dialogue. |
| Approach: | They propose to annotate a corpus of 33 dialogues and over 9,000 utterance units using the Dependency Dialogue Acts framework. |
| Outcome: | The proposed corpus spans four genres of multi-party conversations from different modalities. |
ProMind-LLM: Proactive Mental Health Care via Causal Reasoning with Sensor Data (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for mental health risk assessment rely on subjective textual records . however, these uncertainties can cause inconsistent and unreliable predictions . |
| Approach: | They propose a method that integrates objective behavior data alongside subjective mental records for robust mental health risk assessment. |
| Outcome: | The proposed approach achieves significant improvements over general LLMs. |
Debiasing Online Preference Learning via Preference Feature Preservation (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent preference learning frameworks for large language models (LLMs) simplify human preferences with binary pairwise comparisons and scalar rewards. |
| Approach: | They propose a preference feature preservation framework that preserves the distribution of human preference features and maps them throughout the online preference learning process. |
| Outcome: | The proposed framework maintains the distribution of human preference features and utilizing such rich signals throughout the online preference learning process. |
ShortGPT: Layers in Large Language Models are More Redundant Than You Expect (2025.findings-acl)
Copied to clipboard
Xin Men, Mingyu Xu, Qingyu Zhang, Qianhao Yuan, Bingning Wang, Hongyu Lin, Yaojie Lu, Xianpei Han, Weipeng Chen
| Challenge: | Recent studies have identified significant redundancy in large language models . quantization and pruning are two methods that reduce computational resources . |
| Approach: | They propose simple pruning methods that prune redundant layers based on their BI scores. |
| Outcome: | The proposed pruning methods demonstrate superior performance over previous pruning methods. |
ProjectEval: A Benchmark for Programming Agents Automated Evaluation on Project-Level Code Generation (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing benchmarks lack the ability to automatically evaluate from users’ perspective and lack the explainability of the results of LLM agents’ code generation capabilities. |
| Approach: | They propose a new benchmark for LLM agents' automated evaluation by simulating user interaction. |
| Outcome: | The proposed benchmark can evaluate the generated projects by user interaction simulation and by code similarity through existing objective indicators. |
Unveiling the Lack of LVLM Robustness to Fundamental Visual Variations: Why and Path Forward (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Vision Language Models (LVLMs) have shown impressive performance on various vision-language tasks. |
| Approach: | They propose a benchmark framework for evaluating Visual Variation Robustness of Large Vision Language Models that incorporates automated evaluation dataset generation and principled metrics for thorough robustness assessment. |
| Outcome: | The proposed framework identifies a vulnerability to visual variations affecting even advanced models that excel at complex vision-language tasks but significantly underperform on simple tasks like object recognition. |
DYNTEXT: Semantic-Aware Dynamic Text Sanitization for Privacy-Preserving LLM Inference (2025.findings-acl)
Copied to clipboard
Juhua Zhang, Zhiliang Tian, Minghang Zhu, Yiping Song, Taishu Sheng, Siyi Yang, Qiunan Du, Xinwang Liu, Minlie Huang, Dongsheng Li
| Challenge: | Existing methods to protect privacy of sensitive data are differential privacy (DP) and DP is used to protect users from privacy leakage. |
| Approach: | They propose an LDP-based Dynamic Text sanitization for privacy-preserving LLM inference that dynamically constructs semantic-aware adjacency lists of sensitive tokens to sample non-sensitive tokens for perturbation. |
| Outcome: | The proposed model excels on three datasets. |
InImageTrans: Multimodal LLM-based Text Image Machine Translation (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing multimodal large language models suffer from repetition and omission hallucinations when transferred to text image machine translation task. |
| Approach: | They propose an efficient MLLM named InImageTrans for TiMT and a method for advancing it. |
| Outcome: | The proposed method outperforms existing open-source MLLMs on the MCiT benchmark. |
FRAME: Boosting LLMs with A Four-Quadrant Multi-Stage Pretraining Strategy (2025.findings-acl)
Copied to clipboard
Xuemiao Zhang, Feiyu Duan, Xu Liangyu, Yongwei Zhou, Sirui Wang, Rongxiang Weng, Jingang Wang, Xunliang Cai
| Challenge: | Multi-stage pretraining methods lack quantitative criteria for data partitioning and instead rely on intuitive heuristics. |
| Approach: | They propose a Four-quadRAnt Multi-stage prEtraining strategy that partitions data into four quadrants to achieve significant loss reductions four times. |
| Outcome: | The proposed strategy achieves 16.8% improvement over random across MMLU and CMMLU for the 3B model. |
When Large Language Models Meet Speech: A Survey on Integration Approaches (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent advances in large language models have spurred interest in expanding their application beyond text-based tasks. |
| Approach: | They propose to categorize the integration of speech with LLMs into three main approaches . they demonstrate how these methods are applied across various speech-related applications . |
| Outcome: | The proposed methods are applied across speech-related applications and highlight the challenges in this field to offer inspiration for future research. |
KE-MHISTO: Towards a Multilingual Historical Knowledge Extraction Benchmark for Addressing the Long-Tail Problem (2025.findings-acl)
Copied to clipboard
Arianna Graciotti, Leonardo Piano, Nicolas Lazzari, Enrico Daga, Rocco Tripodi, Valentina Presutti, Livio Pompianu
| Challenge: | Large Language Models struggle when probed for long-tail knowledge due to the inherent sparsity of such data. |
| Approach: | They propose a multilingual benchmark for Entity Linking and Question Answering in the domain of historical music knowledge that provides broader coverage of long-tail knowledge. |
| Outcome: | The proposed model provides broader coverage of long-tail knowledge compared to existing models. |
TailorKV: A Hybrid Framework for Long-Context Inference via Tailored KV Cache Optimization (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing work mitigates memory overhead by offloading or compressing the Key-Value cache. |
| Approach: | They propose a method that integrates quantization and offloading into a generative large language model by using a hybrid compression method. |
| Outcome: | The proposed method outperforms the state-of-the-art in long-context evaluations. |
The Elephant in the Room: Exploring the Role of Neutral Words in Language Model Group-Agnostic Debiasing (2025.findings-acl)
Copied to clipboard
Xinwei Guo, Jiashi Gao, Junlei Zhou, Jiaxin Zhang, Guanhua Chen, Xiangyu Zhao, Quanying Liu, Haiyan Wu, Xin Yao, Xuetao Wei
| Challenge: | Large Language Models (LLMs) are increasingly integrated into our daily lives, raising ethical concerns, especially about perpetuating stereotypes. |
| Approach: | They propose a method that incorporates a neutral word semantics-based loss function to alleviate the deterioration of the LMS during debiasing. |
| Outcome: | The proposed method alleviates the deterioration of the Language Modeling Score (LMS) by incorporating a neutral word semantics-based loss function. |
LLMs Can Achieve High-quality Simultaneous Machine Translation as Efficiently as Offline (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models perform well in offline machine translation when the complete source sentence is provided . however, in many real scenarios, the source tokens arrive in a streaming manner and simultaneous machine translation is required . |
| Approach: | They propose a new paradigm that includes constructing supervised fine-tuning data for simultaneous machine translation (SiMT) to achieve SiMT, source and target tokens are rearranged into interleaved sequences, separated by special tokens according to varying latency requirements. |
| Outcome: | The proposed approach achieves state-of-the-art performance across various SiMT benchmarks and evaluation metrics while maintaining efficient auto-regressive decoding. |
Beyond Completion: A Foundation Model for General Knowledge Graph Reasoning (2025.findings-acl)
Copied to clipboard
Yin Hua, Zhiqiang Liu, Mingyang Chen, Zheng Fang, Chi Man Wong, Lingxiao Li, Chi Man Vong, Huajun Chen, Wen Zhang
| Challenge: | Existing foundation models for general knowledge graph reasoning have focused on their structural aspects, with most efforts restricted to in-KG tasks. |
| Approach: | They propose a conditional encoding architecture that bridges the gap between textual and structural modalities, enabling seamless integration. |
| Outcome: | The proposed model outperforms baseline models on 28 datasets and is generalized to out-of-KG tasks. |
Generative Error Correction for Emotion-aware Speech-to-text Translation (2025.findings-acl)
Copied to clipboard
| Challenge: | Despite recent advances in speech-to-text translation, the impact of the emotion content has been overlooked. |
| Approach: | They propose to use generative error correction (GER) to generate the translation based on the decoded N-best hypotheses and combine emotion and sentiment labels into the LLM finetuning process to enable the model to consider the emotion content. |
| Outcome: | The proposed model can translate speech in English-Chinese using GER and emotion and sentiment labels. |
SynapticRAG: Enhancing Temporal Memory Retrieval in Large Language Models through Synaptic Mechanisms (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing retrieval methods in Large Language Models show degradation in accuracy when handling temporally distributed conversations. |
| Approach: | They propose a method that combines temporal triggers and synaptic-like stimulus propagation to identify relevant dialogue histories. |
| Outcome: | The proposed approach improves on four datasets of English, Chinese and Japanese compared to state-of-the-art retrieval methods by 14.66% points. |
Localizing and Mitigating Errors in Long-form Question Answering (2025.findings-acl)
Copied to clipboard
| Challenge: | Long-form question answering (LFQA) answers are prone to hallucinations and factual inconsistencies, challenging their faithful evaluation. |
| Approach: | They propose a dataset with localized error annotations for human-written and model-generated LFQA answers. |
| Outcome: | The proposed approach reduces errors and improves quality of the answers across multiple models. |
EMGLLM: Data-to-Text Alignment for Electromyogram Diagnosis Generation with Medical Numerical Data Encoding (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing Large Language Models struggle to interpret EMG tables . EMGLLM is a data-to-text model for medical examination tables based on electrical signals . |
| Approach: | They propose a data-to-text model that aligns EMG data into word embeddings that reflect health degree. |
| Outcome: | The proposed model outperforms baseline models in understanding EMG tables and generating high-quality diagnoses. |
LLMVoX: Autoregressive Streaming Text-to-Speech Model for Any LLM (2025.findings-acl)
Copied to clipboard
Sambal Shikhar, Mohammed Irfan Kurpath, Sahal Shaji Mullappilly, Jean Lahoud, Fahad Shahbaz Khan, Rao Muhammad Anwer, Salman Khan, Hisham Cholakkal
| Challenge: | Existing speech-enabled LLMs degrade conversational quality by modifying the LLM, compromising its linguistic capabilities. |
| Approach: | They propose a lightweight 30M-parameter, LLM-agnostic, autoregressive streaming TTS system that generates high-quality speech with low latency. |
| Outcome: | The proposed system achieves a significantly lower word error rate compared to speech-enabled LLMs while operating at comparable latency. |
Act2P: LLM-Driven Online Dialogue Act Classification for Power Analysis (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing studies focus on explicit utterance functions, overlooking the implicit power dynamics embedded in dialogue. |
| Approach: | They propose an online Dialogue Act Classification and Dynamic Power Analysis framework based on large language models to integrate dialogue act classification with power quantification. |
| Outcome: | The proposed framework outperforms existing methods in online scenarios and shows that dialogue power is distributed and dynamically transferred. |
MELABenchv1: Benchmarking Large Language Models against Smaller Fine-Tuned Models for Low-Resource Maltese NLP (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have shown remarkable performance across various NLP tasks, largely due to their generalisability and ability to perform tasks without additional training. |
| Approach: | They evaluate the performance of 55 publicly available Large Language Models on Maltese, a low-resource language, using a newly introduced benchmark covering 11 discriminative and generative tasks. |
| Outcome: | The proposed models perform poorly on discriminative and generative tasks and smaller fine-tuned models perform better across all tasks. |
TRATES: Trait-Specific Rubric-Assisted Cross-Prompt Essay Scoring (2025.findings-acl)
Copied to clipboard
| Challenge: | Automated Essay Scoring (AES) has seen significant progress in assessing writing ability and trait scoring. |
| Approach: | They propose a trait-specific and rubric-based cross-prompt AES framework that is generic yet specific to the underlying trait. |
| Outcome: | The proposed framework achieves state-of-the-art across all traits on a widely-used dataset, with the generated LLM-based features being the most significant. |
DAST: Context-Aware Compression in LLMs via Dynamic Allocation of Soft Tokens (2025.findings-acl)
Copied to clipboard
Shaoshen Chen, Yangning Li, Zishan Xu, Yongqin Zeng, Shunlong Wu, Xinshuo Hu, Zifei Shan, Xin Su, Jiwei Tang, Yinghui Li, Hai-Tao Zheng
| Challenge: | Existing semantic vector-based compression methods do not account for the intrinsic information density variations between context chunks, instead allocating soft tokens uniformly across context chunk. |
| Approach: | They propose a method that leverages the LLM's intrinsic understanding of contextual relevance to guide compression. |
| Outcome: | The proposed method surpasses state-of-the-art methods on long context tasks. |
A Multi-Expert Structural-Semantic Hybrid Framework for Unveiling Historical Patterns in Temporal Knowledge Graphs (2025.findings-acl)
Copied to clipboard
Yimin Deng, Yuxia Wu, Yejing Wang, Guoshuai Zhao, Li Zhu, Qidong Liu, Derong Xu, Zichuan Fu, Xian Wu, Yefeng Zheng, Xiangyu Zhao, Xueming Qian
| Challenge: | Existing methods focus on graph structure learning or semantic reasoning, lacking the capability to capture the inherent differences between historical and non-historical events. |
| Approach: | They propose a temporal knowledge graph reasoning framework that integrates both structural and semantic information to guide the reasoning process for different events. |
| Outcome: | The proposed framework integrates structural and semantic information to predict future events . it can provide evidence for many downstream tasks, including situation analysis and political decision making . |
MWPO: Enhancing LLMs Performance through Multi-Weight Preference Strength and Length Optimization (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing offline alternatives to Reinforcement Learning from Human Feedback (RLHF) are available at https://github.com/AIR-hl/MWPO. |
| Approach: | They propose an offline method to optimize preference pairs based on implicit reward margins and response length margins by reweighting them using a geometric mixture. |
| Outcome: | The proposed method outperforms state-of-the-art methods on four different scales and reduces generation length by 9.4%. |
CLEAR: Character Unlearning in Textual and Visual Modalities (2025.findings-acl)
Copied to clipboard
Alexey Dontsov, Dmitrii Korzh, Alexey Zhavoronkin, Boris Mikheev, Denis Bobkov, Aibek Alanov, Oleg Rogov, Ivan Oseledets, Elena Tutubalina
| Challenge: | Existing methods for removing private or hazardous data from deep learning models are focused on single-modality models. |
| Approach: | They propose CLEAR, the first open-source benchmark specifically for MMU. CLEAR contains 200 fictitious individuals and 3,700 images linked with corresponding question-answer pairs. |
| Outcome: | The proposed benchmarks show that unlearning both modalities outperform single-modality approaches. |
Assessing the Reasoning Capabilities of LLMs in the context of Evidence-based Claim Verification (2025.findings-acl)
Copied to clipboard
John Dougrez-Lewis, Mahmud Elahi Akhter, Federico Ruggeri, Sebastian Löbbers, Yulan He, Maria Liakata
| Challenge: | Large Language Models (LLMs) have shown remarkable proficiency in complex tasks where reasoning capabilities are paramount. |
| Approach: | They propose a framework to break down claims into atomic reasoning types needed for verification. |
| Outcome: | The proposed framework breaks down claims into atomic reasoning types needed for verification. |
Language Models Lack Temporal Generalization and Bigger is Not Better (2025.findings-acl)
Copied to clipboard
| Challenge: | 450 encoder models are fine-tuned on 15 data splits on a task to detect events in Early Modern Dutch archival texts. |
| Approach: | They propose to fine tune six encoder models that have been pretrained with very different data on a task in Early Modern Dutch archival texts. |
| Outcome: | The proposed model is fine tuned with 5 seeds on 15 different data splits and reaches highest F1 performance. |
DiffLM: Controllable Synthetic Data Generation via Diffusion Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent advances in large language models (LLMs) have significantly enhanced their knowledge and generative capabilities, leading to a surge of interest in leveraging LLMs for high-quality data synthesis. |
| Approach: | They propose a controllable data synthesis framework based on variational autoencoder which leverages diffusion models to reserve more information of original distribution and format structure in the learned latent distribution. |
| Outcome: | The proposed framework generates high-quality data with performance exceeding that of real data by 2%–7% on seven real-world datasets. |
Uncertainty Unveiled: Can Exposure to More In-context Examples Mitigate Uncertainty for Large Language Models? (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent advances in handling long sequences have unlocked new possibilities for long-context in-contact learning (ICL). |
| Approach: | They investigate how increased examples influence predictive uncertainty . they quantify uncertainty across different “shot” configurations and focus on EU . |
| Outcome: | The proposed model reduces uncertainty in simple and complex tasks by injecting task-specific knowledge. |
ToolSpectrum: Towards Personalized Tool Utilization for Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing approaches focus on functional tool selection following user instructions while overlooking the critical role of context-aware personalization in tool selection. |
| Approach: | They propose a benchmark to evaluate LLMs’ capabilities in personalized tool utilization. |
| Outcome: | The proposed benchmark evaluates LLMs' capabilities in personalized tool utilization. |
Reverse Preference Optimization for Complex Instruction Following (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for identifying and evaluating preference pairs with multiple constraints are noisy. |
| Approach: | They propose a method that dynamically reverses constraints to ensure the chosen response is perfect. |
| Outcome: | The proposed method reduces noise in preference pairs by reversing constraints to ensure the chosen response is perfect. |
MMS-LLaMA: Efficient LLM-based Audio-Visual Speech Recognition with Minimal Multimodal Speech Tokens (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent Large Language Model (LLM) based AVSR systems incur high computational costs due to high temporal resolution of audio-visual speech. |
| Approach: | They propose an efficient multimodal speech LLM framework that minimizes token length while preserving essential linguistic content. |
| Outcome: | The proposed approach reduces token usage by 86% while using only 3.5 tokens per second. |
Def-DTS: Deductive Reasoning for Open-domain Dialogue Topic Segmentation (2025.findings-acl)
Copied to clipboard
| Challenge: | Dialogue Topic Segmentation (DTS) suffers from data shortage, labeling ambiguity, and incremental complexity of recently proposed solutions. |
| Approach: | They propose a method that employs a structured prompting approach for context summarization, utterance intent classification, and deductive topic shift detection. |
| Outcome: | The proposed method outperforms traditional and state-of-the-art approaches in various dialogue settings. |
Exploring Jailbreak Attacks on LLMs through Intent Concealment and Diversion (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing jailbreak methods face an excessive number of iterative queries and poor generalization across models. |
| Approach: | They propose a jailbreak method that employs **I**ntent **C**oncealment and div**E**rsion to circumvent security constraints. |
| Outcome: | The proposed method outperforms existing jailbreak techniques in question-answering and text-generation tasks. |
Verbosity-Aware Rationale Reduction: Sentence-Level Rationale Reduction for Efficient and Effective Reasoning (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent advances in Large Language Models (LLMs) demonstrate remarkable reasoning capabilities comparable to human cognitive abilities. |
| Approach: | They propose a sentence-level rationale reduction framework that leverages likelihood-based criteria to selectively remove redundant reasoning sentences while preserving reasoning capabilities. |
| Outcome: | The proposed framework improves performance by 7.71% while reducing token generation by 19.87% compared to models trained with complete reasoning paths. |
Exploring the Role of Mental Health Conversational Agents in Training Medical Students and Professionals: A Systematic Literature Review (2025.findings-acl)
Copied to clipboard
| Challenge: | This systematic review analyses 38 studies on AI-powered conversational agents in mental health education and training . traditional training methods provide valuable but expensive and inherently limited learning opportunities . early pioneers like Woebot and Wysa demonstrated a groundbreaking insight: machines could engage in meaningful therapeutic interactions. |
| Approach: | They analyse 38 studies on AI-powered conversational agents in mental health education and training . findings reveal that AI-based approaches dominate the field, with training as the application area being the most prevalent . |
| Outcome: | The systematic review of 38 studies on AI-powered conversational agents in mental health education and training (MHET) reveals that AI-based approaches dominate the field, with training as the application area being the most prevalent. |
Bandit-Based Prompt Design Strategy Selection Improves Prompt Optimizers (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing prompt optimization methods have found effective prompts, but they often differ from sophisticated prompts carefully designed by human experts. |
| Approach: | They propose to integrate prompt design strategies into prompt optimization by using a Thompson sampling-based approach. |
| Outcome: | The proposed method incorporates prompt design strategies into the prompt optimization process. |
STORYTELLER: An Enhanced Plot-Planning Framework for Coherent and Cohesive Story Generation (2025.findings-acl)
Copied to clipboard
Jiaming Li, Yukun Chen, Ziqiang Liu, Minghuan Tan, Lei Zhang, Yunshui Li, Run Luo, Longze Chen, Jing Luo, Ahmadreza Argha, Hamid Alinejad-Rokny, Wei Zhou, Min Yang
| Challenge: | Existing methods for storytelling lack coherence and consistency, compromising the overall storytelling experience. |
| Approach: | They propose a novel approach that improves the coherence and consistency of automatically generated stories by managing plot nodes and enabling dynamic interactions between different parts of the story. |
| Outcome: | The proposed approach outperforms existing methods in 84.33% of the trials. |
SelectLLM: Query-Aware Efficient Selection Algorithm for Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing large language models struggle with complex tasks such as factually-grounded reasoning and planning due to inherent training biases, model size constraints, and the quality or diversity of pre-training datasets. |
| Approach: | They propose a novel algorithm to select the most suitable LLMs from a large pool and use it to efficiently generalize and perform tasks. |
| Outcome: | The proposed model outperforms existing ensemble-based baselines and achieves competitive performance with similarly sized top-performing LLMs while maintaining efficiency. |
SkyLLM: Cross-LLM-APIs Federation for Cost-effective Query Processing (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) have demonstrated exceptional capabilities across a wide range of tasks, from text generation to complex problem-solving. |
| Approach: | They propose a system which federates multiple LLM APIs and dynamically assigns a non-empty subset of these APIs to each query prior to inference. |
| Outcome: | The proposed system can match the most accurate LLM with the lowest cost while cutting costs by 67.8%. |
Matina: A Culturally-Aligned Persian Language Model Using Multiple LoRA Experts (2025.findings-acl)
Copied to clipboard
Sara Bourbour Hosseinbeigi, MohammadAli SeifKashani, Javad Seraj, Fatemeh Taherinezhad, Ali Nafisi, Fatemeh Nadi, Iman Barati, Hosein Hasani, Mostafa Amiri, Mostafa Masoudi
| Challenge: | Existing Large language models fail to accurately model underrepresented languages and cultures, limiting their applicability and acceptance. |
| Approach: | They develop a Persian-focused multi-expert model that incorporates Iranian cultural values and linguistic structures. |
| Outcome: | The proposed model outperforms baseline models in task performance and user satisfaction. |
PM3-KIE: A Probabilistic Multi-Task Meta-Model for Document Key Information Extraction (2025.findings-acl)
Copied to clipboard
| Challenge: | a probabilistic multi-task meta-model for key information extraction is proposed . entity-level models capture spatial and visual cues, while fine-grained token classification models capture logical dependencies. |
| Approach: | They propose a probabilistic multi-task meta-model that integrates fine-grained token classification and coarse-grain entity extraction models. |
| Outcome: | The proposed model outperforms state-of-the-art models and a stacked ensemble on two datasets. |
TechniqueRAG: Retrieval Augmented Generation for Adversarial Technique Annotation in Cyber Threat Intelligence Text (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for identifying adversarial techniques in security texts face a trade-off: generic models with limited domain precision or resource-intensive pipelines. |
| Approach: | They propose a domain-specific retrieval-augmented generation framework that integrates off-the-shelf retrievers, instruction-tuned LLMs, and minimal text–technique pairs. |
| Outcome: | The proposed framework improves retrieval quality and domain specificity without extensive optimizations. |
G2S: A General-to-Specific Learning Framework for Temporal Knowledge Graph Forecasting with Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent studies have introduced Large Language Models (LLMs) for this task to enhance the models’ generalization abilities. |
| Approach: | They propose a General-to-Specific learning framework that disentangles the learning processes of two kinds of knowledge in a temporal temporal structure. |
| Outcome: | The proposed framework disentangles the learning processes of the above two kinds of knowledge and improves their generalization abilities. |
Disentangling Reasoning Tokens and Boilerplate Tokens For Language Model Fine-tuning (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing approaches to enhance agent capabilities for Large Language Models treat all tokens equally . however, reasoning tokens versus boilerplate tokens differ in importance and learning complexity . recent research has focused on enhancing agent capabilities in large language models . |
| Approach: | They propose a Shuffle-Aware Discriminator (SHAD) for adaptive token discrimination . they propose SHAD method which adaptively emphasizes reasoning tokens during fine-tuning . |
| Outcome: | The proposed method improves performance over standard fine-tuning methods. |
APT: Improving Specialist LLM Performance with Weakness Case Acquisition and Iterative Preference Training (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models often require domain-specific fine-tuning to address targeted tasks, which risks degrading their general capabilities. |
| Approach: | They propose to use self-generated dis-preferred weakness data to enhance model performance with a targeted training approach that minimizes interference with existing knowledge base. |
| Outcome: | The proposed approach ensures no reduction in generic capacity and achieves superior performance on downstream tasks compared to existing methods. |
EasyEA: Large Language Model is All You Need in Entity Alignment Between Knowledge Graphs (2025.findings-acl)
Copied to clipboard
| Challenge: | Entity alignment (EA) aims to identify entities in different knowledge graphs (KGs) that represent the same real-world object. |
| Approach: | They propose an end-to-end EA framework based on large language models that requires no training to implement. |
| Outcome: | The proposed framework significantly reduces the reliance on seed entity pairs while achieving state-of-the-art (SOTA) performance on diverse datasets. |
An Adaptive Multi-Threshold Loss and a General Framework for Collaborating Losses in Document-Level Relation Extraction (2025.findings-acl)
Copied to clipboard
| Challenge: | Document-level relation extraction (DocRE) aims to identify relations for a given entity pair within a document. |
| Approach: | They propose to partition the label space into different sub-label spaces and learn an adaptive threshold for each sub-labeled space. |
| Outcome: | The proposed model outperforms single-loss methods on the concurrent application of multiple losses across four datasets. |
RoleMRC: A Fine-Grained Composite Benchmark for Role-Playing and Instruction-Following (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing role-playing datasets mostly contribute to controlling role style and knowledge boundaries, but overlook role-following in instruction-follower scenarios. |
| Approach: | They propose a fine-grained role-playing and instruction-following composite benchmark, named RoleMRC, which includes multi-turn dialogues between ideal roles and humans, including free chats or discussions upon given passages . |
| Outcome: | The proposed model improves instruction-following without compromising general role-playing and reasoning capabilities. |
C²RBench: A Chinese Complex Reasoning Benchmark for Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing benchmarks often fail to capture complex multi-step reasoning demands inherent in real-world scenarios. |
| Approach: | They propose a benchmark to evaluate multi-step, multimodal advanced reasoning of large language models. |
| Outcome: | The proposed benchmark exceeds existing benchmarks in cognitive complexity and accuracy by over 90% . it features 1,115 carefully curated Chinese tasks organized into eight domain-specific subsets . evaluations of 20 LLMs and 24 multimodal large language models reveal critical performance gaps . |
Unlocking LLMs’ Self-Improvement Capacity with Autonomous Learning for Domain Adaptation (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing models that use self-supervised and instruction fine-tuning can be trained using unlabeled corpora. |
| Approach: | They propose to use unlabeled target corpora to adapt large language models to new domains . they propose to employ self-supervised pre-training and instruction fine-tuning methods . |
| Outcome: | The proposed model can adapt to new domains using only a large amount of unlabeled target corpora. |
How Personality Traits Shape LLM Risk-Taking Behaviour (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models are increasingly used as autonomous agents for simulation and decision-making. |
| Approach: | They apply Cumulative Prospect Theory and the Big Five personality framework to investigate the relationship between LLMs’ personality traits and risk-propensity. |
| Outcome: | The proposed models show that they are risk-neutral rational agents, whereas others show lower neuroticism and higher conscientiousness and Agreeability traits. |
Word-Level Detection of Code-Mixed Hate Speech with Multilingual Domain Transfer (2025.findings-acl)
Copied to clipboard
| Challenge: | a growing problem in language detection tasks is code-mixing, a combination of more than one language . lack of available datasets for code-mixing causes the problem . authors propose a multilingual approach to code-matching . |
| Approach: | They propose to use an annotated hate speech dataset to detect code-mixing in profane language . they propose to apply bilingual fine-tuned models to code-mixed hate speech in german rap lyrics . |
| Outcome: | The proposed model can detect code-mixed hate speech and neologisms in German rap lyrics . the proposed model is more nuanced than binary classification . |
Evaluation of Attribution Bias in Generator-Aware Retrieval-Augmented Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Prior work has focused on improving and evaluating the attribution quality of large language models (LLMs) but this may come at the expense of inducing biases in the attributed answers. |
| Approach: | They propose to evaluate attribution sensitivity and bias with respect to authorship information in large language models (LLMs) in retrieval-augmented generation pipelines. |
| Outcome: | The proposed framework can significantly improve the attribution quality of large language models (LLMs) in retrieval-augmented generation pipelines by adding authorship information to source documents. |
Implicit Cross-Lingual Rewarding for Efficient Multilingual Preference Alignment (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing approaches to align English LLMs with human preferences rely on expensive human annotations or advanced multilingual preference alignment models. |
| Approach: | They propose a method that captures learned preferences from English models by implicit rewards . they annotate preference relations in cross-lingual instruction-following pairs using English . |
| Outcome: | The proposed approach captures learned preferences from well-aligned English models by implicit rewards and transfers them to other languages through iterative training. |
Diagnosing Failures in Large Language Models’ Answers: Integrating Error Attribution into Evaluation Framework (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing evaluation models lack error attribution capability due to their proprietary nature. |
| Approach: | They propose a misattribution framework with 6 primary and 15 secondary categories to facilitate in-depth analysis. |
| Outcome: | The proposed framework is based on a dataset specifically designed for error attribution, along with the corresponding scores and feedback. |
Encode Errors: Representational Retrieval of In-Context Demonstrations for Multilingual Grammatical Error Correction (2025.findings-acl)
Copied to clipboard
| Challenge: | a novel method for encoding fine-grained error patterns improves performance on GEC. |
| Approach: | They propose a method for encoding grammatical errors from LLMs' internal states using a GER method. |
| Outcome: | The proposed method significantly boosts performance in ICL settings on multilingual GEC datasets. |
Preference Curriculum: LLMs Should Always Be Pretrained on Their Preferred Data (2025.findings-acl)
Copied to clipboard
Xuemiao Zhang, Xu Liangyu, Feiyu Duan, Yongwei Zhou, Sirui Wang, Rongxiang Weng, Jingang Wang, Xunliang Cai
| Challenge: | Existing methods of uniformly sampling data throughout the pretraining process are suboptimal because they overlook the model's evolving data preferences. |
| Approach: | They propose a Perplexity Difference (PD) based Preference Curriculum learning framework which perceives and uses the data preferred by LLMs as their capabilities improve . they propose PDPC to complete the arrangement of the dataset offline and ensure continuous training without interruption. |
| Outcome: | The proposed framework surpasses baselines on 1.3B and 3B models and achieves an increased average accuracy of over 8.1% across MMLU and CMMLU. |
Can Input Attributions Explain Inductive Reasoning in In-Context Learning? (2025.findings-acl)
Copied to clipboard
| Challenge: | interpreting the internal process of neural models has long been a challenge . despite rapid progress, there are still questions bridging the IA and MI eras . |
| Approach: | They propose to use input attribution methods to interpret in-context learning . they find that a certain simple IA method works best in large models . |
| Outcome: | The proposed method is the best for interpreting LLM-based ICL, but the larger the model, the harder it is to interpret it. |
Modal Dependency Parsing via Biaffine Attention with Self-Loop (2025.findings-acl)
Copied to clipboard
| Challenge: | A modal dependency structure is a web of connections between events and sources of information in a document that allows for tracing of who-said-what with what levels of certainty. |
| Approach: | They propose a modal dependency structure that integrates biaffine attention with a large language model to optimize against domain-specific challenges of modal dependence parsing. |
| Outcome: | The proposed approach outperforms the previous state-of-the-art on English and Chinese datasets by 2% and 4% respectively. |
Beyond Profile: From Surface-Level Facts to Deep Persona Simulation in LLMs (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing approaches to persona simulation large language models (LLMs) focus on learning basic biographical information, or using limited role-play dialogue datasets to capture a character’s responses. |
| Approach: | They propose to train characters using a linguistic structure and a style-tuning mechanism that allows a general linguistic style expert to collaborate with other task-specific experts to better understand their thoughts. |
| Outcome: | The proposed model outperforms baselines on linguistic accuracy and opinion comprehension on three tasks for Lu Xun's essay collection. |
Measuring What Makes You Unique: Difference-Aware User Modeling for Enhancing LLM Personalization (2025.findings-acl)
Copied to clipboard
| Challenge: | Extensive experiments on real-world datasets demonstrate that DPL significantly enhances LLM personalization. |
| Approach: | They propose a novel approach that emphasizes extracting inter-user differences to enhance LLM personalization. |
| Outcome: | The proposed approach extracts inter-user differences to enhance LLM personalization. |
VideoRAG: Retrieval-Augmented Generation over Video Corpus (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing approaches to generating models rely on text and images, but video content is a rich source of multimodal knowledge. |
| Approach: | They propose a framework that dynamically retrieves videos based on their relevance with queries . they use large video language models to represent video content for retrieval . |
| Outcome: | The proposed framework retrieves videos based on relevance with queries and integrates both visual and textual information. |
Synergistic Augmentation: Enhancing Cross-Domain Zero-Shot Slot Filling with Small Model-Assisted Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing approaches to slot filling are limited due to data scarcity and timeconsuming efforts. |
| Approach: | They propose a framework that harnesses the power of a small model to augment inferential capabilities of LLMs without additional training. |
| Outcome: | The proposed framework improves slot filling performance on a spoken language dataset and a NER dataset. |
A Classifier of Word-Level Variants in Witnesses of Biblical Hebrew Manuscripts (2025.findings-acl)
Copied to clipboard
| Challenge: | a strong classifier (F1 value of 0.80) is trained to predict the category of difference between word pairs as present in collated (aligned) pairs of witnesses. |
| Approach: | The project is based on the relationship between available witnesses of biblical Hebrew and a strong classifier (F1 value of 0.80) is trained to predict the category of difference between word pairs as present in collated pairs of witnesses. |
| Outcome: | The proposed model is non-neural and uses part-of-speech tags, hand-crafted rules per category and synthetically derived data. |
NOVA: An Iterative Planning Framework for Enhancing Scientific Innovation with Large Language Models (2025.findings-acl)
Copied to clipboard
Xiang Hu, Hongyu Fu, Jinge Wang, Yifeng Wang, Zhikun Li, Renjun Xu, Yu Lu, Yaochu Jin, Lili Pan, Zhenzhong Lan
| Challenge: | Existing approaches to generate research ideas rely on retrieval or prompt engineering to generate ideas. |
| Approach: | They propose a method that uses iterative planning and search to boost creative potential of LLMs by integrating external knowledge with broader and deeper insights. |
| Outcome: | The proposed method outperforms the current state-of-the-art in generating 2.5 times more top-rated ideas based on 170 seed papers in a Swiss Tournament evaluation. |
Query-Driven Multimodal GraphRAG: Dynamic Local Knowledge Graph Construction for Online Reasoning (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing approaches to build knowledge graphs with LLMs are constrained by static knowledge bases and ineffective multimodal data integration. |
| Approach: | They propose a Query-Driven Multimodal GraphRAG framework that dynamically constructs local knowledge graphs tailored to query semantics. |
| Outcome: | The proposed framework outperforms unsupervised competitors in cross-modal understanding of complex queries. |
A Survey of Uncertainty Estimation Methods on Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) have demonstrated remarkable capabilities but could produce biased, hallucinated, or non-factual responses. |
| Approach: | They propose to conduct extensive experimental evaluations of LLM uncertainty estimation methods . large language models have demonstrated remarkable capabilities across tasks . |
| Outcome: | The proposed method could produce biased, hallucinated, or non-factual responses . a lack of comprehensive surveys on LLM uncertainty estimation is a problem . |
Beyond Single-Value Metrics: Evaluating and Enhancing LLM Unlearning with Cognitive Diagnosis (2025.findings-acl)
Copied to clipboard
Yicheng Lang, Kehan Guo, Yue Huang, Yujun Zhou, Haomin Zhuang, Tianyu Yang, Yao Su, Xiangliang Zhang
| Challenge: | Large Language Models (LLMs) have been used to remove harmful knowledge and undesirable capabilities. |
| Approach: | They propose a framework that leverages Cognitive Diagnosis Modeling to evaluate LLM unlearning. |
| Outcome: | The proposed framework enhances evaluation and facilitates removal of harmful abilities. |
Natural Language Processing in Support of Evidence-based Medicine: A Scoping Review (2025.findings-acl)
Copied to clipboard
| Challenge: | Evidence-based medicine (EBM) is at the forefront of modern healthcare, emphasizing the use of the best available scientific evidence to guide clinical decisions. |
| Approach: | They propose to investigate the use of Natural Language Processing (NLP) techniques to identify, appraise, synthesize, apply, and disseminate evidence in EBM. |
| Outcome: | The proposed methods support the five fundamental steps of EBM—Ask, Acquire, Appraise, Apply, and Assess. |
How do Transformer Embeddings Represent Compositions? A Functional Analysis (2025.findings-acl)
Copied to clipboard
| Challenge: | Despite the popularity of transformer-based models, little is known about how they represent compound words and whether they are compositional. |
| Approach: | They evaluate compositionality in mistral, OpenAI Large, and Google embedding models and compare them with BERT. |
| Outcome: | The proposed models perform best in addition, multiplication, dilation, regression, and the classic vector addition model performs almost as well as any other model. |
Entriever: Energy-based Retriever for Knowledge-Grounded Dialog Systems (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing knowledge retrieval methods fail to account for interrelationship between knowledge pieces . however, current methods fail in a situation where multiple knowledge pieces are relevant . |
| Approach: | They propose an energy-based retriever that directly models the candidate retrieval results as a whole instead of modeling the knowledge pieces separately. |
| Outcome: | The proposed retriever outperforms the baseline energy-based retriever in knowledge retrieval tasks. |
MONTROSE: LLM-driven Monte Carlo Tree Search Self-Refinement for Cross-Domain Rumor Detection (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing feature alignment methods are susceptible to task interference during training. |
| Approach: | MONTROSE is a cross-domain rumor detection method that generates high-quality synthetic data for the target domain and a domain-sharpness-aware approach to train models with these synthetic data. |
| Outcome: | Experiments show that MONTROSE improves in cross-domain rumor detection. |
PEToolLLM: Towards Personalized Tool Learning in Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing tool learning studies focus on general-purpose tool-use capability, but ignore the importance of personalized tool-user preferences. |
| Approach: | They propose a framework to adapt Large Language Models to personalized tool learning task, which is trained through supervised fine-tuning and direct preference optimization. |
| Outcome: | Extensive experiments on PEToolBench show that the proposed framework outperforms existing LLMs in the personalized tool learning task. |
A Comprehensive Graph Framework for Question Answering with Mode-Seeking Preference Alignment (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing studies struggle with achieving global understanding of large language models . GraphMPA is a graph-based framework with mode-seeking preference alignment . |
| Approach: | They propose a graph-based framework with mode-seeking preference alignment to improve model outputs. |
| Outcome: | The proposed framework constructs a hierarchical document graph mimicking human cognitive processes for information understanding and synthesis. |
A MISMATCHED Benchmark for Scientific Natural Language Inference (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing datasets for scientific NLI are derived from various computer science domains, whereas non-CS domains are completely ignored. |
| Approach: | They propose a scientific natural language inference benchmark called MisMatched that incorporates sentence pairs having an implicit scientific NLI relation into model training. |
| Outcome: | The proposed benchmark covers three non-CS domains and contains 2,700 human annotated sentence pairs. |
TagRouter: Learning Route to LLMs through Tags for Open-Domain Text Generation Tasks (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing models with limited performance and limited training can be difficult to use in large-scale applications. |
| Approach: | They propose a training-free model routing method that optimizes synergy among multiple LLMs for open-domain text generation tasks. |
| Outcome: | The proposed method outperforms 13 baseline models and reduces costs by 17.20%. |
The Reasoning-Memorization Interplay in Language Models Is Mediated by a Single Direction (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models excel on a variety of reasoning benchmarks, but struggle to generalize to unseen questions due to over-reliance on memorized training examples. |
| Approach: | They propose to identify a set of linear features in the model’s residual stream that govern the balance between genuine reasoning and memory recall. |
| Outcome: | The proposed model can be manipulated to activate the most relevant problem-solving capabilities during answer generation. |
MPBench: A Comprehensive Multimodal Reasoning Benchmark for Process Errors Identification (2025.findings-acl)
Copied to clipboard
xu Zhao Pan, Pengfei Zhou, Jiaxin Ai, Wangbo Zhao, Kai Wang, Xiaojiang Peng, Wenqi Shao, Hongxun Yao, Kaipeng Zhang
| Challenge: | Existing benchmarks of large language models focus on error detection, neglecting other scenarios like reasoning search. |
| Approach: | et al. propose a multi-task, multimodal benchmark to assess effectiveness of PRMs . step correctness, answers aggregation and reasoning process search are evaluated . ethical principles of MPBench are based on a set of evaluation paradigms based in a text-based benchmark . |
| Outcome: | a new benchmark assesses the effectiveness of large language models (LLMs) in multiple scenarios . it uses three evaluation paradigms to assess the effectiveness and compares them with existing models . a the proposed model improves reasoning accuracy by providing stepwise feedback for multi-step reasoning results . |
CRAB: Cross-environment Agent Benchmark for Multimodal Language Model Agents (2025.findings-acl)
Copied to clipboard
Tianqi Xu, Linyao Chen, Dai-Jie Wu, Yanjun Chen, Zecheng Zhang, Xiang Yao, Zhiqiang Xie, Yongchao Chen, Shilong Liu, Bochen Qian, Anjie Yang, Zhaoxuan Jin, Jianbo Deng, Philip Torr, Bernard Ghanem, Guohao Li
| Challenge: | Existing benchmarks for MLM agents in interactive environments are limited by their focus on a single environment, lack of detailed and generalized evaluation methods, and the complexity of constructing tasks and evaluators. |
| Approach: | They propose a cross-environment agent benchmark framework that integrates graph-based evaluation and task generation methods. |
| Outcome: | The proposed framework supports multiple devices and can be easily extended to any environment with a Python interface. |
Towards A “Novel” Benchmark: Evaluating Literary Fiction with Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent advances in Large Language Models (LLMs) context windows have enabled them to process inputs over 100K tokens and generate outputs of up to 10K token. |
| Approach: | They propose a multi-level evaluation framework that incorporates ten metrics across the Macro, Meso, and Micro levels and an annotated fiction dataset. |
| Outcome: | The proposed framework incorporates ten metrics across the Macro, Meso, and Micro levels and is based on a human-human-AI dataset. |
A Reinforcement Learning Framework for Cross-Lingual Stance Detection Using Chain-of-Thought Alignment (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing approaches to cross-lingual stance detection can't effectively perform cross-linguistic transfer of complex reasoning processes. |
| Approach: | They propose a framework to facilitate cross-lingual transfer of complex reasoning processes in stance detection by using cross-linguistic Chain-of-Thought alignment to obtain high-quality CoTs generated from target language inputs. |
| Outcome: | The proposed framework outperforms competing models on four multilingual datasets. |
CARE-STaR: Constraint-aware Self-taught Reasoner (2025.findings-acl)
Copied to clipboard
Zhiliang Li, Bo Tang, Yijun Niu, Beihong Jin, Qiwen Shi, Yuchen Feng, Zhiyu Li, Jie Hu, Mingchuan Yang, Feiyu Xiong
| Challenge: | Recent research on instruction following has demonstrated that LLMs can handle complex instructions. |
| Approach: | They propose to assign constraints to different levels of constraints in instructions . they use chain-of-thought and self-taught reasoner methods to identify constraints . |
| Outcome: | The proposed method outperforms supervised fine-tuning (SFT) on three instruction-following benchmarks. |
Is It JUST Semantics? A Case Study of Discourse Particle Understanding in LLMs (2025.findings-acl)
Copied to clipboard
William Berkeley Sheffield, Kanishka Misra, Valentina Pyatkin, Ashwini Deo, Kyle Mahowald, Junyi Jessy Li
| Challenge: | Discourse particles are crucial elements that subtly shape the meaning of text. |
| Approach: | They examine the capacity of linguists to distinguish fine-grained senses of English *just* . they find that they struggle to fully capture more subtle nuances of discourse particles . |
| Outcome: | The study shows that linguists struggle to capture subtle nuances of discourse particles. |
War of Thoughts: Competition Stimulates Stronger Reasoning in Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent advances in Large Language Models (LLMs) have reshaped the landscape of reasoning tasks. |
| Approach: | They propose a method that enhances LLM reasoning without finetuning by using test-time scaling. |
| Outcome: | The proposed method outperforms baseline models in both budget and model size. |
Does Rationale Quality Matter? Enhancing Mental Disorder Detection via Selective Reasoning Distillation (2025.findings-acl)
Copied to clipboard
| Challenge: | incorporating clinical symptom information into a model enhances domain expertise, improving its detection and interpretation performance. large language models are effective for generating explanatory rationales, but inconsistencies in relevance and domain alignment of LLM-generated rationale are challenging. |
| Approach: | They propose a framework that fine-tunes smaller language models with rationales exhibiting high domain relevance . they propose 'quality-focused' approach that selects rationale based on their alignment with clinical reasoning . |
| Outcome: | The proposed framework improves mental health detection and interpretation performance by ensuring high-quality rationales with domain relevance. |
Rethinking Table Instruction Tuning (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing studies have overlooked the impact of hyperparameters on table understanding abilities . authors show that smaller learning rates and fewer training instances can enhance table understanding while preserving general capabilities. |
| Approach: | They propose a hyperparameter-based instruction-tuned model for table-related tasks that improves out-of-domain table understanding ability and general capabilities. |
| Outcome: | The proposed model outperforms existing models on table-related tasks while maintaining strong out-of-domain generalization and general capabilities. |
CliniDial: A Naturally Occurring Multimodal Dialogue Dataset for Team Reflection in Action During Clinical Operation (2025.findings-acl)
Copied to clipboard
| Challenge: | Prior studies have shown that sufficient collaboration is the key factor that determines the outcome of an operation. |
| Approach: | They propose to model the communication between team members during an operation using audio data and physiology signals from two camera angles. |
| Outcome: | The proposed model is based on existing frameworks and invites future effort on developing methods that can deal with real-world clinical data. |
Chumor 2.0: Towards Better Benchmarking Chinese Humor Understanding from (Ruo Zhi Ba) (2025.findings-acl)
Copied to clipboard
Ruiqi He, Yushu He, Longju Bai, Jiarui Liu, Zhenjie Sun, Zenghao Tang, He Wang, Hanchen Xia, Rada Mihalcea, Naihao Deng
| Challenge: | Existing studies on humor in non-English languages lack culturally nuanced humor in other languages. |
| Approach: | They construct a Chinese humor explanation dataset using a reddit-like platform . they test ten LLMs and find they are significantly better than existing LLM models . |
| Outcome: | The proposed dataset is the first and largest Chinese humor explanation dataset. |
Explicit Bayesian Inference to Uncover the Latent Themes of Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) have impressive generative capabilities, yet their inner mechanisms remain largely opaque. |
| Approach: | They propose a variational autoencoder-based neural topic model to interpret LLMs generation process through an explicit Bayesian framework by inferring latent topic variables via variational inference. |
| Outcome: | The proposed model outperforms state-of-the-art topic models on intrinsic measures of coherence and diversity on multiple datasets and shows significant gains on classification and summarization tasks. |
Improving Occupational ISCO Classification of Multilingual Swiss Job Postings with LLM-Refined Training Data (2025.findings-acl)
Copied to clipboard
| Challenge: | 80% of job postings are German, 11% French, 8% English, and under 1% Italian. |
| Approach: | They propose a method that refines silver-standard ISCO labels by consolidating them with predictions from pre-fine-tuned models to resolve discrepancies. |
| Outcome: | The proposed method raises Top-1 accuracy on silver data to 58.3% and reaches 80% precision on held-out data. |
Brevity is the soul of sustainability: Characterizing LLM response lengths (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) consume significant energy and carbon emissions due to their inference processes. |
| Approach: | They first benchmark 12 decoder-only LLMs across 5 datasets and then analyze LLM responses to determine their quality. |
| Outcome: | The proposed methods can reduce the length of responses while preserving the quality of the LLMs. |
Adversarial Preference Learning for Robust LLM Alignment (2025.findings-acl)
Copied to clipboard
Yuanfu Wang, Pengyu Wang, Chenyang Xi, Bo Tang, Junyi Zhu, Wenqiang Wei, Chen Chen, Chao Yang, Jingfeng Zhang, Chaochao Lu, Yijun Niu, Keming Mao, Zhiyu Li, Feiyu Xiong, Jie Hu, Mingchuan Yang
| Challenge: | Modern language models rely on Reinforcement Learning from Human Feedback (RLHF) to encourage safe behaviors, but they remain vulnerable to adversarial attacks due to three key limitations: (1) the inefficiency and high cost of human annotation; (2) the vast diversity of potential adversarials; and (3) the risk of feedback bias and reward hacking. |
| Approach: | They propose an iterative adversarial training method that incorporates three key innovations to address these challenges. |
| Outcome: | Experiments on Mistral-7B-Instruct-v0.3 show that the proposed method significantly enhances robustness and reduces harmful outputs from 5.88% to 0.43%. |
gMBA: Expression Semantic Guided Mixed Boolean-Arithmetic Deobfuscation Using Transformer Architectures (2025.findings-acl)
Copied to clipboard
| Challenge: | Mixed Boolean-Arithmetic (MBA) obfuscation protects intellectual property by converting programs into complex forms that are difficult to analyze. |
| Approach: | They propose a mixed-boolean-arithmetic (MBA) obfuscation framework that transforms a Transformer-based neural encoder-decoder into a truth table that is an automatically constructed semantic representation of an expression's behavior. |
| Outcome: | The proposed framework improves performance and highlights the importance of internal semantic expressions in recovering obfuscated code to its original form. |
READoc: A Unified Benchmark for Realistic Document Structured Extraction (2025.findings-acl)
Copied to clipboard
Zichao Li, Aizier Abulaiti, Yaojie Lu, Xuanang Chen, Jia Zheng, Hongyu Lin, Xianpei Han, Shanshan Jiang, Bin Dong, Le Sun
| Challenge: | Document Structured Extraction (DSE) is a field of document structure analysis that aims to extract structured content from raw documents. |
| Approach: | They propose a benchmark to evaluate document structured extraction systems by converting unstructured PDFs into semantically rich Markdown. |
| Outcome: | The proposed benchmark is based on 3,576 diverse and real-world documents from arXiv, GitHub, and Zenodo. |
TicTac: Time-aware Supervised Fine-tuning for Automatic Text Dating (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing models that ignore the temporal relatedness of documents are time-agnostic and therefore fail to perform in automatic text dating. |
| Approach: | They propose a supervised fine-tuning model for automatic text dating that captures temporal semantic information and uses a contrastive learning-based approach to model two types of temporal relations of diachronic documents. |
| Outcome: | The proposed model outperforms state-of-the-art models on two diachronic corpora and captures temporal semantic information. |
Dolphin: Document Image Parsing via Heterogeneous Anchor Prompting (2025.findings-acl)
Copied to clipboard
Hao Feng, Shu Wei, Xiang Fei, Wei Shi, Yingdong Han, Lei Liao, Jinghui Lu, Binghong Wu, Qi Liu, Chunhui Lin, Jingqun Tang, Hao Liu, Can Huang
| Challenge: | Current document image parsing solutions rely on specialized models or generate content autoregressively. |
| Approach: | They propose a multimodal document image parsing model that integrates specialized models with autogeneous content generation. |
| Outcome: | The proposed model achieves state-of-the-art performance across diverse page-level and element-level settings while ensuring superior efficiency. |
FanChuan: A Multilingual and Graph-Structured Benchmark For Parody Detection and Analysis (2025.findings-acl)
Copied to clipboard
Yilun Zheng, Sha Li, Fangkun Wu, Yang Ziyi, Lin Hongchao, Zhichao Hu, Cai Xinjun, Ziming Wang, Jinxuan Chen, Sitao Luan, Jiahao Xu, Lihui Chen
| Challenge: | Parody is an emerging phenomenon on social media, where individuals imitate a role or position opposite to their own . limited available data and deficient diversity in current datasets hinder study of parody . |
| Approach: | They build a dataset of parody users and annotated comments from both English and Chinese corpora to test parody detection and comment sentiment analysis. |
| Outcome: | The proposed datasets provide richer contextual information, which is lacking in existing datasets. |
P-CoT: A Pedagogically-motivated Participatory Chain-of-Thought Prompting for Phonological Reasoning in LLMs (2025.findings-acl)
Copied to clipboard
| Challenge: | Using the PhonologyBench benchmark, we assess tasks like rhyme word generation, g2p conversion, and syllable counting. |
| Approach: | They evaluate phonological reasoning in text-based large language models using the PhonologyBench benchmark and a Pedagogically-motivated Participatory Chain-of-Thought prompt. |
| Outcome: | The proposed model achieves up to 52% improvement and surpasses human baselines in certain tasks. |
DynaCode: A Dynamic Complexity-Aware Code Benchmark for Evaluating Large Language Models in Code Generation (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing code benchmarks for large language models remain static, resulting in data contamination and unreliable evaluation results. |
| Approach: | They propose a dynamic, complexity-aware benchmark that overcomes the limitations of static datasets and provides a memorization-advantaged benchmark. |
| Outcome: | DynaCode generates 189 million unique nested code problems across 4 units of code complexity and 16 types of call graphs. |
Small Encoders Can Rival Large Decoders in Detecting Groundedness (2025.findings-acl)
Copied to clipboard
Istabrak Abbes, Gabriele Prato, Quentin Fournier, Fernando Rodriguez, Alaa Boukhary, Adam Elwood, Sarath Chandar
| Challenge: | Large language models struggle to answer queries reliably when the provided context lacks information, often resorting to ungrounded speculation or internal knowledge. |
| Approach: | They propose to detect whether a given query is grounded in a document provided in context before LLMs generate answers. |
| Outcome: | The proposed model can generate answers that are grounded in the document provided in context while reducing inference latency by orders of magnitude. |
KITAB-Bench: A Comprehensive Multi-Domain Benchmark for Arabic OCR and Document Understanding (2025.findings-acl)
Copied to clipboard
Ahmed Heakl, Muhammad Abdullah Sohail, Mukul Ranjan, Rania Elbadry, Ghazi Shazan Ahmad, Mohamed El-Geish, Omar Maher, Zhiqiang Shen, Fahad Shahbaz Khan, Salman Khan
| Challenge: | Optical Character Recognition (OCR) is a key component of document processing . Arabic text recognition has complex typographic and calligraphic features . |
| Approach: | They propose a comprehensive Arabic OCR benchmark that fills the gaps in evaluation systems. |
| Outcome: | The proposed benchmark outperforms existing models in Arabic by 60% in the character error rate . the best model achieves only 65% accuracy in PDF-to-Markdown conversion . |
Robustness and Confounders in the Demographic Alignment of LLMs with Human Perceptions of Offensiveness (2025.findings-acl)
Copied to clipboard
| Challenge: | despite evidence of demographic bias, reports with whom they align best are hard to generalize or contradictory . confounders introduced in the annotation process account for more variation in alignment patterns than demographic traits . |
| Approach: | They examine the alignment of large language models with human annotations in offensive language datasets. |
| Outcome: | The results show that LLMs align better with human annotations than other models. |
AL-QASIDA: Analyzing LLM Quality and Accuracy Systematically in Dialectal Arabic (2025.findings-acl)
Copied to clipboard
| Challenge: | Dialectal Arabic (DA) varieties are under-served by language technologies, particularly large language models (LLMs). |
| Approach: | They propose a framework that comprehensively assesses LLMs’ DA modeling capabilities across four dimensions: fidelity, understanding, quality, and diglossia. |
| Outcome: | The proposed framework assesses LLMs’ DA modeling capabilities across four dimensions: fidelity, understanding, quality, and diglossia. |
Is Large Language Model Performance on Reasoning Tasks Impacted by Different Ways Questions Are Asked? (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing studies on Large Language Models (LLMs) have not investigated the impact of question types on LLM performance. |
| Approach: | They evaluate the performance of five Large Language Models on reasoning tasks . they use quantitative reasoning tasks and deductive reasoning tasks to evaluate the models . |
| Outcome: | The results show that Reasoning accuracy does not correlate with final selection accuracy. |
MutantPrompt: Prompt Optimization via Mutation Under a Budget on Modest-sized LMs (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have revolutionized the way we learn and process information, but identifying optimal prompts remains a challenge for low-resource languages. |
| Approach: | They propose a framework that leverages multi-armed bandit algorithms to efficiently identify optimal prompts tailored to low-resource languages. |
| Outcome: | The proposed framework is able to find optimal prompts for low-resource languages and significantly improves performance across multiple low-level tasks. |
Heuristic-based Search Algorithm in Automatic Instruction-focused Prompt Optimization: A Survey (2025.findings-acl)
Copied to clipboard
Wendi Cui, Jiaxin Zhang, Zhuohang Li, Hao Sun, Damien Lopez, Kamalika Das, Bradley A. Malin, Sricharan Kumar
| Challenge: | Recent advances in Large Language Models (LLMs) have led to remarkable achievements across a variety of NLP tasks. |
| Approach: | They propose a taxonomy of automatic prompt optimization methods that explore and improve prompts with minimal human oversight. |
| Outcome: | The proposed methods can explore and improve prompts with minimal human oversight. |
CONSENSAGENT: Towards Efficient and Effective Consensus in Multi-Agent LLM Interactions Through Sycophancy Mitigation (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent advances in multi-agent large language model systems have shown remarkable performance in tasks such as reasoning, planning, and decision-making. |
| Approach: | They propose a framework that dynamically refines prompts based on agent interactions to mitigate sycophancy by requiring additional debate rounds to reach consensus. |
| Outcome: | The proposed framework outperforms both single-agent and multi-a agent baselines and achieves state-of-the-art results across all benchmark datasets. |
The Structural Safety Generalization Problem (2025.findings-acl)
Copied to clipboard
Julius Broomfield, Tom Gibbs, George Ingebretsen, Ethan Kosak-Hine, Tia Nasir, Jason Zhang, Reihaneh Iranmanesh, Sara Pieri, Reihaneh Rabbany, Kellin Pelrine
| Challenge: | LLM jailbreaks are a widespread safety challenge. |
| Approach: | They propose a structure-rewriting guardrail that allows for more efficient safety assessment . single-turn attacks are the most extensively explored in the literature . |
| Outcome: | The proposed framework can be used to enable new defenses, the authors show . they show that the proposed framework reduces the risk of harmful inputs . |
DPO Kernels: A Semantically-Aware, Kernel-Enhanced, and Divergence-Rich Paradigm for Direct Preference Optimization (2025.findings-acl)
Copied to clipboard
Amitava Das, Suranjana Trivedy, Danush Khanna, Yaswanth Narsupalli, Basab Ghosh, Rajarshi Roy, Gurpreet Singh, Vinija Jain, Vasu Sharma, Aishwarya Naresh Reganti, Aman Chadha
| Challenge: | Direct Preference Optimization (DPO) is a cornerstone for preference alignment but is constrained by fixed divergence measures and limited feature transformations. |
| Approach: | They propose a new enhancement of Direct Preference Optimization that integrates kernel methods to overcome these challenges. |
| Outcome: | The proposed model improves divergence measures and features by using kernels . the proposed model achieves state-of-the-art generalization in factuality, safety, reasoning, and instruction following . |
Model-Dependent Moderation: Inconsistencies in Hate Speech Detection Across LLM-based Systems (2025.findings-acl)
Copied to clipboard
| Challenge: | Content moderation systems powered by large language models are increasingly deployed to detect hate speech . if two systems produce different outcomes for the same content, it undermines consistency and predictability . |
| Approach: | They analyze 1.3+ million sentences from a factorial design to determine hate speech classification . they find identical content receives markedly different classification values across systems . |
| Outcome: | The proposed model finds that identical content receives markedly different classification values across systems. |
Label-semantics Aware Generative Approach for Domain-Agnostic Multilabel Classification (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing approaches to multi-label text classification are limited by textual data. |
| Approach: | They propose a domain-agnostic generative model framework for multi-label text classification that generates predefined label descriptions and matches them to predefined labels. |
| Outcome: | The proposed model achieves 13.94% and 24.85% performance over all datasets. |
Unsupervised Morphological Tree Tokenizer (2025.findings-acl)
Copied to clipboard
| Challenge: | Conventional statistical tokenizers often disrupt constituent boundaries within words, thereby corrupting semantic information. |
| Approach: | They propose a method that uses morphological structure guidance to induce character-level structures of words by training a deep model. |
| Outcome: | Empirical results show that the proposed method retains complete morphemes and outperforms existing methods on morphological segmentation and language modeling tasks. |
CausalLink: An Interactive Evaluation Framework for Causal Reasoning (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing evaluation frameworks for causal reasoning are unclear . we propose a framework that disentangles reasoning processes from confounding factors . |
| Approach: | They propose a framework that assesses the causal reasoning skill to identify correct interventions in conversational language models. |
| Outcome: | The proposed evaluation framework isolates causal capabilities from confounding effects of world knowledge and semantic cues. |
Toward Global AI Inclusivity: A Large-Scale Multilingual Terminology Dataset (GIST) (2025.findings-acl)
Copied to clipboard
Jiarui Liu, Iman Ouzzani, Wenkai Li, Lechen Zhang, Tianyue Ou, Houda Bouamor, Zhijing Jin, Mona T. Diab
| Challenge: | Despite advances in machine translation, domain-specific terminology translation remains challenging. |
| Approach: | They propose a large-scale multilingual AI terminology dataset that combines LLMs for extraction with human expertise for translation. |
| Outcome: | The proposed framework combines human translation expertise with LLMs to improve translation accuracy and improve BLEU and COMET scores. |
A Joint Optimization Framework for Enhancing Efficiency of Tool Utilization in LLM Agents (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing efforts for tool utilization involve an LLM agent that contains instructions on using the description of the available tools to determine and call the tools required to solve the problem. |
| Approach: | They propose to optimize the context of LLM agents by combining the instructions provided in agent prompts and tool descriptions to enhance their interaction. |
| Outcome: | The proposed framework improves both the instructions provided in agent prompt and tool description, enhancing their interaction. |
When Claims Evolve: Evaluating and Enhancing the Robustness of Embedding Models Against Misinformation Edits (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing claims-matching systems that use sentence embedding models are not robust to edits as users interact with claims online. |
| Approach: | They propose a perturbation framework that generates valid and natural claim variations and evaluate different mitigation approaches to improve their findings. |
| Outcome: | The proposed framework evaluates embedding models in a multi-stage retrieval pipeline and identifies the effectiveness of mitigation approaches. |
Splintering Nonconcatenative Languages for Better Tokenization (2025.findings-acl)
Copied to clipboard
| Challenge: | Subword tokenizers are not effective in nonconcatenative languages such as Hebrew and Arabic, where morphology is encoded in root-template patterns, or Malay and Georgian, where split affixes are common. |
| Approach: | They propose a pre-processing step which rearranges text into a linear form that better represents nonconcatenative morphologies. |
| Outcome: | The proposed method is based on measures evaluating token vocabularies in Hebrew, Arabic, and Malay, and on BERT-architecture models trained for Hebrew. |
Aria-UI: Visual Grounding for GUI Instructions (2025.findings-acl)
Copied to clipboard
| Challenge: | Using a multimodal model, GUI agents can ground from language instructions to target elements . relying on HTML or AXTree inputs is a challenge for GUI agents . |
| Approach: | They propose a large multimodal model specifically designed for GUI grounding that adopts a pure vision approach instead of auxiliary inputs. |
| Outcome: | The proposed model outperforms vision-only and AXTree-reliant models on offline and online agents. |
Revealing Hidden Mechanisms of Cross-Country Content Moderation with Natural Language Processing (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing knowledge on how and why NLP methods make content moderation decisions is limited . authors examine how and when to use LLMs in content modeation . |
| Approach: | They use Shapley values and LLM-guided explanations to reverse-engineer content moderation decisions across countries. |
| Outcome: | The proposed methods show that they reverse-engineer content moderation decisions across countries and over time. |
Unilogit: Robust Machine Unlearning for LLMs Using Uniform-Target Self-Distillation (2025.findings-acl)
Copied to clipboard
| Challenge: | Extensive experiments on public benchmarks and an in-house e-commerce dataset demonstrate Unilogit’s superior performance in balancing forget and retain objectives, outperforming state-of-the-art methods such as NPO and UnDIAL. |
| Approach: | They propose a self-distillation method that dynamically adjusts target logits to achieve a uniform probability for the target token. |
| Outcome: | Extensive experiments on public benchmarks and an in-house e-commerce dataset demonstrate Unilogit’s superior performance in balancing forget and retain objectives. |
Creating a Lens of Chinese Culture: A Multimodal Dataset for Chinese Pun Rebus Art Understanding (2025.findings-acl)
Copied to clipboard
Tuo Zhang, Tiantian Feng, Yibin Ni, Mengqin Cao, Ruying Liu, Kiana Avestimehr, Katharine Butler, Yanjun Weng, Mi Zhang, Shrikanth Narayanan, Salman Avestimehr
| Challenge: | a new study examines the performance of large vision-language models in understanding art . the Pun Rebus Art Dataset is a multimodal dataset for art understanding rooted in traditional Chinese culture . |
| Approach: | They propose a multimodal dataset for art understanding deeply rooted in traditional Chinese culture . they aim to facilitate the development of VLMs that can better understand culturally specific content . |
| Outcome: | The proposed dataset shows that state-of-the-art VLMs struggle with these tasks . the data will facilitate the development of VLM models that can better understand culturally specific content . |
FastDraft: How to Train Your Draft (2025.findings-acl)
Copied to clipboard
| Challenge: | Speculative Decoding relies on the availability of efficient draft models, which are often lacking due to a stringent constraint of vocabulary compatibility. |
| Approach: | They propose a novel approach for pre-training and aligning a draft model to any large language model by incorporating efficient pre-train and fine-tuning over synthetic datasets generated by the target model. |
| Outcome: | The proposed model can be trained on a single server with 8 Intel Gaudi 2 accelerators in under 24 hours and achieves 3x acceptance rate, block efficiency and 2x memory bound speedup. |
SignMusketeers: An Efficient Multi-Stream Approach for Sign Language Translation at Scale (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing work on sign language video processing focuses on the face, hands and body posture of the signer. |
| Approach: | They propose to learn the handshapes and rich facial expressions of sign languages in a self-supervised fashion by learning from individual frames rather than video sequences. |
| Outcome: | The proposed model is more efficient than previous work on sign language pre-training. |
GUI Agents: A Survey (2025.findings-acl)
Copied to clipboard
Dang Nguyen, Jian Chen, Yu Wang, Gang Wu, Namyong Park, Zhengmian Hu, Hanjia Lyu, Junda Wu, Ryan Aponte, Yu Xia, Xintong Li, Jing Shi, Hongjie Chen, Viet Dac Lai, Zhouhang Xie, Sungchul Kim, Ruiyi Zhang, Tong Yu, Mehrab Tanjim, Nesreen K. Ahmed, Puneet Mathur, Seunghyun Yoon, Lina Yao, Branislav Kveton, Jihyung Kil, Thien Huu Nguyen, Trung Bui, Tianyi Zhou, Ryan A. Rossi, Franck Dernoncourt
| Challenge: | Large Foundation Models (LFMs) have transformed the landscape of AI research and day-to-day life. |
| Approach: | They propose a framework that delineates GUI agents' perception, reasoning, planning, and acting capabilities. |
| Outcome: | The proposed framework delineates their perception, reasoning, planning, and acting capabilities. |
MEDEC: A Benchmark for Medical Error Detection and Correction in Clinical Notes (2025.findings-acl)
Copied to clipboard
| Challenge: | Several studies have shown that large language models can answer medical questions correctly, outperforming the average human score in some medical exams. |
| Approach: | They introduce MEDEC, the first publicly available benchmark for medical error detection and correction in clinical notes. |
| Outcome: | The proposed model outperforms medical doctors in errors detection and correction tasks. |
Understanding the Influence of Synthetic Data for Text Embedders (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent advances in general purpose text embedders have been driven by training on synthetic training data. |
| Approach: | They propose to use GPT-4 to produce high quality synthetic data that expands existing training datasets for embeddings to new tasks. |
| Outcome: | The proposed dataset is high quality and leads to consistent improvements in performance. |
Dynamic Knowledge Integration for Evidence-Driven Counter-Argument Generation with Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Argumentation in natural language processing (NLP) is becoming an indispensable tool in many application domains such as public policy, law, medicine, and education. |
| Approach: | They propose a reconstructed dataset of argument and counter-argument pairs . they propose integrating dynamic external knowledge from the web to improve counter-arguments . |
| Outcome: | The proposed method shows stronger correlation with human judgments compared to reference-based metrics. |
Tell, Don’t Show: Leveraging Language Models’ Abstractive Retellings to Model Literary Themes (2025.findings-acl)
Copied to clipboard
| Challenge: | Literature challenges traditional bag-of-words approaches for topic modeling because narrative language focuses on immersive sensory details instead of abstractive description or exposition. |
| Approach: | They propose a topic modeling approach that prompts generative language models to *tell* what passages *show*, thereby translating narratives’ surface forms into higher-level concepts and themes. |
| Outcome: | The proposed model can translate narratives’ surface forms into higher-level concepts and themes than by running LDA alone or directly asking LMs to list topics. |
BottleHumor: Self-Informed Humor Explanation using the Information Bottleneck Principle (2025.findings-acl)
Copied to clipboard
| Challenge: | Humor is an effective communication tool that can manifest in various forms, including puns, exaggerated facial expressions, absurd behaviors, and incongruities. |
| Approach: | They propose a method that elicits relevant world knowledge from vision and language models and refines it to generate an explanation of the humor in an unsupervised manner. |
| Outcome: | The proposed method can be adapted for additional tasks that can benefit from eliciting and conditioning on relevant world knowledge. |
Financial Language Model Evaluation (FLaME) (2025.findings-acl)
Copied to clipboard
| Challenge: | Language Models (LMs) have demonstrated impressive capabilities with core NLP tasks in finance, but their effectiveness is difficult to assess due to gaps in evaluation methodologies. |
| Approach: | They propose to use a framework to evaluate language models against ‘reasoning-reinforced’ LMs to measure their performance on finance NLP tasks. |
| Outcome: | The proposed frameworks are open-source and provide data and data for the study. |
CausalRAG: Integrating Causal Graphs into Retrieval-Augmented Generation (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing RAG frameworks face critical limitations due to text chunking and semantic similarity. |
| Approach: | They propose a framework that incorporates causal graphs into the retrieval process. |
| Outcome: | The proposed framework preserves contextual continuity and improves retrieval precision, leading to more accurate and interpretable responses. |
Towards Safety Reasoning in LLMs: AI-agentic Deliberation for Policy-embedded CoT Data Creation (2025.findings-acl)
Copied to clipboard
Tharindu Kumarage, Ninareh Mehrabi, Anil Ramakrishna, Xinyan Zhao, Richard Zemel, Kai-Wei Chang, Aram Galstyan, Rahul Gupta, Charith Peris
| Challenge: | Safety reasoning paradigms require high-quality policy-embedded chain-of-thought datasets . generating such data through human annotations is prohibitively expensive . |
| Approach: | They propose AIDSAFE: Agentic Iterative Deliberation for Safety Reasoning . AIDS AFE leverages multi-agent deliberation to iteratively expand reasoning on safety policies . |
| Outcome: | The proposed model improves policy adherence and reasoning quality while maintaining acceptable utility and over-refusal accuracy. |
Explain then Rank: Scale Calibration of Neural Rankers Using Natural Language Explanations from LLMs (2025.findings-acl)
Copied to clipboard
| Challenge: | Neural ranking models produce the final document scores, but they are often treated as transient information and only the relative orderings are preserved to produce a ranking. |
| Approach: | They propose to exploit large language models (LLMs) to provide relevance and uncertainty signals for these neural text rankers to produce scale-calibrated scores through Monte Carlo sampling of natural language explanations (NLEs). |
| Outcome: | The proposed approach outperforms previous calibration methods and LLM-based methods for ranking, calibration, and query performance prediction tasks. |
Beyond instruction-conditioning, MoTE: Mixture of Task Experts for Multi-task Embedding Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Instruction-conditioning has become the dominant approach for embedding specialization, but its direct application to low-capacity models imposes representational constraints that limit the performance gains derived from specialization. |
| Approach: | They propose a mixture of task experts transformer block which leverages task-specialized parameters trained with Task-Aware Contrastive Learning to enhance the model’s ability to generate specialized embeddings. |
| Outcome: | The proposed model achieves 64% higher performance gains in retrieval datasets (+3.27 +5.21) and 43% higher performance gain across all datasets (+1.81 2.60). |
Metagent-P: A Neuro-Symbolic Planning Agent with Metacognition for Open Worlds (2025.findings-acl)
Copied to clipboard
YanfangZhou YanfangZhou, Yuntao Liu, Xiaodong Li, Yongqiang Zhao, Xintong Wang, Jinlong Tian, Zhenyu Li, Xinhai Xu
| Challenge: | Recent advances in large language models (LLMs) show promising potential through their world knowledge and language processing capabilities in open-world planning. |
| Approach: | They propose a framework that integrates the world knowledge of large language models, symbolic reasoning capabilities of cognitive architectures, and metacognition to improve experience utilization. |
| Outcome: | The proposed framework outperforms current state-of-the-art methods in Minecraft and reduces the average replanning counts by 34% and exceeds the human success rate by 18.96%. |
Q-STRUM Debate: Query-Driven Contrastive Summarization for Recommendation Comparison (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing contrastive summarization methods such as STRUM-LLM fail to clarify differences between items . emergence of large language models (LLMs) has revolutionized QCS capabilities . |
| Approach: | They propose a new method that generates focused and contrastive summaries by using debate-style prompting. |
| Outcome: | Experiments show that Q-STRUM Debate performs better than existing methods on key contrastive summarization criteria. |
Inductive Linguistic Reasoning with Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Evaluating large language models (LLMs) on their linguistic reasoning capabilities is an important task to understand the gaps in their skills that may surface during large-scale adoption. |
| Approach: | They propose to generate analogical exemplars with a language model and apply them in-context with target language exemplar. |
| Outcome: | The proposed method can be applied to other tasks present in Linguistics Olympiad competitions and achieves state-of-the-art results across nearly all problem types and difficulty levels in the LINGOLY dataset. |
Evaluating LLMs’ Mathematical and Coding Competency through Ontology-guided Interventions (2025.findings-acl)
Copied to clipboard
| Challenge: | Current large language models have shown impressive performance on logical reasoning benchmarks . however, the true depth of their competencies and robustness in reasoning tasks remains an open question . |
| Approach: | They propose a general ontology of perturbations and a semi-automatic method to apply perturbations to arithmetic reasoning and code generation datasets to test their LLMs' capabilities. |
| Outcome: | The proposed model outperforms existing models on arithmetic reasoning and code generation tasks. |
Exploiting Phonetics and Glyph Representation at Radical-level for Classical Chinese Understanding (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing approaches in classical Chinese understanding have integrated heterogeneous linguistic knowledge, spanning lexicalsemantic relationships. |
| Approach: | They propose a radical-level phonetics and glyph representation enhanced Chinese model with powerful fine-grained semantic modeling capabilities. |
| Outcome: | The proposed model establishes robust representations through rules-based radical decomposition and bype pair encoder (BPE) based radical aggregated for structural pattern recognition, phonetic-glyph semantic mapping, and dynamic semantic fusion. |
Tokens for Learning, Tokens for Unlearning: Mitigating Membership Inference Attacks in Large Language Models via Dual-Purpose Training (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing defenses for large language models do not account for the sequential nature of text data. |
| Approach: | They propose a lightweight yet effective empirical privacy defense that leverages token-specific characteristics to protect training data of large language models. |
| Outcome: | The proposed approach provides strong protection against membership inference attacks and improves language modeling performance by 10% across different LLM architectures and datasets compared to baselines. |
Verify with Caution: The Pitfalls of Relying on Imperfect Factuality Metrics (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent advances in large language models have led to optimism that they can serve as reliable evaluators of natural language outputs. |
| Approach: | They propose to use factuality metrics to evaluate natural language outputs . they find they misestimate the factual accuracy of NLG systems . |
| Outcome: | The proposed metrics are inconsistent with each other and often misestimate the factual accuracy of NLG systems, causing biases against paraphrased outputs and outputs that draw upon faraway parts of the source documents. |
TabXEval: Why this is a Bad Table? An eXhaustive Rubric for Table Evaluation (2025.findings-acl)
Copied to clipboard
| Challenge: | a rubric-based evaluation framework for tables often overlooks subtle structural and content-level discrepancies. |
| Approach: | They propose a rubric-based evaluation framework that integrates structural descriptors with contextual signals to enable more precise and consistent table comparison. |
| Outcome: | The proposed framework is based on a multi-domain benchmark with human annotations and perturbations. |
LADDER: Language-Driven Slice Discovery and Error Rectification in Vision Classifiers (2025.findings-acl)
Copied to clipboard
Shantanu Ghosh, Rayan Syed, Chenyu Wang, Vaibhav Choudhary, Binxu Li, Clare B Poynton, Shyam Visweswaran, Kayhan Batmanghelich
| Challenge: | Current slice discovery methods in computer vision rely on converting input images into sets of attributes and testing hypotheses about configurations of pre-computed attributes associated with elevated error patterns. |
| Approach: | They propose a method to identify systematic biases in the mistakes of pre-trained vision models by converting input images into sets of attributes and testing hypotheses about configurations of these attributes. |
| Outcome: | The proposed method outperforms existing methods on 3 natural and 3 medical imaging datasets and generates pseudo-labels for each identified bias. |
GSQ-Tuning: Group-Shared Exponents Integer in Fully Quantized Training for LLMs On-Device Fine-tuning (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) fine-tuning techniques require large Floating Point(FP) computation and are impractical for resource-constrained edge devices. |
| Approach: | They propose a framework for on-device LLM fine-tuning that eliminates the need for floating-point operations in both inference and training. |
| Outcome: | The proposed framework reduces memory and compute costs while reducing memory usage. |
Evaluation of LLMs in Medical Text Summarization: The Role of Vocabulary Adaptation in High OOV Settings (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have been successful in medical text summarization . however, they do not perform fine-grained evaluations under difficult settings . |
| Approach: | They show that large language models show a significant performance drop for data points with high concentration of out-of-vocabulary words or with high novelty. |
| Outcome: | The proposed model shows a significant performance drop for data points with high concentration of out-of-vocabulary words or with high novelty. |
UniT: One Document, Many Revisions, Too Many Edit Intention Taxonomies (2025.findings-acl)
Copied to clipboard
| Challenge: | Current research on edit intentions lacks a comprehensive edit intention taxonomy (EIT) that spans multiple application domains. |
| Approach: | They propose a Unified edit intention taxonomy that integrates existing edit intentions and integrates them into a comprehensive edit intention Taxonomic. |
| Outcome: | The proposed taxonomy achieves higher inter-annotator agreement scores compared to existing taxonomies and is applicable to a large set of application domains. |
Predicting Depression in Screening Interviews from Interactive Multi-Theme Collaboration (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for depression detection do not capture intra-theme and inter-themes correlation and do not allow clinicians to focus on themes of interest. |
| Approach: | They propose an interactive depression detection framework that leverages in-context learning techniques to identify themes in clinical interviews and then models both intra-theme and inter-themes correlation. |
| Outcome: | The proposed framework achieves 12% on Recall and 35% on F1-dep. metrics compared to the previous state-of-the-art model on the depression detection dataset DAIC-WOZ. |
Your Language Model May Think Too Rigidly: Achieving Reasoning Consistency with Symmetry-Enhanced Training (2025.findings-acl)
Copied to clipboard
Yihang Yao, Zhepeng Cen, Miao Li, William Han, Yuyou Zhang, Emerson Liu, Zuxin Liu, Chuang Gan, Ding Zhao
| Challenge: | Large Language Models (LLMs) have demonstrated strong reasoning capabilities across various tasks. |
| Approach: | They propose a data-centric approach that enhances LLMs’ awareness of symmetry in query variations and propose syMmetry-ENhanceD (MEND) data augmentation. |
| Outcome: | Extensive experiments on logical and arithmetic reasoning tasks show that the proposed approach improves model robustness at the knowledge extraction stage through query augmentation. |
TritonBench: Benchmarking Large Language Model Capabilities for Generating Triton Operators (2025.findings-acl)
Copied to clipboard
Jianling Li, ShangZhan Li, Zhenye Gao, Qi Shi, Yuxuan Li, Zefan Wang, Jiacheng Huang, WangHaojie WangHaojie, Jianrong Wang, Xu Han, Zhiyuan Liu, Maosong Sun
| Challenge: | Triton is a high-level Python-like programming language for building efficient GPU kernels. |
| Approach: | They propose a TritonBench benchmark that provides a comprehensive evaluation of Tritonic operators on widely deployed GPUs. |
| Outcome: | The proposed benchmarks show that current LLMs struggle to generate efficient Triton operators on widely deployed GPUs aligned with industry applications. |
Just KIDDIN’ : Knowledge Infusion and Distillation for Detection of INdecent Memes (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing approaches to detect toxicity in online multimodal environments require common-sense reasoning and contextual awareness. |
| Approach: | They propose a hybrid neurosymbolic framework that unifies distillation of implicit contextual knowledge from Large Vision-Language Models and infusion of explicit relational semantics through sub-graphs from Knowledge Graphs. |
| Outcome: | The proposed framework outperforms state-of-the-art models on two datasets with improvements of 0.5%, and 10.6% in HatefulMemes Benchmark. |
Dynamic Personality in LLM Agents: A Framework for Evolutionary Modeling and Behavioral Analysis in the Prisoner’s Dilemma (2025.findings-acl)
Copied to clipboard
| Challenge: | Current models rely on static personality traits but lack natural selection processes and direct psychological metrics, failing to accurately capture authentic dynamic personality variations. |
| Approach: | They propose a framework that uses game payoffs as environmental feedback to drive adaptive personality evolution and analyze correlations between personality metrics and behavior. |
| Outcome: | The proposed framework reveals new behavioral patterns of agents and evaluates personality-behavior relationships, advancing agent-based social simulations and human-AI symbiosis research. |
Building A Proof-Oriented Programmer That Is 64% Better Than GPT-4o Under Data Scarcity (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing proof-oriented programming languages struggle due to data scarcity . existing proof-based models struggle due a lack of sufficient corpora for proof-orientated programming languages such as F* . |
| Approach: | They propose a method that synthesizes proof-oriented programming problems for proficiency in a language and incorporates diverse coding data for reasoning capability elicitation. |
| Outcome: | The proposed method outperforms existing proof-oriented models in function- and repository-level proof-based programming by 64% relative margin and improves GPT-4o's performance by 54% by repairing outputs over GPT-4) |
On the Robust Approximation of ASR Metrics (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for estimating speech recognition metrics depend on ground truth labels. |
| Approach: | They propose a label-free approach to approximating ASR performance metrics . they embed multimodal embeddings in a unified space for speech and transcription representations . |
| Outcome: | The proposed method outperforms baseline models on speech recognition benchmarks by 50%. |
Are the Values of LLMs Structurally Aligned with Humans? A Causal Perspective (2025.findings-acl)
Copied to clipboard
Yipeng Kang, Junqi Wang, Yexin Li, Mengmeng Wang, Wenming Tu, Quansen Wang, Hengli Li, Tingjun Wu, Xue Feng, Fangwei Zhong, Zilong Zheng
| Challenge: | Current approaches to value alignment focus on a few core values, such as helpfulness, harmlessness, and honesty. |
| Approach: | They propose to use latent causal value graphs to guide two lightweight value-steering methods . role-based prompting and sparse autoencoder (SAE) steering are also used . |
| Outcome: | Experiments on Gemma-2B-IT and Llama3-8B- IT show that the proposed methods are effective and controllable. |
LLMs Can Also Do Well! Breaking Barriers in Semantic Role Labeling via Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Semantic role labeling (SRL) is a crucial task of natural language processing (NLP). |
| Approach: | They propose to equip LLMs with retrieval-augmented generation and self-correction mechanisms to enable SRL to perform better in Chinese and English. |
| Outcome: | The proposed method achieves state-of-the-art in Chinese and English on three widely-used benchmarks. |
Lost in Transcription, Found in Distribution Shift: Demystifying Hallucination in Speech Foundation Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Automatic speech recognition systems have seen remarkable improvements in recent years, but evaluation of performance remains dependent on word and character error rate (WER/CER). |
| Approach: | They investigate how distribution shifts, model size and model architecture influence hallucination error rate (HER) HER is a metric used to quantify hallucinosity in automatic speech recognition systems. |
| Outcome: | The proposed model can be used to measure hallucination errors in high-stakes domains such as healthcare, legal, and aviation. |
M2PA: A Multi-Memory Planning Agent for Open Worlds Inspired by Cognitive Theory (2025.findings-acl)
Copied to clipboard
YanfangZhou YanfangZhou, Xiaodong Li, Yuntao Liu, Yongqiang Zhao, Xintong Wang, Zhenyu Li, Jinlong Tian, Xinhai Xu
| Challenge: | Open-world planning poses a challenge due to complex environments and task diversity . recent work shows that large language models (LLMs) lack the ability to connect to agents' experiences . |
| Approach: | They propose an open-world multi-memory planning agent that combines large language models with human-like multi-mesh systems to leverage their strengths. |
| Outcome: | The proposed agent outperforms state-of-the-art agents on 50 Minecraft tasks in zero-shot learning. |
AnnaAgent: Dynamic Evolution Agent System with Multi-Session Memory for Realistic Seeker Simulation (2025.findings-acl)
Copied to clipboard
Ming Wang, Peidong Wang, Lin Wu, Xiaocui Yang, Daling Wang, Shi Feng, Yuxin Chen, Bixuan Wang, Yifei Zhang
| Challenge: | Existing models of seeker simulations are limited by the cost and ethical concerns of involving real seekers in mental health research. |
| Approach: | They propose an emotional and cognitive dynamic agent system equipped with tertiary memory to enable dynamic control of the simulator's configurations. |
| Outcome: | The proposed system achieves more realistic seeker simulation compared to baselines. |
Diversification Catalyzes Language Models’ Instruction Generalization To Unseen Semantics (2025.findings-acl)
Copied to clipboard
| Challenge: | Instruction-tuned language models excel in knowledge, reasoning, and instruction-following . however, the factors enabling generalization to unseen instructions remain underexplored . |
| Approach: | They propose to model instruction-following as a computational process and design controlled experiments inspired by the Turing-complete Markov algorithm to disentangle its dynamics. |
| Outcome: | The proposed model outperforms scaling up data volumes in generalist models by combining in-domain and diverse out-of-domain tasks. |
DecompileBench: A Comprehensive Benchmark for Evaluating Decompilers in Real-World Scenarios (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing approaches focus on syntactic correctness through synthetic micro-benchmarks or subjective human ratings, despite semantic fidelity and usability. |
| Approach: | They propose a framework that enables effective evaluation of decompilers in reverse engineering workflows . they compare six industrial-strength decompils and six recent LLM-powered approaches . |
| Outcome: | The proposed framework outperforms commercial tools in code understandability despite lower functionality correctness . it shows that it can transform human-centric reverse engineering workflows . |
Thinking Before Running! Efficient Code Generation with Thorough Exploration and Optimal Refinement (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent research indicates that large language models (LLMs) have demonstrated remark-able capabilities in various programming-related domains, such as code generation and code refinement. |
| Approach: | They propose a framework that combines exploration with refinement to reduce test-time computation overhead. |
| Outcome: | The proposed framework outperforms SOTA and AgentCoder on humanEval and MBPP benchmarks while reducing test-time computation overhead and scalability. |
Edit Once, Update Everywhere: A Simple Framework for Cross-Lingual Knowledge Synchronization in LLMs (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods to update large language models focus on single-language editing or basic multilingual editing, failing to achieve true cross-linguistic knowledge synchronization. |
| Approach: | They propose a cross-linguistic knowledge democracy edit technique to improve cross-lingual performance. |
| Outcome: | The proposed method improves cross-lingual performance while maintaining high accuracy in monolingual settings. |
SafeChain: Safety of Language Models with Long Chain-of-Thought Reasoning Capabilities (2025.findings-acl)
Copied to clipboard
Fengqing Jiang, Zhangchen Xu, Yuetai Li, Luyao Niu, Zhen Xiang, Bo Li, Bill Yuchen Lin, Radha Poovendran
| Challenge: | Emerging large reasoning models (LRMs) leverage long chain-of-thought (CoT) reasoning to enhance their reasoning capabilities. |
| Approach: | They conduct a systematic study of LRM safety using human annotations to assess their safety. |
| Outcome: | The proposed safety measures are compared to state-of-the-art models on strong and wildjailbreak datasets. |
ETRQA: A Comprehensive Benchmark for Evaluating Event Temporal Reasoning Abilities of Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Event temporal reasoning (ETR) is a significant indicator that a large language model understands the physical world. |
| Approach: | They propose a unified taxonomy for event temporal questions and construct a benchmark based on this taxonomies. |
| Outcome: | The proposed taxonomy inherits and expands existing datasets and contains multiple categories of compound questions. |
The Law of Knowledge Overshadowing: Towards Understanding, Predicting and Preventing LLM Hallucination (2025.findings-acl)
Copied to clipboard
Yuji Zhang, Sha Li, Cheng Qian, Jiateng Liu, Pengfei Yu, Chi Han, Yi R. Fung, Kathleen McKeown, ChengXiang Zhai, Manling Li, Heng Ji
| Challenge: | Hallucination is a persistent challenge in large language models where even with rigorous quality control, models often generate distorted facts. |
| Approach: | They propose a new framework to quantify factual hallucinations by modeling knowledge overshadowing. |
| Outcome: | The proposed framework improves model factuality on Overshadow (27.9%), MemoTrap (13.1%) and NQ-Swap (18.3%). |
LegoMT2: Selective Asynchronous Sharded Data Parallel Training for Massive Neural Machine Translation (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods to train a single model for massive languages have huge communication overheads and parameter interference. |
| Approach: | They propose an efficient training approach with an asymmetric multi-way model architecture for massive multilingual neural machine translation. |
| Outcome: | The proposed model is 16.2 faster than the distributed training method for M2M-100-12B while improving the translation performance by an average of 2.2 BLEU on Flores-101. |
Pruning General Large Language Models into Customized Expert Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) require significant computational resources to maintain their general capabilities. |
| Approach: | They propose a Custom Pruning method to prune a large general model into a smaller lightweight expert model, positioned along the "language", "domain" and "task" dimensions. |
| Outcome: | The proposed method outperforms existing pruning methods and achieves minimal loss in both expert and general capabilities across models from different model families and sizes. |
Enhance Multimodal Consistency and Coherence for Text-Image Plan Generation (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing studies on textual plan generation only focus on LLMs, enabling applications in robotics, virtual assistants, and instruc. |
| Approach: | They propose a framework that generates and refines text-image plans step-by-step . they collect a new benchmark consisting of 1,100 tasks and their text- image pair solutions covering 11 daily topics. |
| Outcome: | The proposed framework generates and refines text-image plans step-by-step and improves on existing models. |
Un-considering Contextual Information: Assessing LLMs’ Understanding of Indexical Elements (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) excel in coreference resolution tasks, but previous studies only assessed performance with nouns and third person pronouns. |
| Approach: | They evaluate LLMs' performance on coreference resolution with indexicals like I, you, here and tomorrow which come with unique challenges due to their linguistic properties. |
| Outcome: | The proposed models perform well with some indexicals while struggling with others. |
Behavioral Analysis of Information Salience in Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models excel at text summarization, but the exact notion of salience remains unclear. |
| Approach: | They propose a framework to derive and investigate information salience in Large Language Models (LLMs) using length-controlled summarization as a behavioral probe into the content selection process. |
| Outcome: | The proposed framework derives a proxy for how models prioritize information in large language models. |
The Behavior Gap: Evaluating Zero-shot LLM Agents in Complex Task-Oriented Dialogs (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent studies show that LLM-based agents struggle to perform in zero-shot scenarios. |
| Approach: | They propose a framework to quantify the behavior gap between AI agents and human experts . they propose to examine discrepancies in dialog acts, tool usage, and knowledge utilization . |
| Outcome: | The proposed framework measures the behavior gap between AI agents and human experts on task-oriented dialogs. |
Task Facet Learning: A Structured Approach To Prompt Optimization (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing approaches to prompt optimization are limited to learning multiple facets of a task from training examples. |
| Approach: | They propose to optimize a text prompt by considering different facets of a task and including them in the prompt. |
| Outcome: | The proposed algorithm can generate long, complex prompts that existing methods are unable to generate. |
LLM as Effective Streaming Processor: Bridging Streaming-Batch Mismatches with Group Position Encoding (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for adapting LLMs to streaming rely on expensive re-encoding or limited scalability. |
| Approach: | They propose a group position encoding paradigm built on batch architectures to enhance consistency between streaming and batch modes. |
| Outcome: | The proposed method outperforms existing methods on cross-lingual and cross-modal tasks. |
YinYang-Align: A new Benchmark for Competing Objectives and Introducing Multi-Objective Preference based Text-to-Image Alignment (2025.findings-acl)
Copied to clipboard
Amitava Das, Yaswanth Narsupalli, Gurpreet Singh, Vinija Jain, Vasu Sharma, Suranjana Trivedy, Aman Chadha, Amit Sheth
| Challenge: | Recent controversies highlight the need for robust alignment mechanisms in text-to-image systems. |
| Approach: | They propose a framework to evaluate T2I systems across six contradictory alignment objectives . objectives highlight key trade-offs such as artistic freedom and cultural sensitivity . |
| Outcome: | The proposed framework achieves superior alignment across all objectives. |
FREE: Fast and Robust Vision Language Models with Early Exits (2025.findings-acl)
Copied to clipboard
| Challenge: | Vision-Language Models (VLMs) have shown remarkable performance improvements in Vision-language tasks, but their large size poses challenges for real-world applications. |
| Approach: | They propose an adversarial approach to train exit classifiers in Vision-Language Models that uses a transformer layer and a classifier to perform input-adaptive inference. |
| Outcome: | The proposed approach speeds up inference speed with minimal drop in performance by 1.51 while retaining comparable performance. |
REPRO-Bench: Can Agentic AI Systems Assess the Reproducibility of Social Science Research? (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing benchmarks for reproducing social science papers focus on reproducing results using provided code and data without assessing their consistency with the paper. |
| Approach: | They propose a benchmark to evaluate agentic AI systems' ability to automate reproducibility assessment. |
| Outcome: | The proposed benchmark oversimplifies real-world scenarios and lacks diversity in data formats and programming languages. |
Time Travel: A Comprehensive Benchmark to Evaluate LMMs on Historical and Cultural Artifacts (2025.findings-acl)
Copied to clipboard
Sara Ghaboura, Ketan Pravin More, Ritesh Thawkar, Wafa Al Ghallabi, Omkar Thawakar, Fahad Shahbaz Khan, Hisham Cholakkal, Salman Khan, Rao Muhammad Anwer
| Challenge: | TimeTravel is a benchmark of 10,250 expert-verified historical artifact samples spanning 266 distinct cultures across 10 major historical regions. |
| Approach: | They evaluate contemporary AI models on TimeTravel, highlighting their strengths and identifying areas for improvement. |
| Outcome: | The timeTravel benchmark covers 266 cultures and 10 major historical regions and aims to establish AI as reliable partner in preserving cultural heritage. |
Unveiling and Addressing Pseudo Forgetting in Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing efforts to mitigate catastrophic forgetting in continual learning have not been studied. |
| Approach: | They propose a rationale-guided replay framework that allows models to leverage their capabilities and provide partial external correct rationales to the original instructions. |
| Outcome: | The proposed framework mitigates pseudo forgetting while maintaining model plasticity. |
Improving MLLM’s Document Image Machine Translation via Synchronously Self-reviewing Its OCR Proficiency (2025.findings-acl)
Copied to clipboard
| Challenge: | Multimodal Large Language Models (MLLMs) have shown strong performance in document image tasks, especially Optical Character Recognition (OCR). However, they struggle with Document Image Machine Translation (DIMT), which requires handling both cross-modal and cross-lingual challenges. |
| Approach: | They propose a novel fine-tuning paradigm that allows the model to generate OCR text before producing translation text, which allows it to leverage its strong monolingual OCR ability while learning to translate text across languages. |
| Outcome: | The proposed model can leverage its strong monolingual OCR ability while learning to translate text across languages. |
HG-InsightLog: Context Prioritization and Reduction for Question Answering with Non-Natural Language Construct Log Data (2025.findings-acl)
Copied to clipboard
| Challenge: | Log files are crucial for monitoring, diagnostics, and root cause analysis in IT systems . their sheer volume makes manual analysis overwhelming and traditional methods are ineffective . |
| Approach: | They propose a framework that constructs a multi-entity temporal hypergraph using log attribute-value pairs as nodes and connects them with hyperedges. |
| Outcome: | The proposed framework is model-agnostic and training-free and scales with open-source LLMs. |
Dialect Normalization using Large Language Models and Morphological Rules (2025.findings-acl)
Copied to clipboard
| Challenge: | Natural language understanding systems struggle with low-resource languages, including many dialects of high-resourced ones. |
| Approach: | They propose a method that combines rule-based linguistically informed transformations and large language models with targeted few-shot prompting without any parallel data. |
| Outcome: | The proposed method is able to transform dialectal text into a standard variety while maintaining as much of the original meaning as possible. |
USDC: A Dataset of ̲User ̲Stance and ̲Dogmatism in Long ̲Conversations (2025.findings-acl)
Copied to clipboard
| Challenge: | Previously, studies on stance and dogmatism in user conversations have focused on training models using annotated datasets at the post level, treating each post as independent and randomly sampling posts from conversation threads. |
| Approach: | They build a dataset for studying user opinion fluctuations in 764 long multi-user Reddit conversation threads, called USDC. |
| Outcome: | The proposed dataset analyzes user opinion fluctuations in 764 long multi-user Reddit conversation threads. |
Learning to Insert [PAUSE] Tokens for Better Reasoning (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing studies have explored incorporating special-purpose tokens into the training process to enhance reasoning capabilities. |
| Approach: | They propose a method for inserting dummy tokens consecutively just before reasoning steps to increase model effectiveness. |
| Outcome: | The proposed method outperforms fine-tuning and previous token insertion methods on multiple datasets and models. |
Understand the Implication: Learning to Think for Pragmatic Understanding (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods rely on annotated labels but overlook the reasoning process humans naturally use to interpret implicit meaning. |
| Approach: | They propose a dataset that includes explicit reasoning for both correct and incorrect interpretations and propose supervised fine-tuning to improve their performance. |
| Outcome: | The proposed dataset improves LLMs' pragmatic understanding by 11.12% across model families and 16.10% over label trained models. |
WASA: WAtermark-based Source Attribution for Large Language Model-Generated Data (2025.findings-acl)
Copied to clipboard
Xinyang Lu, Jingtan Wang, Zitong Zhao, Zhongxiang Dai, Chuan-Sheng Foo, See-Kiong Ng, Bryan Kian Hsiang Low
| Challenge: | Large language models (LLMs) have impressive performance but intellectual property concerns are looming . a framework that can be used to perform source attribution for LLMs can be developed. |
| Approach: | They propose a framework that enables an LLM to generate synthetic texts with embedded watermarks that contain information about their source. |
| Outcome: | The proposed framework achieves source attribution accuracy and robustness against adversaries. |
Dense Retrieval with Quantity Comparison Intent (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing sparse and dense retrieval systems fragment numerals and units that express quantities in arbitrary ways. |
| Approach: | They propose a dense retrieval system built around a density multi-vector index . they propose eliciting and exploiting quantities and associated comparison intents . |
| Outcome: | The proposed system is faster and more accurate than popular PLMs on two public and one proprietary e-commerce benchmarks. |
Reflection on Knowledge Graph for Large Language Models Reasoning (2025.findings-acl)
Copied to clipboard
Yigeng Zhou, Wu Li, Yifan Lu, Jing Li, Fangming Liu, Meishan Zhang, Yequan Wang, Daojing He, Honghai Liu, Min Zhang
| Challenge: | Existing methods for supplementing Large Language Models (LLMs) with knowledge graphs often introduce noise in the retrieval and reasoning pipeline, hindering their ability to integrate external knowledge for complex multi-hop question answering. |
| Approach: | They propose a framework to enhance LLMs' reasoning capabilities through reflective engagement with knowledge graphs by Query Decoupling, LLM-Driven Knowledge Graph Exploration, and Inference with Knowledge Reconstruction. |
| Outcome: | The proposed framework integrates external knowledge into LLMs and trains them to leverage this knowledge for answering questions. |
Revisiting 3D LLM Benchmarks: Are We Really Testing 3D Capabilities? (2025.findings-acl)
Copied to clipboard
| Challenge: | Current 3D LLMs are evaluated on Q&A or captioning tasks rather than specific downstream tasks like object detection. |
| Approach: | They propose principles for better assessing genuine 3D understanding by explicitly separating 3D abilities from 1D or 2D aspects when evaluating 3D LLMs. |
| Outcome: | The proposed methods are based on the “2D-Cheating” problem in 3D LLM evaluation, suggesting that they are ineffective . |
DIESEL: A Lightweight Inference-Time Safety Enhancement for Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models generate outputs that are not aligned with human values, such as toxic content, malicious use cases, and vulnerabilities to adversarial jailbreak attacks. |
| Approach: | They propose a lightweight inference-guidance technique that can be seamlessly integrated into any autoregressive LLM to semantically filter undesirable content during generation. |
| Outcome: | The proposed technique can be integrated into any autoregressive LLM to semantically filter undesirable content during generation. |
Toward Structured Knowledge Reasoning: Contrastive Retrieval-Augmented Generation on Experience (2025.findings-acl)
Copied to clipboard
Jiawei Gu, Ziting Xian, Yuanzhen Xie, Ye Liu, Enjie Liu, Ruichao Zhong, Mochi Gao, Yunzhi Tan, Bo Hu, Zang Li
| Challenge: | Large language models struggle to infer implicit relationships embedded in tabular formats . authors introduce a framework that builds experience memory representations and enhances generalization through contrastive In-Context Learning (ICL). |
| Approach: | They propose a framework that builds experience memory representations and enhances generalization through contrastive In-Context Learning to simulate human-like knowledge transfer. |
| Outcome: | Experiments on Text-to-SQL and TableQA show CoRE significantly improves performance . it achieves gains of 3.44% and 4.24%, with up to 17.2% on challenging tasks . |
Structured Pruning for Diverse Best-of-N Reasoning Optimization (2025.findings-acl)
Copied to clipboard
| Challenge: | Extensive experiments on the MATH dataset demonstrate that our method significantly outperforms traditional best-of-N and random head selection strategies. |
| Approach: | They propose a contrastive learning framework that dynamically selects the optimal head and layer to prune during inference by aligning question embeddings with head embedds. |
| Outcome: | The proposed approach outperforms best-of-N and random head selection strategies on the MATH500 and GSM8K datasets. |
PodAgent: A Comprehensive Framework for Podcast Generation (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing automatic audio generation methods struggle to generate podcast-like audio programs effectively. |
| Approach: | They propose a framework for creating podcast-like audio programs that generates informative topic-discussion content by designing a multi-agent collaboration system, builds a voice pool and uses LLM-enhanced speech synthesis to generate expressive conversational speech. |
| Outcome: | The proposed framework surpasses direct GPT-4 generation in topic-discussion dialogue content, and produces more expressive conversational speech. |
STORM-BORN: A Challenging Mathematical Derivations Dataset Curated via a Human-in-the-Loop Multi-Agent Framework (2025.findings-acl)
Copied to clipboard
Wenhao Liu, Zhenyi Lu, Xinyu Hu, Jerry Zhang, Dailin Li, Jiacheng Cen, Huilin Cao, Haiteng Wang, Yuhan Li, Xie Kun, Dandan Li, Pei Zhang, Chengbo Zhang, Yuxiang Ren, Xiaohong Huang, Yan Ma
| Challenge: | Existing datasets suffer from outdated and insufficient challenging content, neglecting human-like reasoning, and limited reliability due to single-LLM generation. |
| Approach: | They propose a human-in-the-loop, multi-agent data generation framework that integrates reasoning-dense filters, multiagent collaboration, and human mathematicians’ evaluations to ensure the reliability and quality of the dataset. |
| Outcome: | The proposed framework improves accuracy and quality of the 2,000-synthesized datasets by integrating reasoning-dense filters, multi-agent collaboration, and human mathematicians’ evaluations. |
iMOVE : Instance-Motion-Aware Video Understanding (2025.findings-acl)
Copied to clipboard
Jiaze Li, Yaya Shi, Zongyang Ma, Haoran Xu, Yandong.bai Yandong.bai, Huihui Xiao, Ruiwen Kang, Fan Yang, Tingting Gao, Di Zhang
| Challenge: | Recent advances in Video Large Language Models have led to rapid development, significantly enhancing the capture of overall video semantics and achieving remarkable performance in general video understanding tasks. |
| Approach: | They propose a large-scale instance-motion-aware video instruction-tuning dataset iMOVE that utilizes Event-awful Spatiotemporal Efficient Modeling to retain informative instance spatiotemporal motion details while maintaining computational efficiency. |
| Outcome: | The proposed model excels in video temporal understanding and general video understanding. |
SceneGram: Conceptualizing and Describing Tangrams in Scene Context (2025.findings-acl)
Copied to clipboard
| Challenge: | Current systems show mixed results in reproducing human variation in object naming . figurative descriptions for abstract stimuli remain a major challenge in vision and language research . |
| Approach: | They propose to analyze human references to tangrams placed in different scene contexts . they analyze the richness and variability of conceptualizations found in human references . |
| Outcome: | The proposed model does not account for the richness and variability of human references. |
Relevant or Random: Can LLMs Truly Perform Analogical Reasoning? (2025.findings-acl)
Copied to clipboard
| Challenge: | Analogical reasoning is a unique ability of humans to address unfamiliar challenges by transferring strategies from relevant past experiences. |
| Approach: | They propose to use self-generated random examples to improve performance on a variety of reasoning tasks by incorporating relevant examples from relevant past experiences. |
| Outcome: | The proposed methods achieve comparable or even better performance on GSM8K with random biological examples. |
MERIT: Multi-Agent Collaboration for Unsupervised Time Series Representation Learning (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing approaches to time series representation learning are time-consuming and expert-dependent, which are difficult to generalize across different tasks. |
| Approach: | They propose to use large language model agent to guide unsupervised time series representation learning and a framework to integrate three LLM agents to collaboratively generate positive views for time series data. |
| Outcome: | The proposed framework integrates large language model (LLM) agent to guide unsupervised time series representation learning and compares it with state-of-the-art baselines on multiple time series datasets. |
JsonTuning: Towards Generalizable, Robust, and Controllable Instruction Tuning (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing text-to-text methods struggle with issues such as generalization, robustness, and controllability due to their lack of explicit task structures. |
| Approach: | They propose a structure-to-structure approach that uses JSON structures to represent tasks. |
| Outcome: | The proposed method outperforms TextTuning in terms of performance, robustness, and controllability across different scenarios. |
RedundancyLens: Revealing and Exploiting Visual Token Processing Redundancy for Efficient Decoder-Only MLLMs (2025.findings-acl)
Copied to clipboard
| Challenge: | Current decoder-only architectures achieve higher performance but lower efficiency . cross-attention-based architectures skip visual token computations . |
| Approach: | They propose a training-free framework for analyzing trained MLLMs to investigate redundancy . they propose 'probe-activated Dynamic FFN and Hollow Attention' algorithms for visual token reductions and a layer ranking algorithm for inference acceleration. |
| Outcome: | The proposed framework achieves comparable performance to or better than state-of-the-art methods while remaining compatible with them. |
Memory-augmented Query Reconstruction for LLM-based Knowledge Graph Reasoning (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods that confuse tool utilization with knowledge reasoning harm readability and give rise to tool invocation hallucinations. |
| Approach: | They propose to decouple LLM from tool invocation tasks by establishing a memory module with explicit descriptions of query statements and a query memory module to facilitate the KGQA process. |
| Outcome: | The proposed method achieves state-of-the-art on WebQSP and CWQ benchmarks. |
KaFT: Knowledge-aware Fine-tuning for Boosting LLMs’ Domain-specific Question-Answering Performance (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent literature reveals that supervised fine-tuning (SFT) is suboptimal for domain-specific question-answering tasks. |
| Approach: | They propose a query diversification strategy for robust conflict detection and a knowledge-aware fine-tuning approach to effectively boost LLMs’ performance. |
| Outcome: | The proposed approach improves the model generalization and alleviates the hallucination. |
Are Multimodal Large Language Models Pragmatically Competent Listeners in Simple Reference Resolution Tasks? (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing models are unable to resolve references to abstract visual stimuli, such as color patches and color grids, but their pragmatic capabilities are still a challenge for state-of-the-art MLLMs. |
| Approach: | They investigate whether multimodal large language models are able to resolve references to abstract visual stimuli, such as color patches and color grids, in a well-known reference resolution paradigm. |
| Outcome: | The proposed model can resolve references to abstract visual stimuli in dyadic reference games. |
Removing Prompt-template Bias in Reinforcement Learning from Human Feedback (2025.findings-acl)
Copied to clipboard
| Challenge: | Reinforcement Learning from Human Feedback (RLHF) has shown promise for enhancing pre-trained large language models to generate responses that align with human preferences and societal values. |
| Approach: | They propose a method to estimate prompt-template bias term during reward modeling and use it to calibrate reward scores. |
| Outcome: | The proposed method can be flexibly combined with existing algorithms of removing length bias, leading to a further improvement in the aspect of enhancing the quality of generated responses. |
Latent Distribution Decouple for Uncertain-Aware Multimodal Multi-label Emotion Recognition (2025.findings-acl)
Copied to clipboard
Jingwang Huang, Jiang Zhong, Qin Lei, Gaojinpeng Gaojinpeng, Ymyang Ymyang, Sirui Wang, PeiguangLi PeiguangLi, Kaiwen Wei
| Challenge: | Existing studies focus on improving fusion strategies and modeling modality-to-label dependencies, but they overlook the impact of aleatoric uncertainty, which is inherent noise in multimodal data. |
| Approach: | They propose a latent emotional distribution decomposition with uncertainty perception framework to model aleatoric uncertainty in multimodal data. |
| Outcome: | The proposed framework achieves state-of-the-art performance on the CMU-MOSEI and M3ED datasets, highlighting the importance of uncertainty modeling in MMER. |
Are LLMs Rational Investors? A Study on the Financial Bias in LLMs (2025.findings-acl)
Copied to clipboard
Yuhang Zhou, Yuchen Ni, Zhiheng Xi, Zhangyue Yin, Yu He, Gan Yunhui, Xiang Liu, Zhang Jian, Sen Liu, Xipeng Qiu, Yixin Cao, Guangnan Ye, Hongfeng Chai
| Challenge: | Existing studies on biases within specific domains, such as finance, remain limited. |
| Approach: | They propose a framework to detect, detect, analyze and mitigate financial biases in large language models. |
| Outcome: | The proposed framework reduces bias by 68% for the most biased model, according to key metrics. |
Seeing What Tastes Good: Revisiting Multimodal Distributional Semantics in the Billion Parameter Era (2025.findings-acl)
Copied to clipboard
| Challenge: | danoneata, et al., 2021): human learning and conceptual representation is grounded in sensorimotor experience. |
| Approach: | They evaluate image encoders and language-only models to learn which attributes are salient to the models. |
| Outcome: | The proposed models outperform language-only models on attributes predicting extended denser McRae norms and newer Binder datasets. |
Communication-Efficient and Tensorized Federated Fine-Tuning of Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) excel in translation and summarization due to the capabilities of transformer architectures. |
| Approach: | They propose to integrate tensorized adapters into model encoder/decoder blocks to improve model adaptability against data heterogeneity. |
| Outcome: | Experiments on large-scale cross-device FL and large-silo FL show that the proposed methods perform on par or even better than existing federated PEFT approaches while reducing communication cost. |
A rebuttal of two common deflationary stances against LLM cognition (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) are arguably the most predictive models of human cognition available. |
| Approach: | They argue that these deflationary claims need further justification . they argue that large language models are "just" simplistic entities . |
| Outcome: | The proposed models lack critical capacities, but they are not "just" models, the authors argue . they argue that the arguments need to be weighed against the evidence . |
COVER: Context-Driven Over-Refusal Verification in LLMs (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have become increasingly prevalent in the field of Natural Language Processing (NLP), achieving unprecedented performance across linguistic tasks. |
| Approach: | They propose a framework to quantify and analyze context-driven over-refusal . they find that over-fusals depend on the task, system prompts, model family, and the number of retrieved documents. |
| Outcome: | The proposed framework quantifyes and analyzes the concept of context-driven over-refusal on two public corpora. |
MOSAIC: Multiple Observers Spotting AI Content (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have made it easier for all to produce harmful, toxic, faked or forged content. |
| Approach: | They propose to use large language models to automatically discriminate from human-written texts by comparing their probability distributions over a document to see if they can detect forged or harmful content. |
| Outcome: | The proposed approach harnesses each model’s capabilities, leading to strong detection performance on a variety of domains. |
GUIDEX: Guided Synthetic Data Generation for Zero-Shot Information Extraction (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing domain-specific IE systems require expert schema design, data annotation, and model training. |
| Approach: | They propose a method that automatically defines domain-specific schemas and infers guidelines and generates synthetically labeled instances. |
| Outcome: | The proposed method improves on seven zeroshot Named Entity Recognition benchmarks. |
Missing the Margins: A Systematic Literature Review on the Demographic Representativeness of LLMs (2025.findings-acl)
Copied to clipboard
| Challenge: | 211 studies on the demographic representativeness of large language models have conflicting results . 29% of the studies report positive conclusions on the representativeness, 30% do not evaluate LLMs across multiple demographic categories or within demographic subcategories. |
| Approach: | 211 papers review the representativeness of large language models . authors recommend more precise evaluation methods and comprehensive documentation of demographic attributes . |
| Outcome: | 211 studies on the representativeness of large language models are reviewed . 29% of the studies report positive conclusions, but 30% fail to specify subcategories . authors recommend more precise evaluation methods and documentation of demographic attributes . |
LlamaV-o1: Rethinking Step-by-step Visual Reasoning in LLMs (2025.findings-acl)
Copied to clipboard
Omkar Thawakar, Dinura Dissanayake, Ketan Pravin More, Ritesh Thawkar, Ahmed Heakl, Noor Ahsan, Yuhao Li, Ilmuz Zaman Mohammed Zumri, Jean Lahoud, Rao Muhammad Anwer, Hisham Cholakkal, Ivan Laptev, Mubarak Shah, Fahad Shahbaz Khan, Salman Khan
| Challenge: | Existing approaches do not emphasize step-wise problem-solving. |
| Approach: | They propose a visual reasoning chain benchmark and a fine-grained reasoning metric that evaluates correctness and logical coherence at each step. |
| Outcome: | The proposed framework outperforms existing models in six benchmarks and is 5x faster during inference scaling. |
Burn After Reading: Do Multimodal Large Language Models Truly Capture Order of Events in Image Sequences? (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing benchmarks focus on single image settings, but some focus on multi-image settings. |
| Approach: | They introduce the TempVS benchmark which focuses on temporal grounding and reasoning capabilities of Multimodal Large Language Models in image sequences. |
| Outcome: | The proposed model performs poorly compared to human models in vision and language tasks. |
Full-Step-DPO: Self-Supervised Preference Optimization with Step-wise Rewards for Mathematical Reasoning (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing approaches to improve long-chain mathematical reasoning focus on the first erroneous step, but ignore all other steps and rely heavily on external signals. |
| Approach: | They propose a DPO framework that leverages step-wise rewards from the entire reasoning chain instead of optimizing only the first erroneous step. |
| Outcome: | The proposed framework improves on in-domain and out-of-domain mathematical reasoning benchmarks. |
Do Emotions Really Affect Argument Convincingness? A Dynamic Approach with LLM-based Manipulation Checks (2025.findings-acl)
Copied to clipboard
| Challenge: | Emotions have been shown to play a role in argument convincingness, yet this aspect is underexplored in the natural language processing community. |
| Approach: | They propose a framework that examines the extent to which perceived emotional intensity influences perceived convincingness. |
| Outcome: | The proposed framework examines whether emotions influence persuasiveness in humans . it finds that emotions enhance rather than weaken convincingness in human judgments . |
SCOPE: Compress Mathematical Reasoning Steps for Efficient Automated Process Annotation (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing process annotation approaches are computationally expensive. |
| Approach: | They propose a compression-based approach that transforms reasoning steps into code and normalizes them through Abstract Syntax Tree. |
| Outcome: | The proposed method outperforms existing methods on Best-of-N strategy and ProcessBench. |
Compositional Syntactico-SemBanking for English as a Second or Foreign Language (2025.findings-acl)
Copied to clipboard
| Challenge: | Despite the widespread use of English as a Second or Foreign Language (ESFL), developing syntactico-semantic representations for it is limited. |
| Approach: | They propose a Synchronous Hyperedge Replacement Grammar-based constructivist approach to address the challenges in ESFL. |
| Outcome: | The proposed approach bridges the gap between literal cues and intended meaning by using constructions as fundamental units. |
Semantics-aware prompting for translating NOtices To AirMen (2025.findings-acl)
Copied to clipboard
| Challenge: | A NOTAM or NOtice To AirMen is a crucial notification for different stakeholders . writing and understanding these messages puts heavy cognitive load on its end users. |
| Approach: | They propose a framework to translate NOTAMs into English natural language using LLMs . they use context-aware semantic prompting techniques paired with domain-specific rules . |
| Outcome: | The proposed approach improves translation accuracy and clarity using semantic prompting techniques . the proposed framework was evaluated using LLMs of varying sizes and 8 evaluation metrics . |
Stereotype or Personalization? User Identity Biases Chatbot Recommendations (2025.findings-acl)
Copied to clipboard
| Challenge: | We show that when people use large language models to generate recommendations, the LLMs produce responses that reflect both what the user wants and who the user is. |
| Approach: | They propose that chatbots should transparently indicate when user’s revealed identity influences model recommendations but fail to do so . |
| Outcome: | The proposed model generates racially stereotypical recommendations regardless of whether the user revealed their identity intentionally or unintentionally through implicit cues. |
Automated main concept generation for narrative discourse assessment in aphasia (2025.findings-acl)
Copied to clipboard
| Challenge: | Several advances have been made towards developing theoretical and computational methods for understanding narratives. |
| Approach: | They propose a method that generates MCs from novel stories that experts can edit manually. |
| Outcome: | The proposed method can generate most of the gold standard MCs for stories from an existing narrative summarization dataset. |
Can VLMs Actually See and Read? A Survey on Modality Collapse in Vision-Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Vision-language models integrate textual and visual information, enabling them to process visual inputs and generate predictions. |
| Approach: | They review work on modality collapse analysis to provide insights into the reason for this unintended behavior and review probing studies for fine-grained vision-language understanding. |
| Outcome: | The proposed models can achieve competitive performance in vision-language tasks despite relying heavily on textual information and ignoring visual information. |
“You are Beautiful, Body Image Stereotypes are Ugly!” BIStereo: A Benchmark to Measure Body Image Stereotypes in Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | BIStereo is a suite of language models that uncover body image stereotypes in language models. |
| Approach: | They propose a metric, TriSentBias, that captures the biased preferences of LMs towards a certain body type over others. |
| Outcome: | The proposed metric captures biased preferences of LMs towards a certain body type over others. |
Retrieval Models Aren’t Tool-Savvy: Benchmarking Tool Retrieval for Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) suffer from inherent inabilities to interact with the physical world and access vast, up-to-date knowledge. |
| Approach: | They propose a tool retrieval benchmark for large language models (LLMs) that includes 7.6k diverse retrieval tasks and a corpus of 43k tools. |
| Outcome: | The proposed model performs poorly on the heterogeneous tool retrieval benchmark, resulting in low pass rate and low retrieval quality. |
FineCite: A Novel Approach For Fine-Grained Citation Context Analysis (2025.findings-acl)
Copied to clipboard
| Challenge: | citation context analysis (CCA) studies the role and purpose of citations in scientific discourse. |
| Approach: | They construct a first comprehensive context definition based on semantic properties of citing text . they use fine-grained semantic properties to evaluate the definition . |
| Outcome: | The proposed definition shows improvements of up to 25% over state-of-the-art methods. |
Decoupling Reasoning and Knowledge Injection for In-Context Knowledge Editing (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing knowledge editing approaches directly edit model context without isolating target knowledge from the reasoning path of model inference, resulting in unreliable and low-quality outputs, especially in multi-hop tasks. |
| Approach: | They propose a framework that separates model reasoning from knowledge editing and propose 'DecKER' that allows users to modify specific factual associations without retraining the entire model. |
| Outcome: | The proposed framework significantly improves multi-hop reasoning performance by mitigating knowledge conflicts and preserving reasoning integrity. |
Entrospect: Information-Theoretic Self-Reflection Elicits Better Response Refinement of Small Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing approaches to self-reflection fail to deliver robust response refinement for models with parameter sizes of 10 billion or smaller. |
| Approach: | They propose to redesign Self-Refine and introduce an information-theoretic framework based on Chain-of-Thought prompt engineering to improve self-reflection in Small Language Models. |
| Outcome: | The proposed framework improves reasoning accuracy and computational efficiency by up to 36.2% under identical model and data settings. |
Iterative Repair with Weak Verifiers for Few-shot Transfer in KBQA with Unanswerability (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing models for KBQA with unanswerable questions are inadequate for real-world applications. |
| Approach: | They propose a task of few-shot transfer for KBQA with unanswerable questions that extends FuSIC-KBQA to include feedback for unanswered questions. |
| Outcome: | The proposed model outperforms suitable adaptations of multiple LLM-based and supervised SoTA models on the task while establishing a new performance for answerable few-shot transfer as well. |
Safeguarding RAG Pipelines with GMTP: A Gradient-based Masked Token Probability Method for Poisoned Document Detection (2025.findings-acl)
Copied to clipboard
| Challenge: | Retrieval-Augmented Generation (RAG) provides external knowledge for accurate and up-to-date responses, but external knowledge is vulnerable to poisoning and unauthorized injections. |
| Approach: | They propose a Gradient-based Masked Token Probability defense method to detect and filter out adversarially crafted documents by examining gradients of the retriever’s similarity function. |
| Outcome: | Experiments show that the proposed method eliminates over 90% of poisoned content while retaining relevant documents. |
EnSToM: Enhancing Dialogue Systems with Entropy-Scaled Steering Vectors for Topic Maintenance (2025.findings-acl)
Copied to clipboard
| Challenge: | Small large language models (sLLMs) are lightweight and efficient, but struggle to maintain topic consistency in task-oriented dialogue systems. |
| Approach: | They propose an approach to ensure topic consistency in task-oriented dialogue systems by manipulating internal activations during inference. |
| Outcome: | The proposed approach achieves significant performance gain with a relatively small data size compared to fine-tuning approaches. |
MultiTEND: A Multilingual Benchmark for Natural Language to NoSQL Query Translation (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent advances in NoSQL database support focus on English . however, the intricacy and heterogeneity of NoSqL query languages present a formidable challenge . |
| Approach: | They propose a multilingual benchmark for natural language to NoSQL query generation that covers six languages. |
| Outcome: | The proposed framework improves performance in English and non-English settings, while ignoring lexical and syntactic differences. |
Tool learning via Inference-time Scaling and Cycle Verifier (2025.findings-acl)
Copied to clipboard
| Challenge: | In inference-time scaling, Chain-of-Thought (CoT) data is scarce or even unavailable. |
| Approach: | They propose a method which establishes an inference cycle to synthesize user queries and CoT data. |
| Outcome: | The proposed method achieves a 75.4% pass rate and a 79.6% win rate using small models in StableToolBench. |
When Benchmarks Talk: Re-Evaluating Code LLMs with Interactive Feedback (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing static benchmarks that measure task performance often rely on a simple input-output configuration. |
| Approach: | They propose an evaluation pipeline that evaluates code models with different feedback types in an interactive setting. |
| Outcome: | The proposed evaluation pipeline compares model-user collaboration with static benchmarks by obfuscating inputs to a simulated user. |
Reranking-based Generation for Unbiased Perspective Summarization (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing evaluation frameworks rely on traditional metrics for measuring key attributes such as coverage and faithfulness without verifying their applicability. |
| Approach: | They propose to use human annotations to measure perspective summary quality and reranking-based methods yield strong results. |
| Outcome: | The proposed methods show that they perform well with synthetically generated and reranking-labeled data. |
KARPA: A Training-free Method of Adapting Knowledge Graph as References for Large Language Model’s Reasoning Path Aggregation (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for large language models (LLMs) are limited by step-by-step decision-making on KGs, or require fine-tuning or pre-training on specific KG. |
| Approach: | They propose a framework that harnesses the global planning abilities of large language models (LLMs) for efficient and accurate KG reasoning. |
| Outcome: | Extensive experiments show that the proposed framework achieves state-of-the-art performance in KGQA tasks, delivering both high efficiency and accuracy. |
Enhancing LLM-based Hatred and Toxicity Detection with Meta-Toxic Knowledge Graph (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods to address toxicity issues with large language models are inadequate . lack of domain-specific knowledge leads to false negatives and excessive sensitivity to toxic speech limits freedom of speech. |
| Approach: | They propose a method that leverages graph search on a meta-toxic knowledge graph to enhance hatred and toxicity detection. |
| Outcome: | The proposed method lowers false positive rate and improves toxicity detection performance in out-of-domain scenarios. |
Mixture-of-Personas Language Models for Population Simulation (2025.findings-acl)
Copied to clipboard
Ngoc Bui, Hieu Trung Nguyen, Shantanu Kumar, Julian Theodore, Weikang Qiu, Viet Anh Nguyen, Rex Ying
| Challenge: | Pretrained LLMs fail to capture behavioral diversity of target populations due to inherent variability across individuals and groups. |
| Approach: | They propose a probabilistic prompting method that aligns LLM responses with the target population. |
| Outcome: | Experiments show that the proposed method outperforms competing methods in alignment and diversity metrics. |
ClusComp: A Simple Paradigm for Model Compression and Efficient Finetuning (2025.findings-acl)
Copied to clipboard
| Challenge: | Weight-only quantization reduces model size but suffers from performance degradation at lower bit widths. |
| Approach: | They propose a weight-only quantization paradigm that clusters weight matrices into codebooks and finetunes them block-by-block. |
| Outcome: | The proposed paradigm outperforms quantization methods and fine tunes LLMs to 1-bit compression and fine tuning. |
Decomposed Opinion Summarization with Verified Aspect-Aware Modules (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for summarizing opinions from large-scale online reviews are not available for crowdsourcing and are difficult to crowdsource. |
| Approach: | They propose a domain-agnostic modular approach guided by review aspects to separate tasks of aspect identification, opinion consolidation, and meta-review synthesis to enable greater transparency and ease of inspection. |
| Outcome: | The proposed approach generates more grounded summaries than baseline models, as verified through automated and human evaluations. |
Token-Budget-Aware LLM Reasoning (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods to enhance reasoning capabilities of large language models incur significant overhead in token usage, leading to increased costs. |
| Approach: | They propose a token-budget-aware LLM reasoning framework that adjusts the number of reasoning tokens based on the reasoning complexity of each problem. |
| Outcome: | The proposed method reduces token costs in CoT reasoning with only a slight performance reduction. |
HATA: Trainable and Hardware-Efficient Hash-Aware Top-k Attention for Scalable Large Model Inference (2025.findings-acl)
Copied to clipboard
Ping Gong, Jiawei Yi, Shengnan Wang, Juncheng Zhang, Zewen Jin, Ouxiang Zhou, Ruibo Liu, Guanbin Xu, Youhui Bai, Bowen Ye, Kun Yuan, Tong Yang, Gong Zhang, Renhai Chen, Feng Wu, Cheng Li
| Challenge: | Existing top-k attention methods struggle to strike a balance between efficiency and accuracy. |
| Approach: | They propose a top-k attention approach that integrates low-overhead techniques into the Top-k Attention process to achieve 7.2 speedup compared to vanilla full attention. |
| Outcome: | The proposed approach achieves 7.2 speedup compared to current top-k attention methods while maintaining model accuracy. |
Answer When Needed, Forget When Not: Language Models Pretend to Forget via In-Context Knowledge Unlearning (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) are increasingly required to selectively unlearn specific information. |
| Approach: | They propose a method which fine-tunes pre-trained LLMs to enable prompt unlearning of target knowledge within the context while preserving unrelated information. |
| Outcome: | The proposed method achieves up to 95% forget accuracy while retaining 80% of unrelated knowledge, significantly outperforming baselines in both in-domain and out-of-domain scenarios. |
LIST: Linearly Incremental SQL Translator for Single-Hop Reasoning, Generation and Verification (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing schema linking methods are not able to handle complex SQL queries. |
| Approach: | They propose a new algorithm that transforms SQL queries into grammatically verifiable sub-queries which are arranged sequentially to reflect single-hop reasoning steps. |
| Outcome: | The proposed algorithm achieves significant performance gains on the BIRD dataset and surpasses schema linking methods at comparable or better cost. |
MAGI: Multi-Agent Guided Interview for Psychiatric Assessment (2025.findings-acl)
Copied to clipboard
Guanqun Bi, Zhuang Chen, Zhoufu Liu, Hongkai Wang, Xiyao Xiao, Yuqiang Xie, Wen Zhang, Yongkang Huang, Yuxuan Chen, Libiao Peng, Minlie Huang
| Challenge: | Existing large language models (LLMs) do not align with psychiatric diagnostic protocols. |
| Approach: | They propose a framework that transforms the Mini International Neuropsychiatric Interview into automatic computational workflows through coordinated multi-agent collaboration. |
| Outcome: | The proposed framework transforms the gold-standard Mini International Neuropsychiatric Interview (MINI) into automatic computational workflows through coordinated multi-agent collaboration. |
TituLLMs: A Family of Bangla LLMs with Comprehensive Benchmarking (2025.findings-acl)
Copied to clipboard
Shahriar Kabir Nahin, Rabindra Nath Nandi, Sagor Sarker, Quazi Sarwar Muhtaseem, Md Kowsher, Apu Chandraw Shill, Md Ibrahim, Mehadi Hasan Menon, Tareq Al Muntasir, Firoj Alam
| Challenge: | Existing benchmarking datasets for Bangla LLMs are not available for all languages. |
| Approach: | They present TituLLMs, the first large pretrained Bangla LLMs, available in 1b and 3b parameter sizes. |
| Outcome: | The proposed model outperforms existing models in Bangla, but not always in the first place. |
WikiMixQA: A Multimodal Benchmark for Question Answering over Tables and Charts (2025.findings-acl)
Copied to clipboard
Negar Foroutan, Angelika Romanou, Matin Ansaripour, Julian Martin Eisenschlos, Karl Aberer, Rémi Lebret
| Challenge: | Documents are fundamental to preserving and disseminating information, often incorporating complex layouts, tables, and charts that pose significant challenges for automatic document understanding (DU). |
| Approach: | They propose a benchmark for evaluating cross-modal reasoning over tables and charts extracted from 4,000 Wikipedia pages . they evaluate 12 vision-language models that achieve 70% accuracy when provided with direct context . |
| Outcome: | The proposed benchmark evaluates models with high accuracy over tables and charts extracted from 4,000 Wikipedia pages . proprietary models achieve 70% accuracy when provided with direct context, but open-source models perform worse when retrieval from long documents is required. |
Let’s Fuse Step by Step: A Generative Fusion Decoding Algorithm with LLMs for Robust and Instruction-Aware ASR and OCR (2025.findings-acl)
Copied to clipboard
| Challenge: | Various fusion strategies have been explored for integration of large language models into multi-modal systems. |
| Approach: | They propose a framework for deep fusion decoding that integrates large language models into cross-modal text recognition systems. |
| Outcome: | The proposed framework surpasses cascaded methods in English and Mandarin, and significantly reduces WERs by 17.7%. |
HPSS: Heuristic Prompting Strategy Search for LLM Evaluators (2025.findings-acl)
Copied to clipboard
Bosi Wen, Pei Ke, Yufei Sun, Cunxiang Wang, Xiaotao Gu, Jinfeng Zhou, Jie Tang, Hongning Wang, Minlie Huang
| Challenge: | Existing efforts to optimize text evaluation prompts neglect the combinatorial impact of multiple factors, leading to insufficient optimization of the evaluation pipeline. |
| Approach: | They propose to integrate 8 key factors for evaluation prompts and integrate them into an algorithm that searches for well-behaved prompting strategies for LLM evaluators. |
| Outcome: | The proposed method outperforms existing methods and human-designed evaluation prompts on four evaluation tasks. |
A Fully Generative Motivational Interviewing Counsellor Chatbot for Moving Smokers Towards the Decision to Quit (2025.findings-acl)
Copied to clipboard
Zafarullah Mahmood, Soliman Ali, Jiading Zhu, Mohamed Abdelwahab, Michelle Yu Collins, Sihan Chen, Yi Cheng Zhao, Jodi Wolff, Osnat C. Melamed, Nadia Minian, Marta Maslej, Carolynne Cooper, Matt Ratto, Peter Selby, Jonathan Rose
| Challenge: | Large language models (LLMs) are being used to provide automated talk therapy . however, it is crucial to know if they would be effective and adhere to known standards. |
| Approach: | They propose to use large language models to automate talk therapy with a focus on tobacco addiction. |
| Outcome: | The proposed chatbot showed adherence to MI standards in 98% of utterances, higher than human counsellors. |
LegalCore: A Dataset for Event Coreference Resolution in Legal Documents (2025.findings-acl)
Copied to clipboard
Kangda Wei, Xi Shi, Jonathan Tong, null Sai Ramana Reddy, Anandhavelu Natarajan, Rajiv Jain, Aparna Garimella, Ruihong Huang
| Challenge: | Existing research on event coreference resolution is limited to news articles . existing datasets for news articles are limited to events and coreferences . |
| Approach: | They present a dataset for the legal domain LegalCore which has been annotated with event and event coreference information. |
| Outcome: | The legal contract documents annotated in this dataset are several times longer than news articles, with an average length of around 25k tokens per document. |
Rectifying Belief Space via Unlearning to Harness LLMs’ Reasoning (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) exhibit sophisticated reasoning yet still generate incorrect answers. |
| Approach: | They propose a belief space rectification framework that suppresses spurious beliefs and enhances true ones to reduce erroneous reasoning and generalization. |
| Outcome: | The proposed framework reduces erroneous reasoning and improves generalization on three QA datasets and three LLMs. |
MemeDetoxNet: Balancing Toxicity Reduction and Context Preservation (2025.findings-acl)
Copied to clipboard
| Challenge: | Toxic memes spread harmful and offensive content and pose a significant challenge in online environments. |
| Approach: | They propose a framework to mitigate toxicity in toxic memes by leveraging a set of pre-trained models that can interpret the visual and textual components of memes. |
| Outcome: | The proposed framework reduces toxicity on publicly available meme datasets by 10-20% compared to the previous methods. |
Should I Trust You? Detecting Deception in Negotiations using Counterfactual RL (2025.findings-acl)
Copied to clipboard
Wichayaporn Wongkamjan, Yanze Wang, Feng Gu, Denis Peskoff, Jonathan K. Kummerfeld, Jonathan May, Jordan Lee Boyd-Graber
| Challenge: | Future human-AI interaction tools can build on our methods for deception detection by triggering friction to give users a chance to interrogate suspicious proposals. |
| Approach: | They propose to use CTRL-D to detect deception in a board game called Diplomacy . CTRL is a counterfactual RL that has a good recall and almost perfect precision . future tools could build on this to reevaluate trust in suspicious negotiations . |
| Outcome: | The proposed method detects human deception with a high precision when compared to a Large Language Model approach that flags many true messages as deceptive. |
Multi-matrix Factorization Attention (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing variants for Multi-Head Attention (MHA) fail to maintain strong performance under stringent Key-Value cache (KV cache) constraints. |
| Approach: | They propose to use multi-matrix factorization attention and MFA-Key-reuse attention architectures to increase model capacity under tight KV cache constraints. |
| Outcome: | The proposed architecture outperforms existing methods while reducing KV cache usage by 56% and 93.7% in large-scale experiments. |
Self-Training Elicits Concise Reasoning in Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Chain-of-thought reasoning has enabled large language models to use additional computation through intermediate tokens to solve complex tasks, but current models often generate more tokens than necessary to accomplish the task, incurring extraneous inference costs. |
| Approach: | They propose to fine-tune models with self-generated concise reasoning paths obtained by best-of-N sampling and few-shot conditioning in task-specific settings to elicit concise reasoning. |
| Outcome: | The proposed method reduces output tokens by 30% on GSM8K and MATH while maintaining average accuracy. |
Reason from Future: Reverse Thought Chain Enhances LLM Reasoning (2025.findings-acl)
Copied to clipboard
Yinlong Xu, Yanzhao Zheng, Shuoshuo Sun, Shuaihan Huang, Baohua Dong, Zhu Hangcheng, Ruohui Huang, Gang Yu, Hongxia Xu, Jian Wu
| Challenge: | Existing reasoning paradigms that focus on local optimum reasoning lack global perspective. |
| Approach: | They propose a bidirectional reasoning paradigm that generates reasoning paths by bidirectional planning and bottom-up reasoning accumulation. |
| Outcome: | The proposed reasoning paradigm outperforms conventional paradigms with higher accuracy and less searching space to solve complex tasks. |
LLMs as Planning Formalizers: A Survey for Leveraging Large Language Models to Construct Automated Planning Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models excel in various natural language tasks but struggle with long-horizon planning problems requiring structured reasoning. |
| Approach: | They propose to integrate large language models into AP and NLP planning frameworks by reviewing current research and identifying critical challenges and future directions. |
| Outcome: | The proposed frameworks are used to support reliable off-the-shelf AP planners. |
From Conversation to Automation: Leveraging LLMs for Problem-Solving Therapy Analysis (2025.findings-acl)
Copied to clipboard
| Challenge: | Problem-Solving Therapy (PST) is a structured psychological approach that helps individuals manage stress and resolve personal issues. |
| Approach: | They developed a framework for PST annotation using established PST Core Strategies and a set of novel Facilitative Strategies to analyze a corpus of real-world therapy transcripts to determine which strategies are most prevalent. |
| Outcome: | The proposed framework outperforms existing models and LLMs to identify the most prevalent strategies in a corpus of real-world therapy transcripts. |
Revisiting Self-Consistency from Dynamic Distributional Alignment Perspective on Answer Aggregation (2025.findings-acl)
Copied to clipboard
Yiwei Li, Ji Zhang, Shaoxiong Feng, Peiwen Yuan, Xinglin Wang, Jiayi Shi, Yueqi Zhang, Chuyi Tan, Boyuan Pan, Yao Hu, Kan Li
| Challenge: | Existing studies on self-consistency show that it improves reasoning abilities by aggregating diverse stochastic samples. |
| Approach: | They propose a confidence-driven mechanism that dynamically calibrates temperature to align with high probability modes. |
| Outcome: | The proposed method outperforms fixed-diversity baselines on reasoning tasks and improves both average and best-case performance. |
Don’t Say No: Jailbreaking LLM by Suppressing Refusal (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) are vulnerable to "jailbreaking" attacks where crafted prompts manipulate them into producing toxic content. |
| Approach: | They propose to improve the target loss objective by combining a cosine decay schedule method with refusal suppression to achieve higher success rates. |
| Outcome: | The proposed approach outperforms baseline attacks and achieves state-of-the-art attack success rates. |
From Perception to Reasoning: Enhancing Vision-Language Models for Mobile UI Understanding (2025.findings-acl)
Copied to clipboard
Settaluri Lakshmi Sravanthi, Ankit Mishra, Debjyoti Mondal, Subhadarshi Panda, Rituraj Singh, Pushpak Bhattacharyya
| Challenge: | Accurately grounding visual and textual elements within mobile user interfaces remains a challenge for Vision-Language Models (VLMs). |
| Approach: | They propose a mobile UI understanding model trained on a dataset specifically tailored for mobile screen understanding and grounding. |
| Outcome: | The proposed model achieves significant gains in accuracy across all perception tasks and on reasoning benchmarks. |
Lemmas Matter, But Not Like That: Predictors of Lemma-Based Generalization in Morphological Inflection (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent work suggests that lemma overlap drives model performance on morphological inflection tasks, but the impact of lemmm overlap is debated. |
| Approach: | They propose a novel algorithm to investigate predictors of accuracy on seen and unseen lemmas by combining the number of lema in train with the number in train. |
| Outcome: | The proposed algorithm shows that the number of lemmas in train has a stronger effect on accuracy on unseen than seen lemmes. |
Mosaic-IT: Cost-Free Compositional Data Synthesis for Instruction Tuning (2025.findings-acl)
Copied to clipboard
| Challenge: | Current instruction tuning relies on teacher models or human intervention to generate and refine the instructions and responses for training, which are costly, non-sustainable, and may lack diversity. |
| Approach: | They propose a human/model-free compositional data synthesis method that can create rich and diverse augmentations from existing instruction tuning data to enhance large language models. |
| Outcome: | The proposed method improves performance over benchmarks and reduces training costs by 80% compared with original instruction tuning. |
MAM: Modular Multi-Agent Framework for Multi-Modal Medical Diagnosis via Role-Specialized Collaboration (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent advances in medical Large Language Models have demonstrated powerful reasoning and diagnostic capabilities. |
| Approach: | They propose a modular multi-agent framework for multi-modal medical diagnosis . they decompose the medical diagnostic process into specialized roles . |
| Outcome: | The framework decomposes the medical diagnostic process into specialized roles . it achieves significant performance improvements ranging from 18% to 365% compared to baseline models. |
ATLAS: Agent Tuning via Learning Critical Steps (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing agent tuning approaches employ supervised finetuning on entire expert trajectories, but behavior-cloning of full traitories introduces expert bias and weakens generalization to states not covered by the expert data. |
| Approach: | They propose a method that finetunes LLMs on critical steps in expert trajectories and identifies and finetuns them on these steps with reduced costs. |
| Outcome: | The proposed method outperforms existing methods and open-source LLM agents on only 30% critical steps in extensive experiments. |
Syntactic Control of Language Models by Posterior Inference (2025.findings-acl)
Copied to clipboard
| Challenge: | Syntactic control of generated text is crucial for many domain-specific applications of language models, where structural constraints, such as formality, grammatical correctness, or adherence to a given template, can significantly affect usability and readability. |
| Approach: | They propose to use posterior inference to enforce target constituency structure by sampling from a proposal distribution with a syntactic tagger to ensure that each generated token aligns with the desired syntakic structure. |
| Outcome: | The proposed method increases the F1 score from 12.31 (GPT2-large) and 35.33 (Llama3-8B) to about 93 in both cases without compromising the language model’s fluency. |
Small Models Struggle to Learn from Strong Reasoners (2025.findings-acl)
Copied to clipboard
Yuetai Li, Xiang Yue, Zhangchen Xu, Fengqing Jiang, Luyao Niu, Bill Yuchen Lin, Bhaskar Ramasubramanian, Radha Poovendran
| Challenge: | a small learning gap exists between large and small language models . long CoT data and large model responses are not beneficial for small models - a problem that may be due to the small student model's ability to handle distribution shifts. |
| Approach: | They propose a mix distillation strategy that balances reasoning complexity by combining long and short CoT examples or reasoning from both larger and smaller models. |
| Outcome: | The proposed strategy outperforms training on large and small models on short CoT and small model CoT. |
Sparse Rewards Can Self-Train Dialogue Agents (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent advances in large language models have been driven by supervised fine-tuning and high-quality human feedback. however, acquiring meaningful human feedback has become increasingly challenging and costly. |
| Approach: | They propose a method that empowers LLM agents to enhance their performance without external feedback. |
| Outcome: | The proposed method improves tool-based interactions while preserving general model capabilities across diverse benchmarks. |
Almost AI, Almost Human: The Challenge of Detecting AI-Polished Writing (2025.findings-acl)
Copied to clipboard
| Challenge: | a growing use of large language models (LLMs) has led to concerns about AI-generated content detection. |
| Approach: | They evaluate 12 state-of-the-art AI-text detectors using a dataset refined at varying levels of AI involvement. |
| Outcome: | The proposed detectors flag even minimally polished text as AI-generated, struggle to differentiate between degrees of AI involvement, and exhibit biases against older and smaller models. |
The Reader is the Metric: How Textual Features and Reader Profiles Explain Conflicting Evaluations of AI Creative Writing (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent studies comparing AI-generated and human-authored literary texts have produced conflicting results. |
| Approach: | They hypothesize that differences in reading quality can be explained by genuine differences in how readers interpret and value literature . |
| Outcome: | The authors show that the differences in reading quality are largely explained by differences in how readers interpret and value literature, rather than by an intrinsic quality of the texts evaluated. |
Summary Factual Inconsistency Detection Based on LLMs Enhanced by Universal Information Extraction (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent studies have shown that Large language models can detect factual inconsistencies in summaries but they lack the efficiency and explainability needed to be effective. |
| Approach: | They propose to decouple LLMs’ information extraction and reasoning capabilities to address key challenges and propose a framework for UIEFID to guide fine-tuned LLM methods in extracting unified structured information from documents and summaries. |
| Outcome: | The proposed framework improves the detection accuracy and reduces redundant reasoning on the AGGREFACT benchmark. |
ELI-Why: Evaluating the Pedagogical Utility of Language Model Explanations (2025.findings-acl)
Copied to clipboard
Brihi Joshi, Keyu He, Sahana Ramnath, Sadra Sabouri, Kaitlyn Zhou, Souti Chattopadhyay, Swabha Swayamdipta, Xiang Ren
| Challenge: | Language models are widely used in education, yet their ability to tailor responses to learners with varied informational needs and knowledge backgrounds remains under-explored. |
| Approach: | They conduct two extensive human studies to assess the utility of language model-generated explanatory answers (explanations) on a benchmark of 13.4K "Why" questions. |
| Outcome: | The proposed model explanations match learners' educational backgrounds only 50% of the time, compared to 79% for lay explanations. |
Beyond Generation: Leveraging LLM Creativity to Overcome Label Bias in Classification (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods to mitigate label bias by leveraging in-domain data are often unavailable in real-world scenarios. |
| Approach: | They propose a calibration method that generates synthetic in-domain data from a few in-context demonstrations and utilizes it for calibration. |
| Outcome: | The proposed method reduces label bias by leveraging in-domain data from demonstrations. |
CogSteer: Cognition-Inspired Selective Layer Intervention for Efficiently Steering Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) achieve excellent performance through pretraining on extensive data. |
| Approach: | They propose an efficient selective layer intervention based on parameter-efficient fine-tuning methods to select the optimal steering layer to modulate LLM semantics. |
| Outcome: | The proposed approach is based on a model-agnostic framework and is safe to deploy. |
PASTEL : Polarity-Aware Sentiment Triplet Extraction with LLM-as-a-Judge (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for extracting triplets of aspect terms and opinions are inadequate due to complexity of aspect-opinion interactions and implicit nature of sentiment dependencies in natural language. |
| Approach: | They propose a pipeline that decomposes the ASTE task into structured subtasks . they employ fine-tuned LLMs to separately extract the aspect and opinion terms . |
| Outcome: | The proposed pipeline outperforms existing baselines in the ASTE subtask. |
COSMIC: Generalized Refusal Direction Identification in LLM Activations (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods depend on predefined refusal templates detectable in output tokens or manual review. |
| Approach: | They propose a framework that optimally identifies steering directions and target layers using cosine similarity, entirely independent of output text. |
| Outcome: | The proposed framework achieves comparable steering effectiveness without any prior knowledge or assumptions of a model’s refusal behavior such as the use of certain refusal tokens. |
Red Queen: Exposing Latent Multi-Turn Risks in Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing jailbreak strategies are single-turn with explicit malicious intent, failing to reflect real-world scenarios where interactions can be multi-turn and users can conceal intents. |
| Approach: | They propose a jailbreak attack that constructs a multi-turn scenario and conceals malicious intent under the guise of preventing harm. |
| Outcome: | The proposed attack achieves 87.6% attack success rate (ASR) on GPT-4o and 77.1% on Llama3-70B compared to previous jailbreak attacks . |
MDBench: A Synthetic Multi-Document Reasoning Benchmark Generated with Knowledge Guidance (2025.findings-acl)
Copied to clipboard
| Challenge: | Multi-document reasoning is an area of increasing relevance given LLM capabilities in handling longer-context inputs, but few benchmarks exist to rigorously examine model behavior in this setting. |
| Approach: | They propose a new dataset for evaluating LLMs on the task of multi-document reasoning that uses condensed structured seed knowledge to modify it through LLM-assisted edits. |
| Outcome: | The proposed method generates document sets and QA examples on a multi-document reasoning task using a synthetic generation process. |
DiaLLMs: EHR-Enhanced Clinical Conversational System for Clinical Test Recommendation and Diagnosis Prediction (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing medical LLMs focus primarily on diagnosis recommendation, limiting their clinical applicability. |
| Approach: | They propose a medical LLM that integrates heterogeneous EHR data into clinically grounded dialogues. |
| Outcome: | The proposed model outperforms baselines in clinical test recommendation and diagnosis prediction. |
Can Hallucination Correction Improve Video-Language Alignment? (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing work on hallucination correction for large vision-language models focuses on mitigating hallucisations, but a new approach is needed to improve video-language alignment. |
| Approach: | They propose a self-training framework learning to correct hallucinations in descriptions that do not align with the video content. |
| Outcome: | The proposed framework improves video-language alignment by identifying and correcting inconsistencies in descriptions that do not align with the video content. |
IMPARA-GED: Grammatical Error Detection is Boosting Reference-free Grammatical Error Quality Estimator (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing reference-free automatic grammatical error correction methods do not correlate with human evaluation. |
| Approach: | They propose a reference-free automatic grammatical error correction evaluation method with enhanced gramma-ed capabilities. |
| Outcome: | The proposed method achieves highest correlation with human evaluations on a meta-evaluation dataset. |
Do Language Models Mirror Human Confidence? Exploring Psychological Insights to Address Overconfidence in LLMs (2025.findings-acl)
Copied to clipboard
| Challenge: | Psychology research has shown that humans are poor at estimating their performance on tasks, tending towards underconfidence on easy tasks and overconfidence on difficult tasks. |
| Approach: | They propose to use a self-assessment method to assess confidence in large language models (LLMs) they propose to ask for the answer separately and then use them to improve their accuracy. |
| Outcome: | The proposed method improves confidence calibration and interpretability in QA tasks with different personas. |
Why Multi-Interest Fairness Matters: Hypergraph Contrastive Multi-Interest Learning for Fair Conversational Recommender System (2025.findings-acl)
Copied to clipboard
| Challenge: | Unfairness is a well-known challenge in Recommender Systems (RSs) some approaches have started to improve fairness in offline or static contexts, but it often exacerbates over time, leading to significant problems like the Matthew effect, filter bubbles, and echo chambers. |
| Approach: | They propose a framework to promote multi-interest diversity fairness in RSs by establishing diverse hypergraphs through contrastive learning. |
| Outcome: | The proposed framework achieves state-of-the-art performance while effectively alleviating unfairness in two CRS-based datasets. |
Cautious Next Token Prediction (2025.findings-acl)
Copied to clipboard
Yizhou Wang, Lingzhi Zhang, Yue Bai, Mang Tik Chiu, Zhengmian Hu, Mingyuan Zhang, Qihua Dong, Yu Yin, Sohrab Amirghodsi, Yun Fu
| Challenge: | Existing methods for decoding autoregressive models are temperature scaling and nucleus sampling to balance diversity and coherence. |
| Approach: | They propose a training-free decoding strategy that uses a model with a low perplexity score to select the trial with the lowest perplexities as the most probable and reliable path. |
| Outcome: | The proposed approach outperforms existing standard decoding strategies consistently by a clear margin. |
Reasoning with Graphs: Structuring Implicit Knowledge to Enhance LLMs Reasoning (2025.findings-acl)
Copied to clipboard
Haoyu Han, Yaochen Xie, Hui Liu, Xianfeng Tang, Sreyashi Nag, William Headden, Yang Li, Chen Luo, Shuiwang Ji, Qi He, Jiliang Tang
| Challenge: | Large language models (LLMs) have demonstrated remarkable success across a wide range of tasks, however, they still face challenges in reasoning tasks that require understanding and inferring relationships between distinct pieces of information within text sequences. |
| Approach: | They propose to construct explicit graphs from context and leverage them to enhance LLM reasoning performance on reasoning tasks. |
| Outcome: | Extensive experiments show that the proposed method improves both logical reasoning and multi-hop question answering tasks. |
Enhancing Medical Dialogue Generation through Knowledge Refinement and Dynamic Prompt Adjustment (2025.findings-acl)
Copied to clipboard
| Challenge: | Medical dialogue systems (MDS) struggle to identify relevant medical knowledge and generate accurate responses. |
| Approach: | They propose a medical dialogue system that integrates knowledge refining and dynamic prompt adjustment to improve medical knowledge and accuracy. |
| Outcome: | The proposed system outperforms state-of-the-art systems in both generation quality and medical entity accuracy. |
Feature-Level Insights into Artificial Text Detection with Sparse Autoencoders (2025.findings-acl)
Copied to clipboard
Kristian Kuznetsov, Laida Kushnareva, Anton Razzhigaev, Polina Druzhinina, Anastasia Voznyuk, Irina Piontkovskaya, Evgeny Burnaev, Serguei Barannikov
| Challenge: | Existing algorithms for AI text detection lack interpretability, limiting their reliability in highstakes applications. |
| Approach: | They extend existing ATD frameworks by using Sparse Autoencoders to extract features from Gemma-2-2b residual stream. |
| Outcome: | The proposed algorithms can extract human-interpretable features from Gemma-2-2b model. |
Low-Resource Grammatical Error Correction: Selective Data Augmentation with Round-Trip Machine Translation (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for grammatical error correction require large amounts of parallel training data. |
| Approach: | They propose to generate synthetic data through round-trip machine translation by generating a set of character-level errors using a technique known as SeLex-RT. |
| Outcome: | The proposed technique produces errors similar to those observed with language learners, but lacks gold-labeled training data. |
Just Put a Human in the Loop? Investigating LLM-Assisted Annotation for Subjective Tasks (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) have shown impressive performance in many annotation tasks, including subjective tasks common in content moderation and text analysis in the social sciences. |
| Approach: | They propose to give crowdworkers LLM-generated annotation suggestions to "review" LLMs for subjective tasks can impact model performance and analysis downstream . |
| Outcome: | The proposed approach improves self-reported confidence in annotators and models . it also significantly improves model performance by analyzing human-approved datasets. |
Research Community Perspectives on “Intelligence” and Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Despite the widespread use of ‘artificial intelligence’ (AI) framing in NLP research, it is not clear what researchers mean by ”intelligence”. |
| Approach: | They propose to use the term "AI" to describe the perception of a system as intelligent, but note that it is not accepted by the majority of respondents. |
| Outcome: | The results suggest that the perception of the current NLP systems as 'intelligent' is a minority position (29%). |
LEMONADE: A Large Multilingual Expert-Annotated Abstractive Event Dataset for the Real World (2025.findings-acl)
Copied to clipboard
Sina Semnani, Pingyue Zhang, Wanyue Zhai, Haozhuo Li, Ryan Beauchamp, Trey Billing, Katayoun Kishi, Manling Li, Monica Lam
| Challenge: | Using a partially reannotated subset of the Armed Conflict Location & Event Data, we analyze 39,786 conflict events across 20 languages and 171 countries. |
| Approach: | They propose a large-scale conflict event dataset with extensive coverage of region-specific entities. |
| Outcome: | The proposed method detects event arguments and entities through holistic document understanding and normalizes them across the multilingual dataset. |
Memorization vs. Reasoning: Updating LLMs with New Knowledge (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods and benchmarks focus on entity substitutions, failing to capture the full breadth of complex real-world dynamics. |
| Approach: | They propose an automatic pipeline for simulating realistic knowledge updates reflected in an evidence corpus. |
| Outcome: | The proposed method outperforms prior continued pre-training (CPT) baselines on two LLM families and improves direct probing (memorization) results by 25.4%. |
CourtEval: A Courtroom-Based Multi-Agent Evaluation Framework (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing automated evaluation metrics like ROUGE and BLEU show low correlation with human judgments. |
| Approach: | They propose a multi-agent evaluation framework that integrates multiple agents . they use ROUGE and BLEU to evaluate natural language models . |
| Outcome: | The proposed evaluation framework outperforms the current state-of-the-art methods in two meta-evaluation benchmarks. |
Multilingual Definition Modeling (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing definition modeling tasks are mainly encoder-decoder-based, with no explicit definitions. |
| Approach: | They propose a multilingual study on definition modeling using monolingual dictionary data for four new languages. |
| Outcome: | The proposed task is based on monolingual dictionary data for four new languages . results show that multilingual models can perform on-pair with English but cannot leverage potential cross-lingual synergies . |
Human Bias in the Face of AI: Examining Human Judgment Against Text Labeled as AI Generated (2025.findings-acl)
Copied to clipboard
| Challenge: | Prior research on AI mistrust focused primarily on AI's bias towards different human pop-ups. |
| Approach: | They examine how bias shapes the perception of AI versus human generated content . they found that raters favored content labeled "Human Generated" even when labels were deliberately swapped . |
| Outcome: | The findings highlight the limitations of human judgment in interacting with AI and offer a foundation for improving human-AI collaboration. |
Redundancy, Isotropy, and Intrinsic Dimensionality of Prompt-based Text Embeddings (2025.findings-acl)
Copied to clipboard
| Challenge: | Prompt-based text embedding models generate task-specific embeddables but have thousands of dimensions . dimensionality reductions for embedded text can result in performance degradations of only the first 25% of the dimensions resulting in a very small degradation . |
| Approach: | They investigate how post-hoc dimensionality reduction affects performance of various tasks . they find that embeddings for classification and clustering exhibit lower intrinsic dimensionalities . |
| Outcome: | The proposed model generates task-specific embeddings upon receiving tailored prompts, but has thousands of dimensions and high storage costs. |
Harnessing Whisper for Prosodic Stress Analysis (2025.findings-acl)
Copied to clipboard
| Challenge: | Prosody affects how people produce and understand language, yet studies of how it does so have been hindered by the lack of efficient tools for analyzing prosodic stress. |
| Approach: | They propose to use OpenAI Whisper large-v2 to analyze prosodic stress . they propose to characterize decision boundaries used by Whisper for prosodic stress interpretation . |
| Outcome: | The proposed model can learn gender-specific stress patterns and transfer learning from one type of stress to another, surpassing traditional machine learning models. |
Can You Share Your Story? Modeling Clients’ Metacognition and Openness for LLM Therapist Evaluation (2025.findings-acl)
Copied to clipboard
Minju Kim, Dongje Yoo, Yeonjun Hwang, Minseok Kang, Namyoung Kim, Minju Gwak, Beong-woo Kwak, Hyungjoo Chae, Harim Kim, Yunjoong Lee, Min Hee Kim, Dayi Jung, Kyong-Mee Chung, Jinyoung Yeo
| Challenge: | Existing evaluation methods for psychological counseling rely on client simulators that clearly disclose internal states to the therapist, making it difficult to determine whether an LLM therapist can uncover unexpressed perspectives. |
| Approach: | They propose a new evaluation framework featuring a controllable and realistic client simulator which dynamically adapts itself based on the ongoing counseling session. |
| Outcome: | The proposed evaluation framework features a realistic and controllable client simulator which dynamically adapts itself based on the ongoing counseling session, offering a more realistic and challenging evaluation environment. |
Dictionaries to the Rescue: Cross-Lingual Vocabulary Transfer for Low-Resource Languages Using Bilingual Dictionaries (2025.findings-acl)
Copied to clipboard
Haruki Sakajo, Yusuke Ide, Justin Vasselli, Yusuke Sakai, Yingtao Tian, Hidetaka Kamigaito, Taro Watanabe
| Challenge: | Existing approaches to cross-lingual vocabulary transfer face challenges when dealing with low-resource languages. |
| Approach: | They propose a dictionary-based crosslingual vocabulary transfer method that leverages bilingual dictionaries, which are available for many languages thanks to descriptive linguists. |
| Outcome: | The proposed method outperforms existing methods for low-resource languages. |
When Should Dense Retrievers Be Updated in Evolving Corpora? Detecting Out-of-Distribution Corpora Using GradNormIR (2025.findings-acl)
Copied to clipboard
Dayoon Ko, Jinyoung Kim, Sohyeon Kim, Jinhyuk Kim, Jaehoon Lee, Seonghak Song, Minyoung Lee, Gunhee Kim
| Challenge: | Dense retrievers encode text into embeddings to retrieve relevant documents . however, real-world corpora evolve, resulting in degraded retrieval performance . identifying when a dense retriever requires an update is critical for robust retrieval systems . |
| Approach: | They propose a task of predicting whether a corpus is out-of-distribution (OOD) relative to a dense retriever before indexing. |
| Outcome: | The proposed method detects whether a corpus is out-of-distribution (OOD) relative to a dense retriever before indexing. |
The Million Authors Corpus: A Cross-Lingual and Cross-Domain Wikipedia Dataset for Authorship Verification (2025.findings-acl)
Copied to clipboard
| Challenge: | Authorship verification (AV) is a crucial task for identity verification, accountlinking, historical linguistics, and AI-generated text identification. |
| Approach: | They propose to use Wikipedia's Million Authors Corpus to examine authorship verification models on a broad scale. |
| Outcome: | The proposed dataset includes 60.08M textual chunks, contributed by 1.29M Wikipedia authors. |
BridG MT: Enhancing LLMs’ Machine Translation Capabilities with Sentence Bridging and Gradual MT (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent Large Language Models (LLMs) have demonstrated impressive translation performance without the need for fine-tuning on additional parallel corpora. |
| Approach: | They propose a method that combines Sentence Bridging and Gradual MT to enhance LLMs' translation performance even outperforming translation methods that rely on a large number of few-shot examples. |
| Outcome: | The proposed method outperforms translation methods that rely on a large number of few-shot examples even when the source and target languages are low-resource languages. |
Text2World: Benchmarking Large Language Models for Symbolic World Model Generation (2025.findings-acl)
Copied to clipboard
Mengkang Hu, Tianxing Chen, Yude Zou, Yuheng Lei, Qiguang Chen, Ming Li, Yao Mu, Hongyuan Zhang, Wenqi Shao, Ping Luo
| Challenge: | Recent studies have encountered limitations in leveraging large language models to generate symbolic world models. |
| Approach: | They propose a benchmarking framework based on planning domain definition language (PDDL) that employs multi-criteria, execution-based metrics for a more robust evaluation. |
| Outcome: | The proposed model outperforms models trained with large-scale reinforcement learning, but lacks the robustness needed to perform in world modeling. |
Blinded by Context: Unveiling the Halo Effect of MLLM in AI Hiring (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) and Multimodal Large Language Modells (MLLMs) are increasingly being deployed across a range of domains, including finance, law, peer review, and recruitment. |
| Approach: | They investigated how image-based evaluations are influenced by non-job-related information, including extracurricular activities and social media images. |
| Outcome: | The proposed models exhibit significant halo effects in image-based evaluations while text-based assessments showed more resistance to bias. |
CoT-UQ: Improving Response-wise Uncertainty Quantification in LLMs with Chain-of-Thought (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing uncertainty quantification methods for Large language models are primarily prompt-wise rather than response-wise, which leads to inefficiency. |
| Approach: | They propose a new approach to quantify response-wise uncertainty by integrating LLMs’ inherent reasoning capabilities through Chain-of-Thought (CoT) into the UQ process. |
| Outcome: | The proposed framework outperforms existing uncertainty quantification methods and achieves an average improvement of 5.9% AUROC compared to existing methods. |
ADO: Automatic Data Optimization for Inputs in LLM Prompts (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent research has focused on refining instruction components and augmenting input data with in-context examples, but this study explores the potential benefits of optimizing the input data itself. |
| Approach: | They propose a content engineering and structural reformulation strategy to optimize input data within prompts to improve performance of Large Language Models. |
| Outcome: | The proposed approach improves performance of Large Language Models (LLMs) in various tasks, offering a promising avenue for future research in prompt engineering. |
Large Language Models Still Exhibit Bias in Long Text (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing fairness benchmarks for large language models focus on simple tasks . a new framework evaluates biases in LLMs through essay-style prompts . |
| Approach: | They propose a framework that evaluates biases in large language models through essay-style prompts. |
| Outcome: | The proposed framework uncovers subtle biases difficult to detect in simple responses. |
Do Vision-Language Models Have Internal World Models? Towards an Atomic Evaluation (2025.findings-acl)
Copied to clipboard
Qiyue Gao, Xinyu Pi, Kevin Liu, Junrong Chen, Ruolan Yang, Xinqi Huang, Xinyu Fang, Lu Sun, Gautham Kishore, Bo Ai, Stone Tao, Mengyang Liu, Jiaxi Yang, Chao-Jung Lai, Chuanyang Jin, Jiannan Xiang, Benhao Huang, Zeming Chen, David Danks, Hao Su, Tianmin Shu, Ziqiao Ma, Lianhui Qin, Zhiting Hu
| Challenge: | Recent studies have evaluated and shown limitations in specific capabilities such as visual understanding, but a systematic evaluation of VLMs’ fundamental WM abilities remains absent. |
| Approach: | They propose a framework that assesses perception and prediction to provide an atomic evaluation of VLMs as WMs. |
| Outcome: | The proposed framework assesses perception and prediction abilities on 15 latest VLMs and compares them to human-level models. |
Protecting Users From Themselves: Safeguarding Contextual Privacy in Interactions with Conversational Agents (2025.findings-acl)
Copied to clipboard
Ivoline C. Ngong, Swanand Ravindra Kadhe, Hao Wang, Keerthiram Murugesan, Justin D. Weisz, Amit Dhurandhar, Karthikeyan Natesan Ramamurthy
| Challenge: | Conversational agents are increasingly woven into individuals’ personal lives, yet users underestimate the privacy risks associated with them. |
| Approach: | They propose a framework that allows users to reformulate out-of-context information in user prompts by identifying and reformulating out- of-content information in the context. |
| Outcome: | The proposed framework can achieve strong gains in contextual privacy while preserving the user’s intended interaction goals. |
Enhancing Persona Consistency for LLMs’ Role-Playing using Persona-Aware Contrastive Learning (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for analyzing and analyzing large language models (LLMs) lack of emotion and fine-grained role awareness limits the model’s ability to provide personalized and diverse interactions further. |
| Approach: | They propose an annotation-free framework to align LLMs’ behavior during role-playing, enhancing the model’s role consistency. |
| Outcome: | The proposed framework outperforms vanilla LLMs under automatic evaluation methods and human expert evaluation. |
M2-TabFact: Multi-Document Multi-Modal Fact Verification with Visual and Textual Representations of Tabular Data (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing fact-checking systems that can reason over structured data are inefficient compared to humans. |
| Approach: | They propose a multi-modal table-based fact verification task that requires reasoning over visual and textual representations of structured data. |
| Outcome: | The proposed model can reason over visual and textual representations of structured data. |
Fuzzy Speculative Decoding for a Tunable Accuracy-Runtime Tradeoff (2025.findings-acl)
Copied to clipboard
| Challenge: | Speculative Decoding (SD) enforces strict distributional equivalence to the target model when accepting candidate tokens. |
| Approach: | They propose a decoding algorithm that generalizes SD by accepting candidate tokens based on the divergences between the target and draft model distributions. |
| Outcome: | Using Fuzzy Speculative Decoding (FSD) we show that the proposed method can achieve significant runtime improvements of over 5 tokens per second faster than SD at only an approximate 2% reduction in benchmark accuracy. |
PLAY2PROMPT: Zero-shot Tool Instruction Optimization for LLM Agents via Tool Play (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing solutions for large language models rely on manual rewriting or labeled data for validation . Existing approaches rely only on comprehensive tool documentation and in-context demonstrations . |
| Approach: | They propose a framework that "plays" with each tool to explore its input-output behaviors. |
| Outcome: | Experiments show that PLAY2PROMPT improves zero-shot tool performance across open and closed models. |
Towards the Pedagogical Steering of Large Language Models for Tutoring: A Case Study with Modeling Productive Failure (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent studies have shown that LLMs are not able to provide one-to-one tutoring solutions because of their high cost and efficiency. |
| Approach: | They propose an algorithm to optimize LLM prompts and steer it to follow a predefined multi-turn tutoring plan represented as a transition graph. |
| Outcome: | The proposed algorithm is able to optimize LLM prompts and steer it to follow a predefined multi-turn tutoring plan represented as a transition graph. |
Spotting Out-of-Character Behavior: Atomic-Level Evaluation of Persona Fidelity in Open-Ended Generation (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing evaluation methods struggle to capture subtle inconsistencies in large language models. |
| Approach: | They propose an atomic-level evaluation framework that quantifies persona fidelity at a finer granularity. |
| Outcome: | The proposed framework detects inconsistencies that prior evaluation methods overlook . it captures subtle deviations that real users would encounter . |
What Language Do Non-English-Centric Large Language Models Think in? (2025.findings-acl)
Copied to clipboard
Chengzhi Zhong, Qianying Liu, Fei Cheng, Junfeng Jiang, Zhen Wan, Chenhui Chu, Yugo Murawaki, Sadao Kurohashi
| Challenge: | Despite their robust performance in English, these models often exhibit reduced proficiency in non-English languages, and their outputs may reflect an inherent bias toward English-centric perspectives. |
| Approach: | They categorize non-English-centric large language models into two groups: CPMs and BLMs, which are pre-trained on a balanced mix of multiple languages from scratch. |
| Outcome: | The proposed models exhibit a pronounced internal preference for English tokens when projected into the vocabulary space. |
T5Score: A Methodology for Automatically Assessing the Quality of LLM Generated Multi-Document Topic Sets (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing evaluation methods for Multi-Document Topic Extraction are not designed for LLMs and result in low inter-annotator agreement scores. |
| Approach: | They propose an evaluation methodology that decomposes the quality of a topic set into quantifiable aspects, measurable through easy-to-perform annotation tasks. |
| Outcome: | The proposed evaluation methodology decomposes the quality of a topic set into quantifiable aspects, measurable through easy-to-perform annotation tasks. |
Uncertainty-Aware Contrastive Decoding (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models excel in a wide range of tasks, but generating factually accurate outputs remains a challenge. |
| Approach: | They propose a method that dynamically adjusts model contributions at each decoding step based on uncertainty. |
| Outcome: | The proposed method significantly improves factual accuracy and reliability over existing methods. |
GEMS: Generation-Based Event Argument Extraction via Multi-perspective Prompts and Ontology Steering (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for event argument extraction rely on a single prompt . existing methods ignore complex structural and dynamic interdependencies between event arguments . |
| Approach: | They propose a multi-prompt learning framework that generates event arguments via multi-perspective prompts and ontology steering. |
| Outcome: | The proposed framework captures interrelationships between arguments and ontology steering . it uses multiple unfilled prompts for each sentence to generate event arguments . |
RomanLens: The Role Of Latent Romanization In Multilinguality In LLMs (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) exhibit strong multilingual performance despite training on English-centric corpora. |
| Approach: | They propose to use Romanization as a potential bridge in multilingual processing . they propose to encode semantic concepts similarly across native and Romanized scripts . |
| Outcome: | The proposed model encodes semantic concepts across native and Romanized scripts, suggesting a shared underlying representation. |
7 Points to Tsinghua but 10 Points to ? Assessing Large Language Models in Agentic Multilingual National Bias (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models have garnered significant attention for their capabilities in multilingual natural language processing, but studies on risks associated with cross biases are limited to immediate context preferences. |
| Approach: | They investigate multilingual bias in state-of-the-art Large Language Models by analyzing their responses to decision-making tasks across multiple languages. |
| Outcome: | The proposed model can provide personalized advice across university applications, travel, and relocation scenarios. |
Search-in-Context: Efficient Multi-Hop QA over Long Contexts via Monte Carlo Tree Search with Dynamic KV Retrieval (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing approaches to multihop question answering (MHQA) over long contexts are often neglecting explicit reasoning or incurring expensive computational costs due to full-attention mechanisms over long contextuals. |
| Approach: | They propose a framework that integrates Monte Carlo Tree Search (MCTS) with dynamic key-value retrieval to enable iterative, context-aware reasoning. |
| Outcome: | The proposed framework integrates Monte Carlo Tree Search (MCTS) with dynamic key-value (KV) retrieval to enable iterative, context-aware reasoning. |
LLM-as-an-Interviewer: Beyond Static Testing Through Dynamic LLM Evaluation (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent work on LLM-as-a-Judge has reported higher correlations with human judgments due to its static nature. |
| Approach: | They propose a framework that leverages multi-turn interactions where the LLM interviewer actively provides feedback on responses and poses follow-up questions to the evaluated LLM. |
| Outcome: | The proposed framework evaluates six models on reasoning, factuality and instruction-following tasks. |
IntentionESC: An Intention-Centered Framework for Enhancing Emotional Support in Dialogue Systems (2025.findings-acl)
Copied to clipboard
| Challenge: | IntentionESC defines the possible intentions of supporters in emotional support conversations, identifies key emotional state aspects for inferring these intentions, and maps them to appropriate support strategies. |
| Approach: | They propose an Intention-centered Emotional Support Conversation framework which defines the possible intentions of supporters in emotional support conversations, identifies key emotional state aspects for inferring intentions, and maps them to appropriate support strategies. |
| Outcome: | The proposed framework defines the possible intentions of supporters in emotional support conversations, identifies key emotional state aspects for inferring these intentions, and maps them to appropriate support strategies. |
Beyond Context to Cognitive Appraisal: Emotion Reasoning as a Theory of Mind Benchmark for Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent studies have shown that large language models (LLMs) reason about others' emotional states using contextual information, within a Theory-of-Mind framework. |
| Approach: | They propose to use large language models to reason about others’ emotional states using contextual information within a Theory-of-Mind framework. |
| Outcome: | The proposed models can reason about situations and appraisals, but are poor at associating situational outcomes and appraisal with specific emotions. |
CSTRL: Context-Driven Sequential Transfer Learning for Abstractive Radiology Report Summarization (2025.findings-acl)
Copied to clipboard
Mst. Fahmida Sultana Naznin, Adnan Ibney Faruq, Mostafa Rifat Tazwar, Md Jobayer, Md. Mehedi Hasan Shawon, Md Rakibul Hasan
| Challenge: | Pretrained models that excel in abstractive summarization problems face challenges when applied to specialized medical domains due to complex terminology and the necessity for accurate clinical context. |
| Approach: | They propose a sequential transfer learning model that ensures key content extraction and coherent summarization. |
| Outcome: | The proposed model shows 56.2% improvement in BLEU-1, 40.5% in ble-2, 84.3% in blu-3, 28.9% in ROUGE-1, 41.0% in Rough-2 and 26.5% of ROGUE-3 over benchmark studies. |
Rethinking Prompt-based Debiasing in Large Language Model (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing prompt-based methods for debiasing are often superficial and lack a thorough understanding of complex bias concepts. |
| Approach: | They analyze a BBQ and stereoSet benchmarks to examine the assumption that large language models understand biases. |
| Outcome: | The proposed model misclassified 90% of unbiased content as biased despite high accuracy on BBQ dataset . the proposed model may have been flawed in previous attempts to debiase . |
Exploring In-context Example Generation for Machine Translation (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) have demonstrated strong performance across various tasks with just a few examples. |
| Approach: | They propose a method that generates in-context example pairs without external resources. |
| Outcome: | The proposed method builds upon two prior criteria, relevance and diversity, which have been highlighted as key factors for in-context example selection. |
Knowledge Base Construction for Knowledge-Augmented Text-to-SQL (2025.findings-acl)
Copied to clipboard
Jinheon Baek, Horst Samulowitz, Oktie Hassanzadeh, Dharmashankar Subramanian, Sola Shirai, Alfio Gliozzo, Debarun Bhattacharjya
| Challenge: | Existing approaches to translate natural language queries into SQL statements are limited in their parametric knowledge of the database schemas. |
| Approach: | They propose to construct a knowledge base for text-to-SQL, a foundational source of knowledge, from which we retrieve and generate the necessary knowledge for given queries. |
| Outcome: | The proposed approach outperforms baselines on multiple text-to-SQL datasets and shows that it is practical and reliable. |
NBDESCRIB: A Dataset for Text Description Generation from Tables and Code in Jupyter Notebooks with Guidelines (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for Jupyter Notebooks focus on generating cell-level descriptions from code snippets or table outputs independently. |
| Approach: | They propose a task to generate personalized cell-level descriptions using code, tables, and user-written guidelines in Jupyter Notebooks. |
| Outcome: | The proposed task combines code, tables, and user-written guidelines with personalized descriptions to evaluate the performance of existing models. |
ECoRAG: Evidentiality-guided Compression for Long Context RAG (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing compression methods do not focus on filtering out non-evidential information, which limit performance in LLMs. |
| Approach: | They propose a framework that leverages external documents through Retrieval-Augmented Generation (RAG) to reduce RAG overhead. |
| Outcome: | Experiments show that ECoRAG outperforms existing compression methods on ODQA tasks. |
From Complexity to Clarity: AI/NLP’s Role in Regulatory Compliance (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent advances in natural language processing have demonstrated remarkable capabilities in text analysis and reasoning. |
| Approach: | They propose to use standardized evaluation frameworks and balanced human-AI collaboration to address these challenges. |
| Outcome: | The proposed research will focus on standardized evaluation frameworks and balanced human-AI collaboration to address these challenges. |
EXPERT: An Explainable Image Captioning Evaluation Metric with Structured Explanations (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing studies on explainable evaluation metrics generate explanations without standardized criteria and the overall quality of the generated explanations remains unverified. |
| Approach: | They propose a reference-free evaluation metric that provides structured explanations based on fluency, relevance, and descriptiveness. |
| Outcome: | The proposed evaluation template achieves state-of-the-art on benchmark datasets while providing significantly higher-quality explanations than existing metrics. |
Mind Your Theory: Theory of Mind Goes Deeper Than Reasoning (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing benchmarks for Theory of Mind (ToM) focus on whether agents have correct beliefs about others. |
| Approach: | They propose to evaluate Theory of Mind (ToM) capabilities in Large Language Models (LLMs) they propose to use the theory of mind to determine whether and how to invoke ToM . |
| Outcome: | The proposed frameworks can be used to evaluate the performance of large language models (LLMs) in biological agents. |
LLMs are Biased Evaluators But Not Biased for Fact-Centric Retrieval Augmented Generation (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent studies have shown that large language models (LLMs) exhibit significant biases in evaluation tasks, especially in preferentially rating and favoring self-generated content. |
| Approach: | They propose to simulate two critical phases of retrieval-augmented generation (RAG) frameworks where keyword extraction and factual accuracy take precedence over stylistic elements. |
| Outcome: | The proposed model emulates two critical phases of the retrieval-augmented generation framework. |
Standard Quality Criteria Derived from Current NLP Evaluations for Guiding Evaluation Design and Grounding Comparability and AI Compliance Assessments (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing evaluations do not evaluate the same aspect of quality, resulting in unclear comparability and low repeatability. |
| Approach: | They propose to use a standard set of qualitycriterion names and definitions to establish comparability of existing evaluations. |
| Outcome: | The proposed taxonomy combines 114 quality criteria from 3 surveys of 933 evaluations in NLP and is used to establish comparability of existing evaluations and guide the design of new evaluations. |
skLEP: A Slovak General Language Understanding Benchmark (2025.findings-acl)
Copied to clipboard
Marek Suppa, Andrej Ridzik, Daniel Hládek, Tomáš Javůrek, Viktória Ondrejová, Kristína Sásiková, Martin Tamajka, Marian Simko
| Challenge: | skLEP is the first comprehensive benchmark specifically designed for evaluating Slovak natural language understanding models. |
| Approach: | They introduce a benchmark specifically designed for evaluating Slovak natural language understanding models. |
| Outcome: | The proposed benchmark covers nine tasks that span token-level, sentence-pair, document-level tasks. |
Can Vision Language Models Understand Mimed Actions? (2025.findings-acl)
Copied to clipboard
Hyundong Justin Cho, Spencer Lin, Tejas Srinivasan, Michael Saxon, Deuksin Kwon, Natali T. Chavez, Jonathan May
| Challenge: | Nonverbal communication (NVC) is an integral part of human language, but it has been overlooked in natural language processing research. |
| Approach: | They propose a multimodal multimodal recognition task that uses a corpus of mimed gestures to evaluate their understanding of NVC. |
| Outcome: | The proposed task is based on 86 unique gestures with perturbations applied to avatar, background, and viewpoint for evaluating recognition robustness. |
Training Language Model to Critique for Better Refinement (2025.findings-acl)
Copied to clipboard
Tianshu Yu, Chao Xiang, Mingchuan Yang, Pei Ke, Bosi Wen, Cunxiang Wang, Jiale Cheng, Li Zhang, Xinyu Mu, Chuxiong Sun, Minlie Huang
| Challenge: | Large language models (LLMs) have remarkable evaluation and critique capabilities, providing insightful feedback and identifying flaws in various tasks. |
| Approach: | They propose a framework to train critic models using refinement signals to generate feedback loops where critiques guide the model in refining its responses. |
| Outcome: | The proposed framework outperforms traditional methods and open-source models in terms of critique quality and refinement outcomes. |
Dynamic Task Vector Grouping for Efficient Multi-Task Prompt Tuning (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing approaches transfer the soft prompt to low-source targets by combining all source tasks or a single “high-similar” source task one-time-only. |
| Approach: | They propose a method to group similar source tasks based on two metrics: target similarity and knowledge consistency. |
| Outcome: | The proposed method reduces negative transfer and improves performance on low-source targets. |
DICE-BENCH: Evaluating the Tool-Use Capabilities of Large Language Models in Multi-Round, Multi-Party Dialogues (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing function-calling benchmarks focus on single-turn interactions but ignore complexity of real-world scenarios. |
| Approach: | They propose a framework that constructs practical function-calling datasets by synthesizing conversations through a tool graph that maintains dependencies across rounds. |
| Outcome: | The proposed framework synthesizes conversations through a tool graph that maintains dependencies across rounds and a multi-agent system with distinct personas to enhance dialogue naturalness. |
HASH-RAG: Bridging Deep Hashing with Retriever for Efficient, Fine Retrieval and Augmented Generation (2025.findings-acl)
Copied to clipboard
| Challenge: | Experimental evaluations on NQ, TriviaQA, and HotpotQA datasets demonstrate that our approach achieves a 90% reduction in retrieval time compared to conventional methods while maintaining considerate recall performance. |
| Approach: | They propose a framework that integrates deep hashing techniques with systematic optimizations to address these limitations. |
| Outcome: | The proposed framework outperforms retrieval/non-retrieval baselines by 1.4-4.3% in EM scores on NQ, TriviaQA, and HotpotQA datasets. |
A Constrained Text Revision Agent via Iterative Planning and Searching (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing text revision systems are capable of generating fluent and coherent text, but struggle with constrained text revision (CTR). |
| Approach: | They propose a tool that generates revisions tailored to different scenarios using a planner, a reviser and adaptable tools. |
| Outcome: | The proposed agent outperforms baseline approaches in both constraint adherence and revision quality. |
MMRefine: Unveiling the Obstacles to Robust Refinement in Multimodal Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent advances have enabled MLLMs to tackle complex challenges such as mathematical reasoning and multimodal understanding. |
| Approach: | They propose a multimodal refinement benchmark to evaluate the refinement capabilities of Multimodal Large Language Models (MLLMs) the benchmark categorizes errors into six error types to highlight areas for improvement in effective reasoning enhancement. |
| Outcome: | The proposed framework evaluates the refinement capabilities of multimodal large language models across six scenarios. |
How Programming Concepts and Neurons Are Shared in Code Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Several studies have focused on programming languages in a monolingual setting, but most focus on programming language models. |
| Approach: | They perform a few-shot translation task on 21 PL pairs using two Llama-based models and decode the embeddings of intermediate layers. |
| Outcome: | The proposed model assigns high probability to English tokens in the second half of the intermediate layers and language-specific neurons are concentrated in the bottom layers . the model's concept space is closer to English (including PL keywords) and the model is more efficient at identifying language-related neurons. |
DynaQuest: A Dynamic Question Answering Dataset Reflecting Real-World Knowledge Updates (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) are typically trained on static datasets, preventing them from integrating real-time updates. |
| Approach: | They propose a dynamic question-answer answering dataset reflecting real-world knowledge updates that are automatically compared between Wikipedia versions and generating question-anchor pairs based on these updates. |
| Outcome: | The proposed framework improves LLMs' performance on time-sensitive question answering by maintaining a dynamic knowledge updating process. |
ProcrustesGPT: Compressing LLMs with Structured Matrices and Orthogonal Transformations (2025.findings-acl)
Copied to clipboard
| Challenge: | Structured matrix representations of large language models are a promising way to reduce the number of parameters in natural language processing tasks but require a significant amount of computational and memory resources. |
| Approach: | They propose to utilize invariance of the network output under certain orthogonal transformations of weight matrices to identify transformations that improve compressibility of weights within structured classes. |
| Outcome: | The proposed approach is applicable to various types of structured matrices that support efficient projection operations. |
Revisiting In-Context Learning with Long Context Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | In-Context Learning (ICL) is a technique by which language models make predictions based on examples provided in their input context. |
| Approach: | They revisited previous studies using in-context learning techniques . they found that using a data augmentation approach, they significantly improved ICL performance . |
| Outcome: | The proposed approach significantly improves ICL performance on 18 datasets spanning 4 tasks . the proposed approach does not improve performance over a simple random sample selection method . |
Rationalize and Align: Enhancing Writing Assistance with Rationale via Self-Training for Improved Alignment (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing writing assistants rely on supervised fine-tuning to optimize models for multiple revisions. |
| Approach: | They propose a framework that enhances WA performance with rationale and alignment. |
| Outcome: | The proposed framework outperforms state-of-the-art WAs and the closed-source GPT-4o by 3.9 and 7.1 points on average across eight well-established writing-related test sets. |
Accelerating Adaptive Retrieval Augmented Generation via Instruction-Driven Representation Reduction of Retrieval Overlaps (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing approaches to augmented generation ignore the overlap in retrieval results . overlapping content is redundantly represented, affecting the overall efficiency. |
| Approach: | They propose a model-agnostic approach to re-augmented generation that speeds up prefilling and decoding . they propose an instruction-driven module to guide the model to more suitable ways for LLMs . |
| Outcome: | The proposed approach achieves 2.79 and 2.33 times significant acceleration on average for prefilling and decoding respectively while maintaining equal generation quality. |
MEXA: Multilingual Evaluation of English-Centric LLMs via Cross-Lingual Alignment (2025.findings-acl)
Copied to clipboard
Amir Hossein Kargaran, Ali Modarressi, Nafiseh Nikeghbal, Jana Diesner, François Yvon, Hinrich Schuetze
| Challenge: | Existing benchmarks for multilinguality for English-centric large language models focus on classic tasks or cover a minimal number of languages. |
| Approach: | They propose a method to assess multilingual capabilities of pre-trained LLMs using parallel sentences. |
| Outcome: | The proposed method evaluates the multilingual capabilities of pre-trained English-centric models using parallel sentences. |
Automated Fine-Grained Mixture-of-Experts Quantization (2025.findings-acl)
Copied to clipboard
| Challenge: | specialized quantization framework for Mixture of Experts architectures is inadequate for model compression. |
| Approach: | They propose a specialized quantization framework for Mixture of Experts architectures . they find that expert networks exhibit distinctive channel-wise outlier distributions ." |
| Outcome: | The proposed framework improves on the Mixtral-8x7b-v0.1 architecture while maintaining minimal computational overhead. |
Enhancing Complex Reasoning in Knowledge Graph Question Answering through Query Graph Approximation (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing knowledge-grounded question answering frameworks lack essential triplets related to the questions . Existing approaches to knowledge-based QA are incomplete in the context of KGs . |
| Approach: | They propose a framework to provide answers to structured queries by leveraging Knowledge Graphs. |
| Outcome: | The proposed framework outperforms existing methods on QA tasks where KGs are incomplete . the framework is based on a set of data from a dataset of QA questions . |