Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)

Discover and Prove: An Open-source Agentic Framework for Hard Mode Automated Theorem Proving in Lean 4 (2026.acl-long)

Challenge:	Existing summarization systems struggle to address diverse linguistic and cognitive barriers among general readers.
Approach:	They propose a multi-agent framework that integrates template-based planning with an iterative feedback loop guided by simulated readers and domain expert revision to address comprehension barriers such as unknown terms, missing contexts, and confusing sentences.
Outcome:	The proposed framework improves readability and factuality across multiple datasets and human evaluations show that it is more accessible to a wide range of readers.

Copied to clipboard

Chengwu Liu, Yichun Yin, Ye Yuan, Jiaxuan Xie, Botao Li, Siqi Li, Jianhao Shen, Yan Xu, Lifeng Shang, Ming Zhang

Your Inference Request Will Become a Black Box: Confidential Inference for Cloud-based Large Language Models (2026.acl-long)

Challenge:	Existing approaches to solving mathematical problems fall into two broad categories: informal methods and formal methods.
Approach:	They propose to use LLM natural-language reasoning to discover answers . they introduce Discover And Prove framework that rewrites Hard Mode statements into Easy Mode ones for existing ATP provers.
Outcome:	The proposed framework can be used to prove hard mode statements on ATP benchmarks.

Copied to clipboard

Chung-ju Huang, Huiqiang Zhao, Yuanpeng He, Lijian Li, Wenpin Jiao, Zhi Jin, Peixuan Chen, Leye Wang

Rhetorical Questions in LLM Representations: A Linear Probing Study (2026.acl-long)

Challenge:	Existing approaches fail to ensure privacy, maintain model performance, and preserve computational efficiency simultaneously.
Approach:	They propose a confidential inference framework that partitions the LLM pipeline between a client-verified Confidential Virtual Machine (CVM) and the public cloud to protect client data without compromising the cloud’s model intellectual property or inference quality.
Outcome:	The proposed framework can defend against state-of-the-art token inference attacks while preserving model privacy, performance, and efficiency.

Copied to clipboard

Louie Hong Yao, Vishesh Anand, Yuan Zhuang, Tianyu Jiang

Dual-Axis Generative Reward Model Toward Semantic and Turn-taking Robustness in Interactive Spoken Dialogue Models (2026.acl-long)

Challenge:	Rhetorical questions are asked not to seek information, but to persuade or signal stance . how large language models internally represent rhetorical questions remains unclear .
Approach:	They analyze rhetorical questions in LLM representations using linear probes on two social-media datasets with different discourse contexts.
Outcome:	The results show that rhetorical signals emerge early and are most stably captured by last-token representations.

Copied to clipboard

Yifu Chen, Shengpeng Ji, Zhengqing Liu, Qian Chen, Wen Wang, Ziqing Wang, Yangzhuo Li, Tianle Liang, Zhou Zhao

Different types of syntactic agreement recruit the same units within large language models (2026.acl-long)

Challenge:	Reinforcement learning (RL) has improved text- and vision-language models, but its application in SDMs is hindered.
Approach:	They propose a dual-axis Generative Reward Model that provides semantic quality and interaction timing for SDMs.
Outcome:	The proposed model achieves state-of-the-art performance on interaction-quality assessment across a wide spectrum of datasets.

Copied to clipboard

Daria Kryvosheieva, Andrea Gregor de Varda, Evelina Fedorenko, Greta Tuckute

Empowering Tabular Data Preparation with Language Models: Why and How? (2026.acl-long)

Challenge:	Large language models can reliably distinguish grammatical from ungrammatically sentences, but how gramatical knowledge is represented within the models remains an open question.
Approach:	They use a functional localization approach inspired by cognitive neuroscience to identify the LLM units most responsive to 67 English syntactic phenomena in seven open-weight models.
Outcome:	The proposed model is most responsive to 67 English syntactic phenomena and consistently supports model performance.

Copied to clipboard

Mengshi Chen, Yuxiang Sun, Tengchao Li, Jianwei Wang, Kai Wang, Xuemin Lin, Ying Zhang, Wenjie Zhang

Learning Diverse Responses with Prefix-Conditioned Supervised Fine-Tuning (2026.acl-long)

Challenge:	Tabular data preparation is a critical step in enhancing the usability of tabular data.
Approach:	They analyze how LMs can be combined with other components for different tabular data preparation tasks.
Outcome:	The proposed methods lack the ability to capture the relationships within tables and adapt to the tasks involved.

Copied to clipboard

Zhiyuan Fan, Guanqiao Chen, Yanyi Huang, Mingkuan Zhao, Dadi Guo, Yi R. Fung

EASE: Entity-Aware Sub-table Generation for Real-world Multi-table QA (2026.acl-long)

Challenge:	Large language models exhibit highly homogeneous, repetitive responses, resulting in inefficient exploration.
Approach:	They propose a method that constructs semantically consistent yet distributionally distinct prior contents to different responses and decouple the one-to-many mapping.
Outcome:	The proposed method improves absolute performance by 5.3% and increases generation diversity by 198.3% on average while significantly enhancing output diversity and test-time scaling.

Copied to clipboard

Myunghoon Kang, Dahyun Jung, Suhyune Son, Seonmin Koo, Changwoo Chun, Daniel Rim, Haeyoung Kwon, Yuna Hur, Heuiseok Lim

Benchmarking LLM’s Capability in Reasoning over Conflicting Web References (2026.acl-long)

Challenge:	Table-based question answering (table QA) is a powerful tool for analyzing large language models.
Approach:	They propose to use noisy multi-table sets to generate sub-tables for table-based question answering.
Outcome:	The proposed framework efficiently filters out irrelevant information while incorporating pertinent table values.

Copied to clipboard

Yizhen Yuan, Rui Kong, Dongze Li, Yuanchun Li, Yunxin Liu

Parallel Universes, Parallel Languages: A Comprehensive Study on LLM-based Multilingual Counterfactual Example Generation (2026.acl-long)

Challenge:	Large language models (LLMs) integrated with retrieval-augmented generation (RAG) are a dominant framework for building intelligent assistants.
Approach:	They propose a benchmark to evaluate LLMs' reasoning capability over real-world conflicting documents retrieved from the web.
Outcome:	The proposed benchmark evaluates LLMs' reasoning capability over real-world conflicting documents retrieved from the web.

Copied to clipboard

Qianli Wang, Van Bach Nguyen, Yihong Liu, Fedor Splitt, Nils Feldhus, Christin Seifert, Hinrich Schuetze, Sebastian Möller, Vera Schmitt

CLEAR: Cross-Lingual Enhancement in Retrieval via Reverse-training (2026.acl-long)

Challenge:	Large language models excel at generating English counterfactuals but their effectiveness in generating multilingual counterfacts remains unclear.
Approach:	They conduct automatic evaluations on both directly generated and derived counterfactuals in six languages and find that cross-lingual perturbations follow common strategic principles.
Outcome:	The proposed models show that translation-based counterfactuals offer higher validity than their directly generated counterparts, but still fall short of matching the quality of the original English counterf actuals.

Copied to clipboard

Seungyoon Lee, Minhyuk Kim, Seongtae Hong, Youngjoon Jang, Dongsuk Oh, Heuiseok Lim

Aligning Language Models with Real-time Knowledge Editing (2026.acl-long)

Challenge:	Existing multilingual embedding models often struggle to capture cross-lingual alignment during training.
Approach:	They propose a novel loss function that leverages an English passage as a bridge to strengthen alignments between target language and English.
Outcome:	The proposed model improves retrieval performance across cross-lingual scenarios while minimizing performance degradation in English.

Copied to clipboard

Chenming Tang, Yutong Yang, Kexue Wang, Yunfang Wu

PosterForest: Hierarchical Multi-Agent Collaboration for Scientific Poster Generation (2026.acl-long)

Challenge:	Mainstream knowledge editing methods are static and fail to keep pace with the evolving real-world knowledge.
Approach:	They propose a new paradigm for knowledge editing that integrates edit augmentation and self-adaptive post-alignment inference into CRAFT to improve edit success.
Outcome:	The proposed method shows significant performance gain on CRAFT and traditional datasets compared to existing methods.

Copied to clipboard

Jiho Choi, Seojeong Park, Seongjong Song, Hyunjung Shim

SLR: Automated Synthesis for Scalable Logical Reasoning (2026.acl-long)

Challenge:	Existing methods for scientific poster generation lack hierarchical document understanding and coherent content-layout planning.
Approach:	They propose a training-free framework for scientific poster generation that captures document hierarchy and semantics across multiple levels.
Outcome:	The proposed framework outperforms existing methods in both automatic and human evaluations without additional training or domain-specific supervision.

Copied to clipboard

Lukas Helff, Ahmad Omar, Felix Friedrich, Antonia Wüst, Hikaru Shindo, Rupert Mitchell, Tim Woydt, Patrick Schramowski, Wolfgang Stammer, Kristian Kersting

Can Continual Pretraining Bridge the Performance Gap between General-purpose and Specialized Language Models in the Medical Domain? (2026.acl-long)

Challenge:	Existing benchmarks intended to evaluate reasoning capabilities emphasize deductive reasoning, where conclusions necessarily follow from given premises.
Approach:	They propose an end-to-end framework for systematic evaluation and training of Large Language Models via Scalable Logical Reasoning.
Outcome:	The proposed framework doubles Llama-3-8B accuracy on SLR-Bench, achieving parity with Gemini-Flash-Thinking at a fraction of computational cost.

Copied to clipboard

Niclas Doll, Jasper Schulze Buschhoff, Shalaka Satheesh, Hammam Abdelwahab, Héctor Allende-Cid, Katrin Klug

MINTQA: A Multi-Hop Question Answering Benchmark for Evaluating LLMs on New and Long-tail Knowledge (2026.acl-long)

Challenge:	specialized models have a large potential for translation and translation, but they lack the integration of domainspecific knowledge and terminology into clinical workflows.
Approach:	They construct a German medical corpus to continuously pre-train and merge three well-known LLMs and use it to improve model performance.
Outcome:	The proposed model family significantly outperforms the mistral-Small-24B-Instruct model family on German medical benchmarks.

Copied to clipboard

Jie He, Nan Hu, Wanqiu Long, Jiaoyan Chen, Jeff Z. Pan

Sense and Sensitivity: Examining the Influence of Semantic Recall on Long Context Code Understanding (2026.acl-long)

Challenge:	Existing studies have shown that large language models can handle knowledge with varying familiarity.
Approach:	They propose a benchmark to evaluate multi-hop question answering on new and tail knowledge . they use RAG to integrate external knowledge into large language models .
Outcome:	The proposed benchmark evaluates the multi-hop reasoning ability of large language models . it primarily evaluates their ability to handle knowledge with different levels of familiarity .

Copied to clipboard

Adam Štorek, Mukur Gupta, Samira Hajizadeh, Prashast Srivastava, Suman Jana

SenseRel: A Sense-Level Benchmark for Denotational and Connotational Meaning Relations (2026.acl-long)

Challenge:	Large language models are increasingly used for understanding large codebases, but whether they understand operational semantics of long code context is unclear.
Approach:	They propose a task that achieves high semantic recall sensitivity through unpredictable operations.
Outcome:	The proposed task SemTrace achieves high semantic recall sensitivity through unpredictable operations.

Copied to clipboard

Pierluigi Cassotti, Naomi Baes, Stefano De Pascale, Jáder Martins Camboim de Sá, Francesco Periti, Nick Haslam, Dirk Geeraerts, Nina Tahmasebi

WebSTAR: Scalable Data Synthesis for Computer Use Agents with Step-Level Filtering (2026.acl-long)

Challenge:	Polysemy enables a single word to convey multiple related meanings . a word's sense is extended to new contexts and concepts, a process called semantic change is gradual .
Approach:	They propose a benchmark for modeling semantic relations between word senses . they use a model that distinguishes denotational and connotationally related aspects of meaning .
Outcome:	The proposed model is able to distinguish between denotational and connotationalist aspects of meaning . it is compared with models with GPT-4o, Llama 3.1, and DeepSeek .

Copied to clipboard

Yifei He, Pranit Chawla, Yaser Souri, Subhojit Som, Xia Song

PR-XAI: PageRank-Based Feature Attribution for Transformers (2026.acl-long)

Challenge:	Existing datasets rely on human demonstrations, limiting scalability.
Approach:	They propose a scalable data synthesis pipeline that transforms noisy rollouts into reliable supervision without human annotation.
Outcome:	The proposed pipeline transforms noisy rollouts into reliable supervision without human annotation.

Copied to clipboard

Behrooz Azarkhalili, Linyi Li, Maxwell W. Libbrecht

CAPruner: Conceptual-Adjacent Scene Graph Pruner for Enhancing 3D Spatial Reasoning of Large Language Models (2026.acl-long)

Challenge:	Existing feature attribution methods for transformer models suffer from limitations that undermine their efficacy.
Approach:	They propose a feature attribution method for transformer models based on PageRank . they propose attribution methods that apply PageRank to attention-derived graphs .
Outcome:	The proposed method outperforms state-of-the-art methods in faithfulness and classification metrics with significant gains on long-form text.

Copied to clipboard

Shengli Zhou, Xiangchen Wang, Guanhua Chen, Feng Zheng

MauBERT: Universal Phonetic Inductive Biases for Few-Shot Acoustic Units Discovery (2026.acl-long)

Challenge:	Existing pruning methods rely on spatial proximity and remove relevant relations, thereby undermining reliable spatial reasoning.
Approach:	They propose a scene graph pruning model that integrates fuzzy semantic relevance with spatial proximity to estimate the importance of relations.
Outcome:	Experiments show that CAPruner outperforms proximity-based pruning with negligible cost savings.

Copied to clipboard

Angelo Ortiz Tandazo, Manel Khentout, Youssef Benchekroun, Thomas Hueber, Emmanuel Dupoux

Your Reasoning Model Knows What Counts: Self-Guided Chain-of-Thought Pruning for Efficient Reasoning (2026.acl-long)

Challenge:	MauBERT models learn from multilingual data to predict articulatory features or phones, resulting in language-independent phonetic representations.
Approach:	They introduce a multilingual extension of HuBERT that leverages articulatory features for robust cross-lingual phonetic representation learning.
Outcome:	The proposed model can predict phonetic features in 55 languages with minimal fine-tuning (10 hours of speech) it is more context-invariant than state-of-the-art models and adapts to unseen languages and casual speech with minimal self-supervised fine- tuning (10 hours)

Copied to clipboard

Zi-Ao Ma, Xian-Ling Mao, Tian Lan, Chen Xu, Zhijing Wu

ImReasoner: Improving Memory-based Language Models for Reasoning-in-a-Haystack Tasks (2026.acl-long)

Challenge:	Existing approaches to Chain-of-Thought reasoning are often degraded because they disregard the model’s intrinsic reasoning dependency.
Approach:	They propose a self-guided pruning framework that leverages the model’s intrinsic likelihood landscape to identify segments that are extraneous to its specific reasoning pattern.
Outcome:	The proposed framework reduces output length while maintaining or improving accuracy on multiple benchmarks.

Copied to clipboard

Ching-Yun Ko, Payel Das, Sihui Dai, Georgios Kollias, Subhajit Chaudhury, Aurelie C. Lozano, Pin-Yu Chen

How Memory Management Impacts LLM Agents: An Empirical Study of Experience-Following Behavior (2026.acl-long)

Challenge:	despite advances, large language models exhibit brittleness on tasks that require multi-step reasoning over long contexts.
Approach:	They propose to explicitly encode contexts as ordered memory and perform iterative retrieval to construct reasoning chains.
Outcome:	The proposed frameworks fail to show emergent reasoning generalization in a weakly supervised scenario . the proposed framework is based on a synthetic benchmark to stress-test the models .

Copied to clipboard

Zidi Xiong, Yuping Lin, Wenya Xie, Pengfei He, Zirui Liu, Jiliang Tang, Himabindu Lakkaraju, Zhen Xiang

SACTOR: LLM-Driven Correct and Idiomatic C to Rust Translation with Static Analysis and FFI-Based Verification (2026.acl-long)

Challenge:	In practice, memory designs vary widely across agents due to their diverse objectives and functionalities.
Approach:	They conduct an empirical study on how memory management choices impact the LLM agents’ behavior, especially their long-term performance.
Outcome:	The proposed methods show that LLM agents display an experience-following property, which results in highly similar agent outputs.

Copied to clipboard

Tianyang Zhou, Ziyi Zhang, Haowen Lin, Somesh Jha, Mihai Christodorescu, Kirill Levchenko, Varun Chandrasekaran

IntrAgent: An LLM Agent for Content-Grounded Information Retrieval through Literature Review (2026.acl-long)

Challenge:	Large language models (LLMs) have shown promise in producing idiomatic translations, but offer no correctness guarantees.
Approach:	They propose a C-to-Rust translation tool that uses an initial "unidiomatic" translation followed by an "idiomatic refinement" they evaluate SACTOR on 200 programs from two datasets and two more complex scenarios .
Outcome:	The proposed tool delivers high end-to-end correctness and produces safe, idiomatic Rust with up to 7 fewer Clippy warnings.

Copied to clipboard

Fengbo Ma, Zixin Rao, Xiaoting Li, Zhetao Chen, Hongyue Sun, Yiping Zhao, Xianyan Chen, Zhen Xiang

Reasoning over Precedents Alongside Statutes: Case-Augmented Deliberative Alignment for LLM Safety (2026.acl-long)

Challenge:	Scientific research relies on accurate information retrieval from literature to support analytical decisions.
Approach:	They propose a task that automates fine-grained information retrieval faithfully grounded in the provided content in response to research-driven queries.
Outcome:	The proposed agent achieves 13.2% higher cross-domain accuracy than state-of-the-art RAG and research-agent baselines across seven backbone LLMs.

Copied to clipboard

Can Jin, Rui Wu, Tong Che, Qixin Zhang, Hongwu Peng, Jiahui Zhao, Zhenting Wang, Wenqi Wei, Ligong Han, Zhao Zhang, Yuan Cao, Ruixiang Tang, Dimitris N. Metaxas

Hybrid Autoregressive-Diffusion Model for Real-Time Sign Language Production (2026.acl-long)

Challenge:	OpenAI introduces deliberative alignment (DA) to enhance safety of its o-series models, but effectiveness of this approach in open-source LLMs is understudied.
Approach:	They propose a case-augmented deliberative alignment method for large language models . they propose to use reinforcement learning on self-generated safety reasoning chains .
Outcome:	The proposed method avoids narrowly enumerated rules and allows broader adaptability.

Copied to clipboard

Maoxiao Ye, Xinfeng Ye, Sathiamoorthy Manoharan

Challenge:	Existing approaches to sign language production use autoregressive or diffusion models that generate one-by-one output tokens but suffer from exposure bias during inference.
Approach:	They propose a hybrid autoregressive-diffusion model that combines iterative refinement and sequential dependency modeling for Sign Language production.
Outcome:	The proposed model improves sign language production quality and real-time efficiency on PHOENIX14T and How2Sign.

Augur: Modeling Covariate Causal Associations in Time Series via Large Language Models (2026.acl-long)

Copied to clipboard

Zhiqing Cui, Binwu Wang, Qingxiang Liu, Yeqiang Wang, Zhengyang Zhou, Yuxuan Liang, Yang Wang

AgentRouter: A Knowledge-Graph-Guided LLM Router for Collaborative Multi-Agent Question Answering (2026.acl-long)

Challenge:	Large language models (LLMs) have emerged as a promising avenue for time series forecasting . existing approaches face limitations such as marginalized role in model architectures and lack of interpretability.
Approach:	They propose a framework that exploits LLM causal reasoning to discover and use directed causal associations among covariates.
Outcome:	The proposed model improves predictive accuracy while yielding transparent, traceable reasoning about variable interactions.

Copied to clipboard

Zheyuan Zhang, Kaiwen Shi, Zhengqing Yuan, Zehong Wang, Tianyi Ma, Keerthiram Murugesan, Vincent Galassi, Chuxu Zhang, Yanfang Ye

MoCa: Modality-aware Continual Pre-training Makes Better Bidirectional Multimodal Embeddings (2026.acl-long)

Challenge:	Existing approaches to agent routing emphasize cost efficiency while overlooking the fine-grained contextual and relational structure inherent in QA tasks.
Approach:	They propose a framework that formulates multi-agent QA as a knowledge-graph-guided routing problem supervised by empirical performance signals.
Outcome:	The proposed framework outperforms single-agent and ensemble baselines while generalizing across benchmarks and LLM backbones.

Copied to clipboard

Haonan Chen, Hong Liu, Yuping Luo, Liang Wang, Nan Yang, Furu Wei, Zhicheng Dou

ReEx-SQL: Reasoning with Execution-Aware Reinforcement Learning for Text-to-SQL (2026.acl-long)

Challenge:	Existing approaches to embed multimodal models face limitations such as suboptimal causal attention in VLMs and limited diversity in training objectives and data.
Approach:	They propose a framework for transforming pre-trained VLMs into bidirectional multimodal embedding models.
Outcome:	The proposed model improves performance across MMEB and ViDoRe-v2 benchmarks and exhibits strong scalability with both model size and training data on MMEF.

Copied to clipboard

Yaxun Dai, Wenxuan Xie, Xialie Zhuang, Tianyu Yang, Ziyi Liu, Haiqin Yang, Yiying Yang, Yuhang Zhao, Pingfu Chao, Wenhao Jiang

DecIF: Improving Instruction-Following through Decomposition (2026.acl-long)

Challenge:	Current Text-to-SQL reasoning models lack integrated execution feedback during generation.
Approach:	They propose a text-to-SQL framework that interacts with the SQL execution engine during decoding and dynamically adjusts reasoning based on execution feedback.
Outcome:	The proposed framework achieves 89.1% accuracy on Spider and 65.3% on BIRD at the 7B scale.

Copied to clipboard

Tingfeng Hui, Pengyu Zhu, Bowen Ping, Ling Tang, Guanting Dong, Yaqi Zhang, Sen Su

Challenge:	Existing approaches to obtain high-quality instruction-following data rely heavily on existing documents and existing methods.
Approach:	They propose a data synthesis framework, DecIF, which automatically generates accurate and diverse instruction-following data from scratch for supervised fine-tuning and reinforcement learning.
Outcome:	Extensive experiments show that the proposed framework can synthesize accurate instruction-following data for both SFT and RL paradigms compared to baselines.

NavA3: Understanding Any Instruction, Navigating Anywhere, Finding Anything (2026.acl-long)

Copied to clipboard

Lingfeng Zhang, Xiaoshuai Hao, Yingbo Tang, Haoxiang Fu, Xinyu Zheng, Pengwei Wang, Zhongyuan Wang, Wenbo Ding, Shanghang Zhang

FastV-RAG: Towards Fast and Fine-Grained Video QA with Retrieval-Augmented Generation (2026.acl-long)

Challenge:	Existing embodied navigation methods struggle with such tasks due to their limitations in comprehending high-level human instructions and localizing objects with an open vocabulary.
Approach:	They propose a hierarchical framework for long-horizon navigation that integrates human instructions with 3D scene views.
Outcome:	The proposed model achieves SOTA results and can complete long-horizon navigation tasks across different robot embodiments in real-world environments.

Copied to clipboard

Gen Li, Peiyu Liu

Efficient Provably Secure Linguistic Steganography via Range Coding (2026.acl-long)

Challenge:	Existing methods for retrieval-augmented generation are inefficient and often fail to maintain high answer quality.
Approach:	They propose an efficient VLM-based RAG framework built on a speculative decoding pipeline and a similarity-based filtering strategy to mitigate errors.
Outcome:	The proposed framework reduces inference latency without sacrificing correctness . it achieves comparable or higher accuracy than standard approaches while speeding up inference by approximately 2x .

Copied to clipboard

Ruiyi Yan, Yugo Murawaki

Uni-MMMU: A Massive Multi-discipline Multimodal Unified Benchmark (2026.acl-long)

Challenge:	Linguistic steganography is a promising field in safeguarding information . previous methods have achieved perfect imperceptibility but at the expense of embedding capacity.
Approach:	They propose to use a classical entropy coding method to achieve secure steganography . they propose to employ a rotation mechanism to achieve embedding efficiency .
Outcome:	The proposed method outperforms existing methods in embedding capacity and embeddability.

Copied to clipboard

Kai Zou, Ziqi Huang, Yuhao Dong, Shulin Tian, Dian Zheng, Hongbo Liu, Jingwen He, Bin Liu, Yu Qiao, Ziwei Liu

MemSearch-o1: Empowering Large Language Models with Reasoning-Aligned Memory Growth in Agentic Search (2026.acl-long)

Challenge:	Existing evaluations treat visual understanding and generation in isolation or overlook tasks that inherently couple them.
Approach:	They propose a benchmark that examines the bidirectional synergy between generation and understanding across eight reasoning-centric domains.
Outcome:	The proposed model systematically unfolds the bidirectional synergy between generation and understanding across eight reasoning-centric domains.

Copied to clipboard

Sheng Zhang, Junyi Li, Yingyi Zhang, Pengyue Jia, Yichao Wang, Xiaowei Qian, Wenlin Zhang, Maolin Wang, Yong Liu, Xiangyu Zhao

CAPC-CG: A Large-Scale, Expert-Directed LLM-Annotated Corpus of Adaptive Policy Communication in China (2026.acl-long)

Challenge:	Existing methods for memory management struggle to capture fine-grained semantic relations between queries and documents.
Approach:	They propose a framework for reasoning and agentic search that grows fine-grained memory fragments from seed tokens from queries, then retraces and deep refines the memory via a contribution function.
Outcome:	Experiments on eight benchmark datasets show that MemSearch-o1 significantly mitigates memory dilution and more effectively activates reasoning potential of diverse LLMs.

Copied to clipboard

Bolun Sun, Charles Chang, Yuen Yuen Ang, Ruotong Mu, Yuchen Xu, Zhengxin Zhang, Pingxu Hao

Challenge:	Adaptive policy communication is a theory of governance in large, decentralized organizations where leaders exercise influence rather than precise control by combining clear and ambiguous instructions to calibrate discipline and flexibility.
Approach:	They propose an expert-directed annotation method that integrates codebook design, structured training, a two-step workflow, and LLM-based scaling.
Outcome:	The proposed method achieves a Fleiss’ kappa of 0.86 on directive labels, indicating high reliability.

MathFlow: Enhancing the Perceptual Flow of MLLMs for Visual Mathematical Problems (2026.acl-long)

Copied to clipboard

Shuhang Chen, Hangjie Yuan, Yunqiu Xu, Pengwei Liu, Tao Feng, Jun Cen, Zeying Huang, Yi Yang

Anchored Sliding Window: Toward Robust and Imperceptible Linguistic Steganography (2026.acl-long)

Challenge:	Existing multimodal large language models (MLLMs) exhibit significant limitations when extracting essential information and reasoned properties from diagrams and performing complex reasoning based on these visual inputs.
Approach:	They propose a benchmark that provides a fine-grained evaluation of MLLMs’ perception and reasoning capabilities.
Outcome:	The proposed benchmark shows that existing MLLMs exhibit limitations when extracting essential information and reasoned properties from diagrams and performing complex reasoning based on these visual inputs.

Copied to clipboard

Ruiyi Yan, Shiao Meng, Yugo Murawaki

What Makes Good Instruction-Tuning Data? An In-Context Learning Perspective (2026.acl-long)

Challenge:	linguistic steganography assumes that stegographic texts are fragile to even minor modifications, compromising text quality.
Approach:	They propose an anchored sliding window framework to improve imperceptibility and robustness . they propose to include the prompt and a bridge context within the context window .
Outcome:	The proposed framework outperforms the baseline method in text quality, imperceptibility and robustness across diverse settings.

Copied to clipboard

Guangzeng Han, Xiaolei Huang

RoBSA: RoPE-based Blockwise Sparse Multi-head Latent Attention (2026.acl-long)

Challenge:	Existing methods for instruction-tuning data contain redundancy and low-quality samples.
Approach:	They propose an instruction data selection framework based on weighted in-context influence . they show that sample difficulty negatively correlates with in-constext influence.
Outcome:	The proposed method outperforms baselines under constrained data budgets while demonstrating that sample difficulty negatively correlates with in-context influence.

Copied to clipboard

Xinyu Shi, Kairong Luo, Zhen Zheng, Wenguang Chen

XtraGPT: Context-Aware and Controllable Academic Paper Revision via Human-AI Collaboration (2026.acl-long)

Challenge:	Large Language Models (LLMs) have advanced in recent years, scaling up in both parameter count and context length.
Approach:	They propose a method to compute attention over a subset of context tokens and to implement token selection in a blockwise manner.
Outcome:	The proposed method reduces end-to-end inference latency by up to 2.55x with minimal accuracy loss compared to full attention in long-context scenarios for very large models.

Copied to clipboard

Nuo Chen, Andre Lin HuiKai, Jiaying Wu, Junyi Hou, Zining Zhang, Qian Wang, Xidong Wang, Bingsheng He

Challenge:	Existing systems are designed for general-purpose scientific text generation and fail to support high-quality scientific writing beyond surface-level polishing.
Approach:	They propose a human-AI collaboration framework for academic paper revision based on criteria-guided intent alignment and context-aware modeling.
Outcome:	The proposed framework outperforms existing LLMs and rivals the quality of proprietary ones.

Learn to Relax with Large Language Models: Solving Constraint Optimization Problems via Bidirectional Coevolution (2026.acl-long)

Copied to clipboard

Beidan Liu, Zhengqiu Zhu, Chen Gao, Tianle Pu, Yong Zhao, Wei Qi, Quanjun Yin

A Game-Theoretica Negotiation Framework for Cross-Cultural Consensus (2026.acl-long)

Challenge:	Large Language Model (LLM)-based optimization has shown promise for autonomous problem solving, but most approaches cast LLMs as passive constraint checkers rather than proactive strategy designers.
Approach:	They propose an end-to-end Automated Constraint Optimization method that tightly couples operations-research principles of constraint relaxation with LLM reasoning.
Outcome:	Extensive experiments on three challenging COP benchmarks validate AutoCO’s consistent effectiveness and superior performance, especially in hard regimes where current methods degrade.

Copied to clipboard

Guoxi Zhang, Jiawei Chen, Tianzhuo Yang, Jiaming Ji, Yaodong Yang, Juntao Dai

Focusing Condition: Inference-Time Self-Contrastive Steering Elicits Better Conditional Text Embeddings in LLMs (2026.acl-long)

Challenge:	Large language models exhibit pronounced WEIRD cultural bias, marginalizing diverse viewpoints and posing challenges for reconciling diverse populations with varying cultural backgrounds and value systems.
Approach:	They propose a framework for cross-cultural fairness using a Nash Equilibrium . they propose equilibriums that iteratively propose and refine natural-language guidelines .
Outcome:	The proposed framework generates higher-quality and more balanced consensus . it finetunes diverse LLM architectures with negotiation data, reducing cultural distances by 95.53%.

Copied to clipboard

Zifeng Cheng, Lingyun Qian, Zhiwei Jiang, Cong Wang, Yafeng Yin, Fei Shen, Ao Zhou, Qing Gu

Exploring Attention Attractors in Large Language Models (2026.acl-long)

Challenge:	Existing methods for extracting conditional text embeddings from large language models (LLMs) relying on prompts often fails to produce high-quality conditional embeddables, resulting in degradation of quality.
Approach:	They propose a plug-and-play method that constructs unconditional general text embeddings and uses them to refine conditional text embeds.
Outcome:	The proposed method improves performance of prompt-based methods on clustering, Semantic Textual Similarity, and triplet alignment datasets.

Copied to clipboard

Ziheng Wang, Zihao Yue, Wenxuan Wang, Qin Jin

Identifying the Periodicity of Information in Natural Language (2026.acl-long)

Challenge:	Existing studies have suggested that attention attractors function as "summary tokens" while others speculate that tokens with weaker semantics attract high attention, they act as attention sinks that offload excessive attention.
Approach:	They examine attention attractors, tokens that draw significantly high attention, in large language models.
Outcome:	The proposed models are able to capture long-range dependencies within a given context.

Copied to clipboard

Yulin OU, Yu Wang, Yang Xu, Hendrik Buschmeier

Discourse & Pragmatics

Challenge:	Existing methods to detect periodicity of information in natural language are based on a canonical periodicity detection algorithm.
Approach:	They propose a method to detect periods in surprisal sequences in natural language . they propose to use this method to identify periods outside the distributions of typical units .
Outcome:	The proposed method can detect significant periods in a single document.

EmoHarbor: Evaluating Personalized Emotional Support by Simulating the User’s Internal World (2026.acl-long)

Copied to clipboard

Jing Ye, Lu Xiang, Yaping Zhang, Chengqing Zong

ReasonEmbed: Enhanced Text Embeddings for Reasoning-Intensive Document Retrieval (2026.acl-long)

Challenge:	EmoHarbor is an evaluation framework that rewards generic empathetic responses but fails to assess whether the support is genuinely personalized to users’ unique psychological profiles and contextual needs.
Approach:	They propose an automated evaluation framework that adopts a User-as-a-Judge paradigm by simulating the user's inner world.
Outcome:	The proposed framework decomposes users' internal processes into three specialized roles and defines 10 evaluation dimensions of personalized support quality.

Copied to clipboard

Jianlyu Chen, Junwei Lan, Chaofan Li, Defu Lian, Zheng Liu

SPAGBias: Uncovering and Tracing Structured Spatial Gender Bias in Large Language Models (2026.acl-long)

Challenge:	Recent studies suggest that traditional retrievers struggle with reasoningintensive tasks such as personal assistants and scientific research.
Approach:	They propose a new data synthesis method that overcomes the triviality problem prevalent in previous synthetic datasets and propose 'ReMixer', a data fusion method that generates 82K high-quality training samples.
Outcome:	The proposed model outperforms existing models on reasoning-intensive retrieval tasks.

Copied to clipboard

Binxian Su, Haoye Lou, Shucheng Zhu, Weikang Wang, Ying Liu, Dong Yu, Pengyuan Liu

VALU: A Benchmark for Video Anomaly Temporal Localization and Understanding at Multiple Semantic Levels (2026.acl-long)

Challenge:	Large language models (LLMs) are being used in urban planning but there is concern that they reproduce or amplify such biases.
Approach:	They propose a framework to evaluate spatial gender bias in large language models . they use a taxonomy of 62 urban micro-spaces, a prompt library and three diagnostic layers .
Outcome:	The proposed framework identifies structured gender-space associations that go beyond the public-private divide, forming nuanced micro-level mappings.

Copied to clipboard

Yixiao He, Menghao Zhang, Haifeng Sun, Jing Wang, Kangheng Lin, Jinghan Wang, Chenye Xu, Pengfei Ren, Qi Qi, Jingyu Wang

CoreGaze: Core Subgraph-Driven Visual Gaze Diffusion for Training-Free Referring Multimodal Large Language Models (2026.acl-long)

Challenge:	Recent advances in Video Large Language Models (Video-LLMs) enhance the ability of VAU models to describe and interpret anomalies.
Approach:	They propose a benchmark that explicitly defines anomalies across five semantic levels and provides detailed temporal boundaries and detailed textual descriptions for each.
Outcome:	The proposed benchmark defines anomalies across five semantic levels and provides detailed descriptions for each.

Copied to clipboard

Xiaoyang Yi, Jing Chen, Yuru Bao, Jian Zhang

RespiraMFM: A Multimodal Foundation Model with Contrastive Audio-Language Alignment for Respiratory Disease Identification (2026.acl-long)

Challenge:	Existing methods rely on extensive fine-tuning to mitigate attention distraction, leading to redundant outputs or hallucinations.
Approach:	They propose a training-free framework that simulates human visual gaze diffusion for fine-grained comprehension by combining a sparse semantic graph with a core subgraph with amplified initial influence.
Outcome:	The proposed framework simulates human visual gaze diffusion for fine-grained comprehension.

Copied to clipboard

Shakhrul Iman Siam, Tiantian Feng, Jiankun Zhang, Shrikanth Narayanan, Mi Zhang

Beyond Detection: Evaluating Fallacy Awareness of LLMs in Interactive Scenarios (2026.acl-long)

Challenge:	Existing models for respiratory diseases rely on audio inputs, but they lack generalizability and diagnostic precision.
Approach:	They propose a multimodal foundation model that integrates respiratory sounds with medical history and symptoms to enhance diagnostic accuracy and disease detection capabilities.
Outcome:	The proposed model improves AUROC and zero-shot tasks across five respiratory diseases using real-world datasets.

Copied to clipboard

Conghui Niu, Ningxin Wu, Ziran Zhao, Dong Yu, Chen Kang, Pengyuan Liu

Crosscoding Through Time: Tracking Emergence & Consolidation Of Linguistic Representations Throughout LLM Pretraining (2026.acl-long)

Challenge:	Large Language Models fail to recognize fallacious reasoning in real-world interactions despite strong performance on static fallacy detection tasks.
Approach:	They propose a Chinese benchmark to assess fallacy awareness without explicit cues . they propose 'fate' evaluation framework that assesses fallacy without explicit .
Outcome:	The proposed framework assesses fallacy awareness without explicit cues, combining natural dialogue responses and reasoning-based decisions.

Copied to clipboard

Deniz Bayazit, Aaron Mueller, Antoine Bosselut

Is EEG-to-Text Feasible in Real-World Scenarios? An In-Depth Analysis Using a Neuropsychology-Inspired Benchmark (2026.acl-long)

Challenge:	Large language models learn non-trivial abstractions during pretraining, but it is not well understood when and how these specific linguistic abilities emerge.
Approach:	They propose a method to track the evolution of linguistic features during pretraining by using sparse crosscoders to discover and align features across model checkpoints.
Outcome:	The proposed approach can detect features emergence, maintenance, and discontinuation during training stages.

Copied to clipboard

Zihan Zhang, Yu Bao, Xiao Ding, Tianyi Jiang, Kai Xiong

Challenge:	Existing benchmarks for EEG2Text have neglected EEG instability, a problem that has confounded inference and sparked debate.
Approach:	They propose to use a 128-channel high-density EEG cap to evaluate EEG2Text models . they find existing benchmarks have neglected EEG instability, a flaw that has confounded inferences and sparked debate .
Outcome:	The proposed benchmarks provide key evidence for teacher-forcing-free decoding of EEG2Text models.

ControlAudio: Tackling Text-Guided, Timing-Indicated and Intelligible Audio Generation via Progressive Diffusion Modeling (2026.acl-long)

Copied to clipboard

Yuxuan Jiang, Zehua Chen, Zeqian Ju, Yusheng Dai, Weibei Dou, Jun Zhu

PrefixNLI: Detecting Factual Inconsistencies as Soon as They Arise (2026.acl-long)

Challenge:	Recent efforts on text-to-audio generation are exploring fine-grained controllability . however, their performance at scale is limited due to data scarcity .
Approach:	They propose a multi-task learning problem for high-controllability text-to-audio generation . they propose scalable diffusion transformers that augment condition information in sequence .
Outcome:	The proposed method outperforms existing methods on objective and subjective evaluations.

Copied to clipboard

Sapir Harary, Eran Hirsch, Aviv Slobodkin, David Wan, Mohit Bansal, Ido Dagan

Textual Inference

Challenge:	Large Language Models (LLMs) are prone to factually inconsistent statements, known as hallucinations.
Approach:	They propose to train a specialized model that detects inconsistencies over text prefixes to improve generation faithfulness by 5-14 F1 points.
Outcome:	The proposed model outperforms baseline models by 5-14 F1 points in prefix-level entailment.

RAM-SD: Retrieval-Augmented Multi-agent framework for Sarcasm Detection (2026.acl-long)

Copied to clipboard

Ziyang Zhou, Ziqi Liu, Yan Wang, Yiming Lin, Yangbin Chen

On the Emergence and Test-Time Use of Structural Information in Large Language Models (2026.acl-long)

Challenge:	Existing approaches to sarcastic detection use a uniform reasoning strategy . existing approaches lack a framework to deal with the diverse analytical demands of sarcasm .
Approach:	They propose a Retrieval-Augmented Multi-Agent framework for Sarcasm Detection . the framework provides transparent and interpretable reasoning traces .
Outcome:	The proposed framework outperforms existing methods on four benchmarks and outperformed the strong GPT-4o+CoC baseline.

Copied to clipboard

Michelle Chao Chen, Moritz Miller, Bernhard Schölkopf, Siyuan Guo

SAGE: A Search-AuGmented Evaluation of Large Language Models on Free-Form QA (2026.acl-long)

Challenge:	a controlled environment is required to study how language models learn structural information from observational data.
Approach:	They propose a natural language dataset based on linguistic structural transformations to study how language models learn abstract structures and utilize the learnt structural information at test-time.
Outcome:	The proposed model can generate new knowledge outside the training corpus in a controlled environment.

Copied to clipboard

Sher Badshah, Ali Emami, Hassan Sajjad

J4R: Learning to Judge with Equivalent Initial State Group Relative Policy Optimization (2026.acl-long)

Challenge:	Large Language Models (LLMs) are prone to hallucination and rely on static, pre-annotated references for evaluation.
Approach:	They propose a framework to assess large language models without fixed ground-truth answers by iteratively generating web queries and synthesizing external evidence.
Outcome:	The proposed framework achieves substantial to perfect agreement with human evaluations on multiple free-form QA benchmarks.

Copied to clipboard

Austin Xu, Yilun Zhou, Xuan-Phi Nguyen, Caiming Xiong, Shafiq Joty

A Functionality-Grounded Benchmark for Evaluating Web Agents in E-commerce Domains (2026.acl-long)

Challenge:	Large language models (LLMs) are increasingly being used for reasoning intensive tasks.
Approach:	They propose an algorithm that trains judges to be robust to positional biases . they also propose a benchmark that evaluates judges in diverse reasoning settings .
Outcome:	The proposed algorithm outperforms GPT-4o and the next best small judge by 6.7% and 9% on ReasoningJudgeBench and JudgeBench.

Copied to clipboard

Xianren Zhang, Shreyas Prasad, Di Wang, Qiuhai Zeng, Suhang Wang, Wenbo Yan, Mat Hans

Challenge:	Existing benchmarks focus on product search tasks, but ignore potential risks.
Approach:	They propose a data generation pipeline that leverages webpage content and interactive elements to create diverse, functionality-grounded user queries.
Outcome:	The proposed framework assesses the performance and safety of web agents under dynamic, real-world e-commerce environments.

Reinforcement Learning for Self-Improving Agent with Skill Library (2026.acl-long)

Copied to clipboard

Jiongxiao Wang, Qiaojing Yan, Yawei Wang, Yijun Tian, Soumya Smruti Mishra, Zhichao Xu, Megha Gandhi, Panpan Xu, Lin Lee Cheong

Challenge:	Large Language Model (LLM)-based agents have demonstrated remarkable capabilities in complex reasoning and multi-turn interactions but struggle to continuously improve and adapt when deployed in new environments.
Approach:	They propose a Reinforcement Learning-based approach to enhance agents’ self-improvement capabilities with a skill library.
Outcome:	The proposed framework achieves 8.9% higher Scenario Goal Completion when applied to supervised-finetuned model with expert experience while requiring 26% fewer interaction steps and generating 59% fewer tokens.

CachePrune: Teaching LLMs What Not to Follow via KV-Cache Editing (2026.acl-long)

Copied to clipboard

Rui Wang, Junda Wu, Yu Xia, Tong Yu, Ruiyi Zhang, Ryan A. Rossi, Subrata Mitra, Lina Yao, Julian McAuley

Flip-Flop Consistency: Unsupervised Training for Robustness to Prompt Perturbations in LLMs (2026.acl-long)

Challenge:	Existing Large Language Models exhibit critical vulnerability to indirect prompt injection attacks, where instructions injected within in the prompt context can override the user's intent.
Approach:	They propose a neural pruning algorithm that prunes neurons associated with instruction-following during KV cache encoding of the prompt context.
Outcome:	The proposed approach significantly reduces the attack success rate while preserving the model's ability to follow user instructions.

Copied to clipboard

Parsa Hejabi, Elnaz Rahmati, Alireza Salkhordeh Ziabari, Morteza Dehghani

Leave My Images Alone: Preventing Multi-Modal Large Language Models from Analyzing Images via Visual Prompt Injection (2026.acl-long)

Challenge:	Large Language Models (LLMs) often produce inconsistent answers when faced with different phrasings of the same prompt.
Approach:	They propose an unsupervised method that flips the phrasings of prompts into a hard pseudo-label . they use Consensus Cross-Entropy to create a consensus, and representation alignment loss to pull lower-confidence predictors toward consensus .
Outcome:	The proposed method raises observed agreement by 11.62% and improves mean F1 by 8.94% on 11 datasets spanning four NLP tasks .

Copied to clipboard

Zedian Shao, Hongbin Liu, Yuepeng Hu, Neil Zhenqiang Gong

Logic Matters in Lightweight Hallucination Classification for RAG System (2026.acl-long)

Challenge:	Multimodal large language models are powerful tools for analyzing Internet-scale image data.
Approach:	They propose a method to protect images from unauthorized analysis by MLLMs . they embed a perturbation that acts as a visual prompt injection attack on MLMLs if a malicious actor downloads and queries an image .
Outcome:	The proposed method protects images from unauthorized analysis by MLLMs . it embeds a perturbation that acts as a visual prompt injection attack on MLMLs if a malicious actor downloads and queries the protected image .

Copied to clipboard

Ningyuan Yang, Kaizhu Huang

Harnessing Negative Signals: Reinforcement Distillation from Teacher Data for LLM Reasoning (2026.acl-long)

Challenge:	Existing hallucination detection frameworks for RAGs lack robustness and performance . a compact model may lose track of precise information in retrieved segments or misinterpret a document's entailment score.
Approach:	They propose a lightweight, modular framework for hallucination detection in RAG systems . they capture logical relationships among retrieved documents within the vector space .
Outcome:	The proposed framework improves hallucination detection in RAG systems without complex architectures or pre-training on datasets.

Copied to clipboard

Shuyao Xu, Cheng Peng, Jiangxuan Long, Weidi Xu, Wei Chu, Yuan Qi

Grammar Search for Multi-Agent Systems (2026.acl-long)

Challenge:	Recent advances in model distillation show that data from advanced reasoning models can effectively train smaller student models.
Approach:	They propose a method to use both positive and negative distilled reasoning traces to maximize LLM reasoning performance in offline settings.
Outcome:	The proposed model outperforms existing methods in the distillation context.

Copied to clipboard

Mayank Singh, Vikas Yadav, Shiva Krishna Reddy Malay, Shravan Nayak, Sai Rajeswar, Sathwik Tejaswi Madhusudhan, Eduardo Blanco

ZoomR: Memory Efficient Reasoning through Multi-Granularity Key Value Retrieval (2026.acl-long)

Challenge:	Several prior approaches have relied on LLM-based free-form search over the code space.
Approach:	They propose a more structured framework that explores the same space through a fixed set of composable components.
Outcome:	The proposed framework outperforms existing approaches on most benchmarks across two backbone LLMs and two domains: mathematics and question answering.

Copied to clipboard

David H. Yang, Yuxuan Zhu, Mohammad Mohammadi Amiri, Keerthiram Murugesan, Tejaswini Pedapati, Subhajit Chaudhury, Pin-Yu Chen

Explicit Trait Inference for Multi-Agent Coordination (2026.acl-long)

Challenge:	Large language models generate long chain of thoughts but memory footprint grows with output length . prior work on KV cache optimization focused on compressing long input context .
Approach:	They propose a new approach that compresses verbose reasoning thoughts into summaries . they use a dynamic KV cache selection policy that leverages these summary keys .
Outcome:	The proposed approach reduces memory usage while avoiding full-cache attention at each step.

Copied to clipboard

Suhaib Abdurahman, Etsuko Ishii, Katerina Margatina, Divya Bhargavi, Monica Sunkara, Yi Zhang

Challenge:	Large language model (LLM) based multi-agent systems (MAS) show promise on complex tasks but remain prone to failures of coordination, such as goal drift, error cascades, and misaligned behaviors.
Approach:	They propose a psychologically grounded method for improving coordination using Explicit Trait Inference (ETI) ETI enables agents to infer and track partner characteristics along two established psychological dimensions—warmth (e.g., trust) and competence (eg. skill)
Outcome:	The proposed method reduces payoff loss in controlled and realistic multi-agent settings by 45–77% and improves performance by 3–29% depending on scenario and model.

SFTMix: Elevating Language Model Instruction Tuning with Mixup Recipe (2026.acl-long)

Copied to clipboard

Yuxin Xiao, Shujian Zhang, Marzyeh Ghassemi, Wenxuan Zhou

Psychological Steering in LLMs: An Evaluation of Effectiveness and Trustworthiness (2026.acl-long)

Challenge:	Efforts to improve instruction tuning often focus on higher-quality supervised fine-tuning datasets, typically requiring data filtering with proprietary LLMs or human annotation.
Approach:	They propose a Mixup-based recipe that elevates LLM instruction tuning without relying on well-curated datasets.
Outcome:	The proposed model improves instruction-following and healthcare-specific tasks with consistent improvements across LLM families and SFT datasets.

Copied to clipboard

Amin Banayeeanzade, Ala N. Tak, Fatemeh Bahrani, Anahita Bolourani, Leonardo Blas, Emilio Ferrara, Jonathan Gratch, Sai Praneeth Karimireddy

Hierarchical Policy Optimization for Simultaneous Translation of Unbounded Speech (2026.acl-long)

Challenge:	Using a model with a high degree of emotion and personality control, large language models can be used to control socially interactive interactions.
Approach:	They propose a Psychologically-informed benchmark to evaluate LLM steering effectiveness and trustworthiness across emotion and personality domains.
Outcome:	The framework establishes the first holistic evaluation of emotion and personality steering, offering insights into its interpretability and reliability for socially interactive applications.

Copied to clipboard

Siqi Ouyang, Shuoyang Ding, Oleksii Hrinchuk, Vitaly Lavrukhin, Brian Yan, Boris Ginsburg, Lei Li

Act as you think: Reinforcing Consistent Reasoning in Medical Visual Question Answering (2026.acl-long)

Challenge:	Existing synthesis methods cannot guarantee data quality.
Approach:	They propose a hierarchical reward that balances translation quality and latency objectives by combining supervised fine-tuning data with supervised inputs.
Outcome:	The proposed model can reuse key-value caches across both modalities and eliminate redundant feature recomputation.

Copied to clipboard

Songtao Jiang, Yuan Wang, Ruizhe Chen, Yan Zhang, Ruilin Luo, Bohan Lei, Yeying Jin, Sibo Song, ZhiBo Yang, Jimeng Sun, Jian Wu, Zuozhu Liu

CRISP: Persistent Concept Unlearning via Sparse Autoencoders (2026.acl-long)

Challenge:	Recent advances have improved the accuracy of medical visual question answering (Med-VQA) however, the high stakes nature of the medical domain has precipitated a shift towards interpretability and transparency of reasoning processes.
Approach:	They propose a reinforcement learning from verifiable rewards framework that rewards internal consistency and logical coherence.
Outcome:	The proposed framework rewards internal consistency and logical coherence, and is highly versatile, the authors show.

Copied to clipboard

Tomer Ashuach, Dana Arad, Aaron Mueller, Martin Tutek, Yonatan Belinkov

LongSpec: Long-Context Lossless Speculative Decoding with Efficient Drafting and Verification (2026.acl-long)

Challenge:	Recent work has explored sparse autoencoders (SAEs) to perform precise interventions on monosemantic features, but most SAE-based methods operate at inference time, which does not create persistent changes in the model’s parameters.
Approach:	They propose a parameter-efficient method for persistent concept unlearning using SAEs that automatically identifies salient SAE features across multiple layers and suppresses their activations.
Outcome:	The proposed method outperforms previous methods on safety-critical unlearning tasks from the WMDP benchmark, successfully removing harmful knowledge while preserving general and in-domain capabilities.

Copied to clipboard

Penghui Yang, Cunxiao Du, Fengzhuo Zhang, Haonan Wang, Tianyu Pang, Chao Du, Bo An

Using Perspectival Words Is Harder Than Vocabulary Words for Humans —and Even More So for Multimodal Language Models (2026.acl-long)

Challenge:	Large Language Models (LLMs) can process extremely long contexts, requiring efficient inference over extended inputs.
Approach:	They propose a model that uses a constant-sized key-value cache to train long-context models.
Outcome:	Experimental results show that LongSpec achieves 3.26x speedup over strong Flash Attention baselines and 2.34x wall clock time on four math reasoning tasks.

Copied to clipboard

Dota Tianai Dong, Yifan Luo, Po-Ya Angela Wang, Asli Ozyurek, Paula Rubio-Fernandez

CHIMERA: A Knowledge Base of Scientific Idea Recombinations for Research Analysis and Ideation (2026.acl-long)

Challenge:	Existing evaluations of multimodal language models focus on vocabulary words with relatively stable, context-independent meanings in conversation, such as object names, colors, and verbs.
Approach:	They compare human and multimodal language models in their use of three word types: vocabulary, possessives, and demonstratives.
Outcome:	The models approach human-level performance on using vocabulary, but exhibit clear deficits with possessives and even greater difficulties with demonstratives.

Copied to clipboard

Noy Sternlicht, Tom Hope

Knowledge Representation

Challenge:	a hallmark of human innovation is recombination.
Approach:	They propose a task to extract recombination instances from scientific literature . they analyze patterns of recombined concepts and apply it to a broad corpus of AI papers .
Outcome:	The proposed model can predict cross-disciplinary research directions . it can predict recombinations across areas and link methods and concepts .

Towards Robust Real-World Spreadsheet Understanding with Multi-Agent Multi-Format Reasoning (2026.acl-long)

Copied to clipboard

Houxing Ren, Mingjie Zhan, Zimu Lu, Ke Wang, Yunqiao Yang, Haotian Hou, Hongsheng Li

GATE: Graph-based Adaptive Tool Evolution Across Diverse Tasks (2026.acl-long)

Challenge:	Spreadsheets are among the most widely used data formats in real-world applications . existing large language models treat tables as plain text, overlooking layout cues and visual semantics.
Approach:	They propose a two-stage multi-agent framework for spreadsheet understanding that adopts a step-by-step reading and reasoning paradigm.
Outcome:	Extensive experiments on two spreadsheet datasets show the proposed framework outperforms existing methods on Spreadsheet Bench.

Copied to clipboard

Jianwen Luo, Yiming Huang, Jinxiang Meng, Fangyu Lei, Shizhu He, Xiao Liu, Shanshan Jiang, Bin Dong, Jun Zhao, Kang Liu

Erasing Without Remembering: Implicit Knowledge Forgetting in Large Language Models (2026.acl-long)

Challenge:	Existing toolsets that use large language models are limited to single-task settings.
Approach:	They propose a framework that dynamically constructs and evolves a hierarchical graph of reusable tools across multiple scenarios.
Outcome:	The proposed framework achieves up to 4.3 faster milestone completion in Minecraft compared to the previous state-of-the-art method and provides an average improvement of 9.23% over existing tool-making methods in code generation tasks and 10.03% in agent tasks.

Copied to clipboard

Huazheng Wang, Yongcheng Jing, Haifeng Sun, Yingjie Wang, Jingyu Wang, Jianxin Liao, Dacheng Tao

Mitigating Hallucinations in Large Vision-Language Models without Performance Degradation (2026.acl-long)

Challenge:	a new method for unlearning large language models is proposed to improve the performance of large language model models.
Approach:	They propose a probability perturbation-based unlearning paradigm that allows models to forget implicit knowledge in large language models with a focus on generalisation.
Outcome:	The proposed model improves unlearning vanilla target data while forgetting implicit knowledge.

Copied to clipboard

Xingyu Zhu, Junfeng Fang, Shuo Wang, Beier Zhu, Zhicai Wang, Yonghui Yang, Xiangnan He

APB-V: Accelerating Long-Video Understanding via Sequence-Parallelism-aware Approximate Attention (2026.acl-long)

Challenge:	Recent advances in large vision-language models produce hallucinations that compromise output reliability.
Approach:	They propose a dual-stage framework for mitigating hallucinations without performance degradation . they propose semantic-aware component disentanglement and interpretable parameter updates .
Outcome:	The proposed model reduces hallucinations by 23.4% while maintaining 97.4% of general generative capability.

Copied to clipboard

Yuxiang Huang, Mingye Li, Xu Han, Chaojun Xiao, Weilin Zhao, Ao Sun, Ziqi Yuan, Hao Zhou, Fandong Meng, Zhiyuan Liu

AraVQA: Building a New Arabic Factoid Visual Question Answering Dataset from Wikipedia (2026.acl-long)

Challenge:	Existing methods for long-video inference use compression or sparse attention . existing methods restrict LMMs from handling longer, more complex videos .
Approach:	They propose a sequence-parallel framework with optimized attention that accelerates long-video inference across multiple GPUs.
Outcome:	The proposed framework delivers speedups of 12.72x, 1.70x, and 1.18x over FlashAttn, ZigZagRing, and APB without significant performance loss.

Copied to clipboard

Sultan Alrowili, Younes Samih, Abed Alhakim Freihat, Mathan Kumar Eswaran

What is a protest anyway? Codebook conceptualization is still a first-order concern in LLM-era classification (2026.acl-long)

Challenge:	Existing Arabic VQA datasets focus on culturally-specific and dialect-aware domains.
Approach:	They propose a pipeline that leverages Wikipedia template tags to extract relevant information for each image and utilize it to generate a new visual question answering dataset.
Outcome:	The proposed pipeline can enhance existing VLMs on Arabic VQA tasks by leveraging Wikipedia template tags.

Copied to clipboard

Andrew Halterman, Katherine A. Keith

When Misinformation Speaks and Converses: Rethinking Fact-Checking in Audio Platforms (2026.acl-long)

Challenge:	generative large language models (LLMs) are used extensively for text classification in computational social science . conceptualization of categories to classify and using LLM predictions can tempt analysts to skip conceptualization altogether.
Approach:	They argue that LLMs can tempt analysts to skip conceptualization altogether . they argue that conceptualization failures induce downstream inferential bias .
Outcome:	The proposed model can tempt analysts to skip conceptualization altogether . the proposed model is a first-order concern in the LLM-era .

Copied to clipboard

Chaewan Chun, Delvin Ce Zhang, Dongwon Lee

Ethical NLP

Challenge:	Existing fact-checking pipelines focus on written claims, not on audio . authors argue that audio misinformation is structurally different because it is both spoken and conversational .
Approach:	They argue that audio misinformation is structurally different because it is both spoken and conversational . they argue that advancing fact-checking requires rethinking verification pipelines around spoken and conversations .
Outcome:	The proposed method fails on audio because it is both spoken and conversational . podcasts exceed 4.3 million distinct shows, reaching an estimated 500 million listeners globally .

Contextual Relevance and Adaptive Sampling for LLM-Based Document Reranking (2026.acl-long)

Copied to clipboard

Jerry Huang, Siddarth Madala, Cheng Niu, Julia Hockenmaier, Tong Zhang

Quantifying Aleatoric Uncertainty of In-Context Learning for Robust Measure of LLM Prediction Confidence (2026.acl-long)

Challenge:	identifying relevant documents for Reasoning-intensive queries remains a challenge . large language models have shown strong performance in zero-shot document reranking .
Approach:	They propose a reranking algorithm that estimates contextual relevance by aggregating LLMs' relevance judgments across batches.
Outcome:	The proposed algorithm improves nDCG@10 over retrieval and reranking baselines by 15% and 6–21% respectively.

Copied to clipboard

Jinseok Chung, Minkyoung Song, Hyunji Jung, Namhoon Lee

SATQuest: A Verifier for Logical Reasoning Evaluation and Reinforcement Fine-Tuning of LLMs (2026.acl-long)

Challenge:	Existing methods for standard generation tasks fail to capture the unique dynamics of ICL.
Approach:	They propose a concept of self-function vectors that leverage Bayesian views and the mechanistic interpretability of ICL to model latent concept learned during in-context prompting.
Outcome:	The proposed framework can be used for trustworthy-related applications, such as hallucination detection.

Copied to clipboard

Yanxiao Zhao, Yaqian Li, Zi-Hao Bo, Rinyoichi Takezoe, Haojia Hui, Mo Guang, null Renlei, Xiaolin Qin, Kaiwen Long

Not All Directions Matter: Towards Structured and Task-Aware Low-Rank Model Adaptation (2026.acl-long)

Challenge:	Large language models exhibit strong general reasoning abilities, yet the community lacks controllable, scalable, and verifiable tools to analyze and improve them.
Approach:	They propose a verifier that generates diverse SAT-based reasoning tasks from CNF instances and checks answers objectively with PySAT.
Outcome:	The proposed verifier generates diverse SAT-based reasoning tasks from CNF instances and checks answers objectively with PySAT.

Copied to clipboard

Xi Xiao, Chenrui Ma, Yunbei Zhang, Chen Liu, Zhuxuanzi Wang, Yanshu Li, Lin Zhao, Guosheng Hu, Tianyang Wang, Hao Xu

ReContraster: Making Your Posters Stand Out with Regional Contrast (2026.acl-long)

Challenge:	Low-Rank Adaptation (LoRA) is a key parameter-efficient fine-tuning method . however, its effectiveness is hampered by semantic drift and structural incoherence .
Approach:	They propose a low-rank Adaptation framework that tackles semantic drift and structural incoherence by pruning task-irrelevant directions.
Outcome:	Experiments on large language models, vision models, and vision models show that the proposed framework outperforms LoRA and advanced dynamic rank allocation and sparsity-based methods.

Copied to clipboard

Peixuan Zhang, Zijian Jia, Ziqi Cai, Shuchen Weng, Si Li, Boxin Shi

Cognitive Modeling

Challenge:	Effective poster design requires rapidly capturing attention and clearly conveying messages.
Approach:	They propose a poster-based model that leverages regional contrast to make posters stand out.
Outcome:	The proposed model outperforms state-of-the-art methods in producing striking posters.

Mechanistic Interpretability Should Prioritize Feature Consistency in Sparse Autoencoders (2026.acl-long)

Copied to clipboard

Xiangchen Song, Aashiq Muhamed, Yujia Zheng, Lingjing Kong, Zeyu Tang, Mona T. Diab, Virginia Smith, Kun Zhang

EquivPruner: Boosting Efficiency and Quality in LLM-Based Search via Action Pruning (2026.acl-long)

Challenge:	Sparse Autoencoders (SAEs) are a tool in mechanistic interpretability (MI) but the aspiration to identify a canonical set of features is challenged by the observed inconsistency of learned SAE features across different training runs.
Approach:	They propose to use the Pairwise Dictionary Mean Correlation Coefficient to quantify SAE feature consistency as an evaluation axis alongside reconstruction and sparsity.
Outcome:	The proposed measure is based on the pairwise dictionary mean correlation coefficient (PW-MCC) on LLM activations.

Copied to clipboard

Jiawei Liu, Qisi Chen, Jianshu Zhang, Quan Liu, Defu Lian

From Data-Centric to Sample-Centric: Enhancing LLM Reasoning via Progressive Optimization (2026.acl-long)

Challenge:	Existing semantic similarity methods struggle to accurately identify semantically equivalent steps in domain-specific contexts like mathematical reasoning.
Approach:	They propose a simple yet effective approach that identifies and prunes semantically equivalent actions during LLM reasoning search.
Outcome:	The proposed approach reduces token consumption while improving reasoning efficiency and accuracy on Qwen2.5-Math-7B-Instruct on GSM8K.

Copied to clipboard

Xinjie Chen, Minpeng Liao, Guoxin Chen, Chengxi Li, Biao Fu, Kai Fan, Xinggao Liu

CARES: Context-Aware Resolution Selector for VLMs (2026.acl-long)

Challenge:	Reinforcement learning with verifiable rewards (RLVR) has recently advanced the reasoning capabilities of large language models (LLMs).
Approach:	They propose a method that incorporates partial solution prefixes from expert demonstrations to guide the policy.
Outcome:	The proposed methods outperform strong baselines, yielding faster convergence and a higher performance ceiling.

Copied to clipboard

Moshe Kimhi, Nimrod Shabtay, Raja Giryes, Chaim Baskin, Eli Schwartz

ChatR1: Reinforcement Learning for Conversational Reasoning and Retrieval Augmented Question Answering (2026.acl-long)

Challenge:	Large vision–language models process images at native or high resolution to remain effective across tasks.
Approach:	They propose a lightweight preprocessing module that predicts the minimum sufficient input resolution for large vision–language models.
Outcome:	CARES predicts when a pre-trained VLM's response converges to its peak ability to answer correctly, reducing compute by up to 80%.

Copied to clipboard

Simon Lupart, Mohammad Aliannejadi, Evangelos Kanoulas

Think in Sentences: Explicit Sentence Boundaries Enhance Language Model’s Capabilities (2026.acl-long)

Challenge:	Unlike static ‘rewrite, retrieve, and generate’ pipelines, ChatR1 interleaves search and reasoning across turns, enabling exploratory and adaptive behaviors learned through RL.
Approach:	They propose a reasoning framework based on reinforcement learning (RL) for conversational question answering that interleaves search and reasoning across turns and provides turn-level feedback.
Outcome:	The proposed framework outperforms competing models on five CQA datasets, measured by different metrics (F1, BERTScore, and LLM-as-judge).

Copied to clipboard

Zhichen Liu, Yongyuan Li, Yang Xu

MORPHOGEN: A Multilingual Benchmark for Evaluating Gender-Aware Morphological Generation (2026.acl-long)

Challenge:	Existing studies focus on dummy tokens but fail to leverage the inherent sentence-level structure of natural language.
Approach:	They propose a method that inserts delimiters at sentence boundaries to enhance large language models' capabilities.
Outcome:	The proposed method improves performance on 7B LLMs to 600B Deepseek-V3 with 7.7% gains on GSM8k and 12.5% on DROP.

Copied to clipboard

Mehul Agarwal, Aditya Aggarwal, Arnav Goel, Medha Hira, Anubha Gupta

TeRA: Vector-based Random Tensor Network for High-Rank Adaptation of Large Language Models (2026.acl-long)

Challenge:	In morphologically rich languages, gender influences verb conjugation, pronouns, and even first-person constructions with explicit and implicit mentions of gender.
Approach:	They propose a morphologically grounded large-scale benchmark dataset for evaluating gender-aware generation in three typologically diverse grammatically gendered languages: French, Arabic, and Hindi.
Outcome:	The proposed dataset compares 15 popular multilingual large language models on their ability to handle morphological gender and morphology agreement.

Copied to clipboard

Yuxuan Gu, Wuyang Zhou, Giorgos Iacovides, Danilo Mandic

GenDis: Generative-Discriminative Dual-View Co-Training for Generalized Category Discovery (2026.acl-long)

Challenge:	Low-Rank Adaptation (LoRA) methods have significantly reduced the number of trainable parameters needed in fine-tuning large language models.
Approach:	They propose a vector-based random Tensor network for high-Rank Adaptation method that achieves high-rank weight updates while retaining parameter efficiency.
Outcome:	The proposed method outperforms existing PEFT methods while keeping low-rank parameters.

Copied to clipboard

Xi Chen, Chuan Qin, Jinpeng Li, Shasha Hu, Chao Wang, Hengshu Zhu, Hui Xiong

Challenge:	Existing methods rely on one-hot discriminative supervision, leading to overfitting on seen classes and poor generalization to unseen ones.
Approach:	They propose a Generative–Discriminative Dual-View Co-Training framework that unifies discriminative classification and semantic label generation within an LLM.
Outcome:	The proposed framework outperforms existing methods on five benchmarks on the generalized category discovery (GCD) task.

CodeHacker: Automated Test Case Generation for Detecting Vulnerabilities in Competitive Programming Solutions (2026.acl-long)

Copied to clipboard

Jingwei Shi, Xinxiang Yin, Jing Huang, Shengyu Tao, Jinman Zhao

Identifying Bias in Machine-generated Text Detection (2026.acl-long)

Challenge:	Existing benchmarks for Large Language Models often lack coverage for subtle corner cases . a substantial amount of effort has been applied to address this challenge .
Approach:	They propose a framework that generates adversarial test cases that expose latent vulnerabilities in code submissions.
Outcome:	The proposed framework improves the True Negative Rate (TNR) of existing datasets and generates superior adversarial cases on liveCodeBench.

Copied to clipboard

Kevin Stowe, Svetlana Afanaseva, Rodolfo C. Raimundo, Yitao Sun, Kailash Patil

DE-CLIP: Few-Shot Anomaly Detection via Difference-Guided Embedding Editing (2026.acl-long)

Challenge:	a growing number of generative AI systems are detecting text generated by a model or written by . humans perform poorly at the detection task, but show no significant biases on the studied attributes.
Approach:	They examine gender, race/ethnicity, English-language learner status, and economic status . they find several models tend to classify disadvantaged groups as machine-generated .
Outcome:	The proposed models show strong performance but can cause negative impacts . the models classify disadvantaged groups as machine-generated, while economically disadvantaged students' essays are less likely to be classified as machine generated .

Copied to clipboard

Yage Zhang, Yukun Jiang, Michael Backes, Yang Zhang

Culinary Crossroads: A RAG Framework for Enhancing Diversity in Cross-Cultural Recipe Adaptation (2026.acl-long)

Challenge:	Existing approaches to detect anomalies are limited due to the lack of anomalous samples .
Approach:	They propose a framework that edits text embeddings based on the differences between normal and anomalous samples.
Outcome:	The proposed framework achieves 96.6% and 96.99% AUROC on MVTec datasets.

Copied to clipboard

Tianyi Hu, Andrea Morales-Garzón, Jingyi Zheng, Maria Maistro, Daniel Hershcovich

Mixture-of-Minds: Multi-Agent Reinforcement Learning for Table Understanding (2026.acl-long)

Challenge:	Retrieval-augmented generation (RAG) is a promising approach for cross-cultural recipe adaptation, but it fails to generate diverse results even when provided with varied contextual inputs.
Approach:	They propose a plug-and-play RAG framework that enhances diversity in both retrieval and context organization to generate diverse outputs to accommodate multiple user preferences.
Outcome:	The proposed framework achieves Pareto efficiency in terms of diversity and quality of recipe adaptation compared to closed-book LLMs.

Copied to clipboard

Yuhang Zhou, Mingrui Zhang, Ke Li, Mingyi Wang, Qiao Liu, Qifei Wang, Jiayi Liu, Fei Liu, Serena Li, Weiwei LI, Mingze Gao, Abhishek Kumar, Xiangjun Fan, Zhuokai Zhao, Lizhu Zhang

MMCLIP: Cross-Modal Attention Masked Modelling for Medical Language-Image Pre-Training (2026.acl-long)

Challenge:	Large language models (LLMs) have shown promise on understanding and reasoning over tables, but current approaches remain limited.
Approach:	They propose a multi-agent framework that decomposes table reasoning into three specialized roles: planning, coding, and answering.
Outcome:	The proposed framework decomposes table reasoning into three specialized roles: planning, coding, and answering.

Copied to clipboard

Biao Wu, Yutong Xie, Zeyu Zhang, Vu Minh Hieu Phan, Qi Chen, Ling Chen, Qi Wu

MMSearch-R1: Incentivizing LMMs to Search (2026.acl-long)

Challenge:	Existing vision-and-language pretraining methods face challenges in reconstructing pathological features due to limited data.
Approach:	They propose a method that uses masked modeling to enhance visual and linguistic learning.
Outcome:	MMCLIP integrates unpaired data through disease-kind prompts to achieve state-of-the-art performance in zero-shot and fine-tuning across five benchmarks.

Copied to clipboard

Jinming Wu, Zihao Deng, Wei Li, Yiding Liu, Bo You, Bo Li, Zejun MA, Ziwei Liu

Toward Secure Tuning: Mitigating Security Risks from Instruction Fine-Tuning (2026.acl-long)

Challenge:	Existing approaches to deploying large multimodal models rely on rigid pipelines . Existing methods such as retrieval-augmented generation and prompt engineered search rely only on rigid knowledge sources.
Approach:	They propose a framework that enables LMMs to perform on-demand, multi-turn search in real-world Internet environments.
Outcome:	The proposed model outperforms existing models while reducing search calls by over 30%.

Copied to clipboard

Yanrui Du, Fenglei Fan, Sendong Zhao, Jiawei Cao, Ming Ma, Danyang Zhao, Shuren Qi, Ting Liu, Bing Qin

SDE-SQL: Enhancing Text-to-SQL Generation in Large Language Models via Self-Driven Exploration with SQL Probes (2026.acl-long)

Challenge:	Instruction Fine-Tuning (IFT) has emerged as a critical technique for customizing Large Language Models (LLMs) however, recent studies have revealed that IFT can compromise the built-in security mechanisms of LLMs, posing significant security risks.
Approach:	They propose a method that shifts learning burden onto security-robust parameters and propose 'warm-up' phase that preferentially trains Mods_Rob to learn low-level features with minimal security risk.
Outcome:	The proposed method reduces security risks without sacrificing performance gains across knowledge-intensive datasets.

Copied to clipboard

Wenxuan Xie, Yaxun Dai, Wenhao Jiang

Splits! Flexible Sociocultural Linguistic Investigation at Scale (2026.acl-long)

Challenge:	Existing approaches depend on static, pre-processed database information, which restricts the model’s capacity to deeply comprehend the underlying database content.
Approach:	They propose a framework that empowers LLMs to perform Self-Driven Exploration of databases during inference.
Outcome:	Evaluated on the BIRD benchmark with Qwen2.5-72B-Instruct, SDE-SQL achieves an 8.02 % improvement in execution accuracy over the baseline.

Copied to clipboard

Eylon Caplan, Tania Chakraborty, Dan Goldwasser

Challenge:	Variation in language use offers a rich lens into cultural perspectives, values, and opinions.
Approach:	They propose to construct a "sandbox" for systematic and flexible sociolinguistic research by splitting a reddit dataset into demographically/topically split SLPs.
Outcome:	The proposed method analyzes a demographically/topically split Reddit dataset validated by self-identification and replicating several known SLPs from existing literature.

Thinking beyond the anthropomorphic paradigm benefits LLM research (2026.acl-long)

Copied to clipboard

Lujain Ibrahim, Myra Cheng

Challenge:	anthropomorphism is an automatic and unconscious response that occurs even in advanced technical expertise.
Approach:	They argue that anthropomorphism is an automatic and unconscious response . they identify and examine five assumptions that shape research across the LLM development lifecycle .
Outcome:	The proposed framework challenges assumptions that shape research across the LLM development lifecycle and offers promising directions for LLMs.

Coordinating Search-Informed Reasoning and Reasoning-Guided Search in Claim Verification (2026.acl-long)

Copied to clipboard

Qisheng Hu, Quanyu Long, Wenya Wang

Breaking the Generator Barrier: Disentangled Representation for Generalizable AI-Text Detection (2026.acl-long)

Challenge:	Existing methods for multi-hop claim verification require multi-step reasoning to construct verification chains while iterating for information to uncover hidden bridging facts.
Approach:	They propose a hierarchical agent reasoning and information search model that integrates reasoning and search-informed reasoning.
Outcome:	Experimental results show that HARIS improves multi-hop claim verification accuracy and interpretability.

Copied to clipboard

Xiao Pu, Zepeng Cheng, Lin Yuan, Yu Wu, Xiuli Bi

PUPPET: Neural-Symbolic Standardized Patients for Mental Health (2026.acl-long)

Challenge:	a rapid proliferation of large language models (LLMs) generate text that increasingly resembles human writing . this makes it difficult to capture subtle cues that distinguish AI-generated content from human-written content .
Approach:	They propose a framework that disentangles AI-detection semantics from generator-aware artifacts by latent encoding and perturbation-based regularization.
Outcome:	The proposed framework disentangles AI-detection semantics from generator-aware artifacts on 20 representative LLMs across 7 categories.

Copied to clipboard

Chen Xu, Yu ji, Zhenyu Lv, Yang Yi, Yizhe Yang, Luyao Ji, Chaoyi Chen, Xianyang Wang, Tian Lan, Zhihua Wang, Juan Wang, Xunde Dong, Fuze Tian, Qunxi Dong, Bin Hu

LightReasoner: Can Small Language Models Teach Large Language Models Reasoning? (2026.acl-long)

Challenge:	Existing LLM-based training approaches lack faithful responses to clinical errors and explainable feedback.
Approach:	They propose a neural-symbolic virtual standardized patient governed by an OBSERVE-THINK-BEHAVE architecture that embeds LLM reasoning into a symbolic system where experts implant causal associations between intervention logic and patient mental states.
Outcome:	The proposed model outperforms baselines in faithfulness and pedagogical value.

Copied to clipboard

Jingyuan Wang, Yankai Chen, Zhonghang Li, Chao Huang

OneRec-Think: In-Text Reasoning for Generative Recommendation (2026.acl-long)

Challenge:	Large language models (LLMs) have demonstrated remarkable progress in reasoning, but are resource-intensive and require large curated datasets.
Approach:	They propose a framework that leverages the behavioral divergence between a stronger expert model and a weaker amateur model.
Outcome:	The proposed framework improves accuracy by up to 28.1% while reducing time consumption by 90% and tuning token usage by 99%.

Copied to clipboard

Zhanyu Liu, Shiyao Wang, Xingmei Wang, Rongzhou Zhang, Jiaxin Deng, Honghui Bao, Jinghao Zhang, Wuchao Li, PengFei Zheng, Xiangyu Wu, Yifei Hu, Qigen Hu, Xinchen Luo, Lejian Ren, Zhang Zixing, Qianqian Wang, Kuo Cai, Yunfan Wu, Hongtao Cheng, Zexuan Cheng, Lu Ren, Huanjie Wang, Yi Su, Ruiming Tang, Kun Gai, Guorui Zhou

AwarenessBench: Assessing Cognitive Capabilities of Language Models (2026.acl-long)

Challenge:	Existing generative models lack the capacity for explicit and controllable reasoning, a key advantage of LLMs.
Approach:	They propose a framework that integrates dialogue, reasoning, and personalized recommendation.
Outcome:	Experiments across public benchmarks show state-of-the-art performance.

Copied to clipboard

Xiaojian Li, Rongwu Xu, Tianyun Zhang, Yue Wang, Shuo Chen, Qiner Lyu, Briana Zhang, Peiran Yang, Kyle Xue Chen, Haoyuan Shi, Yu Wang, Wei Xu

Beyond Surface-Level Detection: Towards Cognitive-Driven Defense Against Jailbreak Attacks via Meta-Operations Reasoning (2026.acl-long)

Challenge:	Language models exhibit increasingly consciousness-like behaviors, requiring a baseline to evaluate their cognitive abilities.
Approach:	They propose a benchmark to assess the cognitive abilities of language models (LMs) they compare 18 state-of-the-art LMs to human models in metacognition, self-awareness, social awareness and situational awareness .
Outcome:	Evaluating 18 state-of-the-art LMs, they find they consistently surpass baselines . but most models fall short in metacognition and self-awareness, the study finds .

Copied to clipboard

Rui Pu, Chaozhuo Li, Rui Ha, Litian Zhang, Lirong Qiu, Xi Zhang

SCAN: Structured Capability Assessment and Navigation for LLMs (2026.acl-long)

Challenge:	Existing defenses rely on shallow pattern matching, which struggles to generalize to novel and unseen attack strategies.
Approach:	They propose a framework which emulates human cognitive reasoning through a structured reasoning chain.
Outcome:	The proposed framework achieves state-of-the-art performance and exhibits strong generalization to unseen attacks.

Copied to clipboard

Zongqi Wang, Tianle Gu, Chen Gong, Xin Tian, Siqi Bao, Yujiu Yang

EMCEE: Improving Multilingual Capability of LLMs via Bridging Knowledge and Reasoning with Extracted Synthetic Multilingual Context (2026.acl-long)

Challenge:	Existing research has focused on approximating model rankings, but such benchmarks fail to provide users and developers with a comprehensive and fine-grained understanding of a specific model’s capabilities.
Approach:	They propose a framework that enables detailed characterization of LLM capabilities through comprehensive and fine-grained evaluation.
Outcome:	The proposed framework enables detailed characterization of large language models through comprehensive and fine-grained evaluation.

Copied to clipboard

Hamin Koo, Jaehyung Kim

Can LLMs Learn to Map the World from Local Descriptions? (2026.acl-long)

Challenge:	Existing methods emphasize reformulating queries into English, but fail to incorporate language- and culture-specific grounding that is essential for some queries.
Approach:	They propose a framework that extracts query-relevant knowledge from the LLM itself.
Outcome:	The proposed framework outperforms existing approaches on four multilingual benchmarks covering diverse languages and tasks.

Copied to clipboard

Sirui Xia, Aili Chen, Xintao Wang, Tinghui Zhu, Yikai Zhang, Jiangjie Chen, Yanghua Xiao

Challenge:	Recent advances in large language models have demonstrated strong capabilities in tasks such as code generation and mathematical reasoning.
Approach:	They investigate whether large language models can construct coherent global spatial cognition by integrating fragmented relational descriptions.
Outcome:	The proposed models can generalize to unseen spatial relationships and exhibit latent representations aligned with real-world spatial distributions.

LEASH: Adaptive Length Penalty and Reward Shaping for Efficient Large Reasoning Model (2026.acl-long)

Copied to clipboard

Yanhao Li, Lu Ma, Jiaran Zhang, Lexiang Tang, Wentao Zhang, Guibo Luo

ALIGN: Word Association Learning for Cultural Alignment in Large Language Models (2026.acl-long)

Challenge:	Existing approaches to long reasoning traces are hard to tune and fail to adapt to evolving LLMs.
Approach:	They propose a reinforcement learning framework that optimizes the length of reasoning traces by a Lagrangian primal–dual method.
Outcome:	The proposed framework reduces the average reasoning length by 60% across diverse tasks while maintaining competitive performance.

Copied to clipboard

Chunhua Liu, Kabir Manandhar Shrestha, Sukai Huang

RealTalk-CN: A Realistic Chinese Speech Task-Oriented Dialogue Benchmark with Cross-Modal Analysis (2026.acl-long)

Challenge:	Large language models exhibit cultural bias from over-represented viewpoints in training data, yet cultural alignment remains a challenge due to limited cultural knowledge and a lack of exploration into effective learning approaches.
Approach:	They propose a cost-efficient method for fine-tuning large language models on native speakers’ word-association norms and a preference optimization method to improve cultural alignment.
Outcome:	The proposed model trains Llama-3.1-8B and Qwen-2.5-7B on native speakers’ word-association norms and shows that such associations capture cultural knowledge.

Copied to clipboard

Enzhi Wang, Jiaming Zhou, Yuhang Jia, Aobo Kong, Qicheng Li, Yong Qin

HOPE: Hybrid Optimized Parallel Encoding with Supervised and Unsupervised Semantic Fusion for Depression Symptom Detection (2026.acl-long)

Challenge:	Recent advances in speech large language models have enabled end-to-end spoken interactions, but their robustness in real-world applications remains unclear.
Approach:	They propose a multi-turn, multi-domain speech–text TOD dataset for Chinese users . it contains 5.4k dialogues with annotations for dialogue states, disfluency types, speaker characteristics .
Outcome:	The proposed model can be used to evaluate speech large language models in real-world scenarios . the proposed model is based on 5.4k real human-to-human dialogues with annotations .

Copied to clipboard

Tu-Phuong Mai, Minh-Ha H. Le, Duc-Luong Tran, Phuong-Anh Chu, Duy-Cat Can, Hoang-Quynh Le

Bias Fitting to Mitigate Length Bias of Reward Model in RLHF (2026.acl-long)

Challenge:	HOPE is a framework for detecting depression symptoms from social media data . it combines supervised symptom relevance signals with unsupervised intrinsic semantic clustering .
Approach:	They propose a Hybrid Optimized Parallel Encoding framework that combines supervised symptom relevance signals with unsupervised intrinsic semantic clustering.
Outcome:	The proposed framework outperforms existing methods on multiple benchmark datasets and shows that it can detect fine-grained symptoms and early warning of mental health risk.

Copied to clipboard

Kangwen Zhao, Jianfeng Cai, Jinhua Zhu, Ruopei Sun, Dongyun Xue, Wengang Zhou, Li Li, Houqiang Li

Challenge:	Existing approaches to tackling length bias are limited by their complexity or lack of a linear length-reward relation.
Approach:	They propose a framework that learns and corrects underlying bias patterns by fitting a length-reward relationship into a reward model.
Outcome:	The proposed framework improves length-controlled win rate and reduces verbosity without compromising performance.

Rethinking Sample Polarity in Reinforcement Learning with Verifiable Rewards (2026.acl-long)

Copied to clipboard

Xinyu Tang, Yuliang Zhan, Zhixun Li, Xin Zhao, Zhenduo Zhang, Zujie Wen, Zhiqiang Zhang, Jun Zhou

Beyond Examples: Towards Automated Thought-level In-Context Reasoning for Large Language Models (2026.acl-long)

Challenge:	Large reasoning models are typically trained using reinforcement learning with verifiable reward (RLVR) positive and negative self-generated rollouts are used to update the model's policy . positive samples sharpen existing correct reasoning patterns, while negative samples encourage exploration of new reasoning paths.
Approach:	They propose a method that allocates advantage signals to key tokens across different polarities.
Outcome:	The proposed method improves the ability of large reasoning models to learn from their own generated rollouts.

Copied to clipboard

Jinyang Wu, Mingkuan Feng, Shuai Zhang, Feihu Che, Zhengqi Wen, Chonghua Liao, Ling Yang, Haoran Luo, Zheng Lian, Jianhua Tao

Two-Stage Regularization-Based Structured Pruning for LLMs (2026.acl-long)

Challenge:	In-context learning (ICL) struggles with complex reasoning due to superficial, example-level implicit imitation.
Approach:	They propose an automated method that shifts from surface-level examples to more guidance-oriented thought patterns.
Outcome:	The proposed method achieves 80.6% accuracy on MATH and 62.5% on AMC, surpassing GPT-4o’s 77.2% and 57.5% accuracy.

Copied to clipboard

Mingkuan Feng, Jinyang Wu, Siyuan Liu, Shuai Zhang, Hongjian Fang, Ruihan Jin, Feihu Che, Pengpeng Shao, Zhengqi Wen, Jianhua Tao

K-Merge: Online Continual Merging of Adapters for On-device Large Language Models (2026.acl-long)

Challenge:	Structural pruning is a promising solution for large language models . prior structured pruning methods remove unimportant parameters based on certain metrics .
Approach:	They propose a structural pruning method that iteratively learns the weights of transformer layers by adding their l1-norm to the loss function.
Outcome:	The proposed pruning method outperforms strong layer-wise pruning methods without requiring retraining.

Copied to clipboard

Donald Shenaj, Ondrej Bohdal, Taha Ceritli, Mete Ozay, Pietro Zanuttigh, Umberto Michieli

SAC: Neural Speech Codec with Semantic-Acoustic Dual-Stream Quantization (2026.acl-long)

Challenge:	Large Language Models (LLMs) are powerful general-purpose models that can be adapted to a wide range of problem types in many languages.
Approach:	They propose a method for on-device online continual merging to integrate new LoRAs when a new one becomes available.
Outcome:	The proposed approach outperforms other methods while adhering to storage budget constraints.

Copied to clipboard

Wenxi Chen, Ruiqi Yan, Yushen Chen, Zhikang Niu, Ziyang Ma, Xiquan Li, Yuzhe Liang, null Wenhanlin, Shunshun Yin, Ming Tao, Xinsheng Wang, Xie Chen

CriticLean: Critic-Guided Reinforcement Learning for Mathematical Formalization (2026.acl-long)

Challenge:	Existing speech codecs struggle to balance high-quality reconstruction with semantically rich representations, limiting their effectiveness in both generative and understanding tasks.
Approach:	They propose a neural speech codec with semantic-acoustic dual-stream quantization that disentangles semantic and acousian modeling into two dedicated streams.
Outcome:	The proposed codec outperforms state-of-the-art speech tokenizers in auto-propagating text-to-speech models.

Copied to clipboard

Zhongyuan Peng, Yifan Yao, Kaijing Ma, Shuyue Guo, Yizhe Li, Yichi Zhang, Chenchen Zhang, Yifan Zhang, Zhouliang Yu, Luming Li, Minghao Liu, Yihang Xia, Jiawei Shen, Yuchen Wu, Yixin Cao, Zhaoxiang Zhang, Wenhao Huang, Jiaheng Liu, Ge Zhang

TRACE: Evaluating Execution Efficiency of LLM-Based Code Translation (2026.acl-long)

Challenge:	Existing approaches to formalizing mathematical statements face limitations in accuracy, especially in the context of complex, highlevel problems that involve sophisticated mathematical reasoning.
Approach:	They propose a CriticLean framework that elevates the role of the critic from a passive validator to an active learning component and introduce a benchmark to measure models’ ability to distinguish semantically correct from incorrect formalizations.
Outcome:	The proposed framework outperforms open- and closed-source benchmarks and shows that it significantly outperformed existing models.

Copied to clipboard

Zhihao Gong, Zeyu Sun, Dong Huang, Qingyuan Liang, Jie M. Zhang, Dan Hao

One Tokenizer To Rule Them All: Emergent Language Plasticity via Multilingual Tokenizers (2026.acl-long)

Challenge:	Large Language Models (LLMs) have improved the functional correctness of code translation, but execution efficiency remains overlooked.
Approach:	They propose a benchmark to explicitly assess execution efficiency in LLM-translated code.
Outcome:	The proposed benchmark identifies that execution efficiency is an essential dimension of code translation . the results highlight that correctness and efficiency are often misaligned .

Copied to clipboard

Diana Abagyan, Alejandro R. Salamanca, Andres Felipe Cruz-Salinas, Kris Cao, Hangyu Lin, Acyr Locatelli, Marzieh Fadaee, Ahmet Üstün, Sara Hooker

It’s Not What You Say, It’s How You Say It: Evaluating LLM Responses to Expressions of Belief (2026.acl-long)

Challenge:	Existing approaches to train multilingual large language models for many languages at once are limited due to limited model capacity, scarce high-quality data, and compute constraints.
Approach:	They propose to use a universal tokenizer to improve language plasticity and adaptability to new languages by up to 20%.
Outcome:	The proposed tokenizer improves language plasticity and improves plasticity towards languages that are completely unseen in the tokenizer and pretraining, by up to 5% win rate gain.

Copied to clipboard

Kevin Du, Clara Kümpel, Michelle Wastl, Alex Warstadt

CiPO: Counterfactual Unlearning for Large Reasoning Models through Iterative Preference Optimization (2026.acl-long)

Challenge:	a typology is grounded in four linguistically motivated dimensions: form, evidentiality, epistemic stance, and tone.
Approach:	They propose a typology to evaluate how different EoBs affect whether models follow context versus prior knowledge.
Outcome:	The proposed model systematically evaluates 16 LLMs that differ in architecture, scale, and training stages . human listeners subconsciously interpret the belief based on how it is expressed, i.e., its explicitness, tone, or contextual cues.

Copied to clipboard

Junyi Li, Yongqiang Chen, Ningning Ding

DIA-HARM: Dialectal Disparities in Harmful Content Detection Across 50 English Dialects (2026.acl-long)

Challenge:	Existing methods to unlearning large reasoning models do not remove unwanted knowledge from CoT traces or interfere with the reasoning process.
Approach:	They propose a framework that targets the CoT reasoning in Large Reasoning Models by generating a valid counterfactual reasoning trace for preference tuning.
Outcome:	Experiments on large LRMs show that CiPO completely removes knowledge from the intermediate CoT steps and the final answer while preserving the reasoning abilities of LRM.

Copied to clipboard

Jason S Lucas, Matt Murtagh White, Ali Al-Lawati, Uchendu Uchendu, Adaku Uchendu, Dongwon Lee

Towards Scalable Lifelong Knowledge Editing with Selective Knowledge Suppression (2026.acl-long)

Challenge:	Current disinformation detection systems are predominantly developed and evaluated on Standard American English (SAE) . however, their robustness to dialectal variation is unexplored.
Approach:	They propose a benchmark for evaluating disinformation detection robustness across 50 English dialects . they use multi-value's linguistically-grounded transformations to introduce D-CUBE (Dialectal Disinformation Detection Corpus)
Outcome:	The proposed model outperforms zero-shot LLMs in human-written dialects while AI-generated content remains stable.

Copied to clipboard

Dahyun Jung, Jaewook Lee, Heuiseok Lim

Optimizing Length Compression in Large Reasoning Models (2026.acl-long)

Challenge:	Existing methods to modify knowledge are limited due to high training costs and lack stability during sequential edits due to catastrophic forgetting.
Approach:	They propose a framework to modify specific knowledge of large language models without retraining the entire model.
Outcome:	Extensive experiments on ZSRE, Counterfact, and RIPE show that LightEdit outperforms existing lifelong knowledge editing methods.

Copied to clipboard

Zhengxiang Cheng, Dongping Chen, Mingyang Fu, Tianyi Zhou

Bridging Language and Items for Retrieval and Recommendation: Benchmarking LLMs as Semantic Encoders (2026.acl-long)

Challenge:	Large Reasoning Models suffer from producing unnecessary and verbose reasoning chains.
Approach:	They propose a post-training method that uses a Length Reward and a Compress Reward to remove the invalid portion of the thinking process.
Outcome:	The proposed method reduces sequence length by 50% with only a marginal (2%) drop in accuracy.

Copied to clipboard

Yupeng Hou, Jiacheng Li, Xiangjun Fu, Zhankui He, An Yan, Xiusi Chen, Julian McAuley

Evolutionary Guided Decoding: Iterative Value Refinement for LLMs (2026.acl-long)

Challenge:	Recent advances in large language models have enabled their use as semantic encoders for recommendation, but their roles and behaviors in this setting are still not well understood.
Approach:	They propose a benchmark to evaluate large language models as semantic encoders in recommendation scenarios.
Outcome:	The proposed benchmark shows that ranking of 11 leading LLMs is low compared to MTEB, highlighting the unique challenges of semantic encoding in recommendation.

Copied to clipboard

Zhenhua Liu, Lijun Li, Ruizhe Chen, Yuxian Jiang, Tong Zhu, Zhaochen Su, Wenliang Chen, Jing Shao

CRAFT: Training-Free Cascaded Retrieval for Tabular QA (2026.acl-long)

Challenge:	Existing methods for directing language model outputs are limited in their accuracy due to a distributional gap . existing methods train static value functions on trajectories sampled exclusively from the base policy .
Approach:	They propose a framework to bridge a distributional gap in the accuracy of value functions . they propose RLHF to align language models with human values and task requirements .
Outcome:	The proposed framework reduces computational costs and improves value function accuracy by leveraging principled value function optimization.

Copied to clipboard

Adarsh Singh, Kushal Raj Bhandari, Jianxi Gao, Soham Dan, Vivek Gupta

Evaluating Legal Reasoning Traces with Legal Issue Tree Rubrics (2026.acl-long)

Challenge:	Existing methods for open-domain table question answering require retraining or fine-tuning on new datasets.
Approach:	They propose a zero-shot, cascaded retrieval approach that uses a sparse retrieval model to filter a subset of candidates before applying more expensive dense models as re-rankers.
Outcome:	The proposed method outperforms state-of-the-art retrieval models on the NQ-Tables dataset.

Copied to clipboard

Jinu Lee, Kyoung-Woon On, Sophia Simeng Han, Arman Cohan, Julia Hockenmaier

KG-MuLQA: A Framework for KG-based Multi-Level QA Extraction and Long-Context LLM Evaluation (2026.acl-long)

Challenge:	Evaluating the quality of LLM-generated reasoning traces in expert domains is essential for ensuring credibility and explainability, yet remains challenging due to the inherent complexity of such reasoning tasks.
Approach:	They propose a large-scale legal reasoning dataset with an emphasis on reasoning trace evaluation that converts court judgments into hierarchical trees of opposing parties’ arguments and the court’s conclusions.
Outcome:	The proposed model improves the quality of LLM-generated reasoning traces in legal domains, whereas RL improves correctness albeit with reduced coverage.

Copied to clipboard

Nikita Tatarinov, Vidhyakshaya Kannan, Haricharana Srinivasa, Arnav Raj, Harpreet Singh Anand, Varun Singh, Aditya Luthra, Ravij Lade, Agam Shah, Sudheer Chava

COMPASS: Enhancing Agent Long-Horizon Reasoning with Evolving Context (2026.acl-long)

Challenge:	KG-MulQA extracts QA pairs at multiple complexity levels along three key dimensions: multi-hop retrieval, set operations, and answer plurality.
Approach:	They propose a framework that extracts QA pairs at multiple complexity levels along three key dimensions: multi-hop retrieval, set operations, and answer plurality.
Outcome:	The framework extracts QA pairs at multiple complexity levels along key dimensions . it enables fine-grained assessment of model performance across controlled difficulty levels.

Copied to clipboard

Guangya Wan, Mingyang Ling, Xiaoqi Ren, Rujun Han, Sheng Li, Zizhao Zhang

BracketRank: Large Language Model Document Ranking via Reasoning-based Competitive Elimination (2026.acl-long)

Challenge:	Long-horizon tasks that require sustained reasoning and multiple tool interactions remain challenging for LLM agents.
Approach:	They propose a framework that separates tactical execution, strategic oversight, and context organization into three specialized components.
Outcome:	The proposed framework improves accuracy by 20% relative to baselines on GAIA, BrowseComp, and Humanity’s Last Exam tasks.

Copied to clipboard

Abdelrahman Abdallah, Mohammed Ali, Bhawna Piryani, Adam Jatowt

Multi-component Causal Tracing in Large Language Models (2026.acl-long)

Challenge:	Existing lists of document ranking methods lack robust performance across domains.
Approach:	They propose a reasoning-driven competitive elimination framework that optimises group sizes based on LLM context limits and reasoning-enhanced prompts.
Outcome:	The proposed method outperforms RankGPT and other state-of-the-art methods on datasets with a 77.90 NDCG@5 score and 54.66 average NDGC@10 on BEIR datasets.

Copied to clipboard

Zirui Yan, Dennis Wei, Dmitriy A Katz, Prasanna Sattigeri, Ali Tajer

Sparse Feature Coactivation Reveals Causal Semantic Modules in Large Language Models (2026.acl-long)

Challenge:	Large language models (LLMs) are prone to various forms of safety risks, such as learning and propagating societal biases and even creating harmful or deceptive content through jailbreak attacks.
Approach:	They propose a framework for causally tracing multiple components simultaneously that systematically identifies the subsets of components most critical to a desired performance metric.
Outcome:	The proposed method outperforms existing methods in identifying components critical to a desired performance metric.

Copied to clipboard

Ruixuan Deng, Xiaoyang Hu, Miles Gilberti, Shane Storks, Aman Taxali, Mike Angstadt, Chandra Sripada, Joyce Chai

TensorLens: End-to-End Transformer Analysis via High-Order Attention Tensors (2026.acl-long)

Challenge:	Recent work has focused on layerwise interpretations, lacking fine-grained interpretation of specific features and their interaction.
Approach:	They identify semantically coherent, context-consistent network components in large language models . they use sparse autoencoders to coactivate sparsity features from a handful of prompts .
Outcome:	The proposed model can capture concepts and relations more comprehensively than individual features while maintaining specificity.

Copied to clipboard

Ido Andrew Atad, Itamar Zimerman, Shahar Katz, Lior Wolf

Inferring Events from Time Series using Language Models (2026.acl-long)

Challenge:	Existing attention-aggregation methods focus on individual attention heads or layers, failing to account for the model’s global behavior.
Approach:	They propose a unified attention representation that captures the entire transformer as a single, input-dependent linear operator expressed through a high-order attention-interaction tensor.
Outcome:	The proposed model encapsulates the entire transformer as a single, input-dependent linear operator expressed through a high-order attention-interaction tensor.

Copied to clipboard

Mingtian Tan, Mike A Merrill, Zachary Gottesman, Tim Althoff, David Evans, Thomas Hartvigsen

OASIS: Online Sample Selection for Continual Instruction Tuning (2026.acl-long)

Challenge:	Prior work on reasoning about time series in conjunction with natural language has largely overlooked event descriptions and focused on tasks involving just numeric data like trend analysis or anomaly detection.
Approach:	They propose a method for generating tasks that test a model’s ability to reason about events associated with time series data based on sports data and develop a benchmarking method.
Outcome:	The proposed method can infer unobserved events from time series data, even when providing minimal context.

Copied to clipboard

Minjae Lee, Minhyuk Seo, Tingyu Qu, Tinne Tuytelaars, Jonghyun Choi

Challenge:	Existing methods for continual instruction tuning (CIT) use pre-trained reference models, which are impractical in CIT setups since future data are unknown.
Approach:	They propose an adaptive online sample selection approach that estimates each sample’s informativeness relative to all previously seen data and minimizes informative redundancy through iterative selection score updates.
Outcome:	Experiments on various large foundation models show that using only 25% of the data achieves comparable performance to full-data training and outperforms the state-of-the-art sampling methods.

Make Mechanistic Interpretability Auditable: A Call to Develop Guidelines via Continuous Collaborative Reviewing (2026.acl-long)

Copied to clipboard

Michael Lan, Narmeen Fatimah Oozeer, Chaithanya Bandi, Philip Quirke, Austin Meek, Fazl Barez, Amir Abdullah

Mitigating Context Interference for Reliable and Efficient Search Agents (2026.acl-long)

Challenge:	a recent paper found conflicting conclusions for the same behavior in a neural network . authors propose auditing MI itself is essential for its application in AI safety, industry, and governance .
Approach:	They propose to develop a system that can audit experiments to ensure validity . authors propose to generalize good practices found on platform into expert-verified guidelines .
Outcome:	a new review system could be developed that can be standardized and audited . authors argue that auditing MI is essential for its application in AI safety, industry, and governance .

Copied to clipboard

Boyang Xue, Bin Wu, Shuofei Qiao, Sheng Wang, Rui Wang, Yiming Du, Hongru Wang, Jeff Z. Pan, Emine Yilmaz, Kam-Fai Wong, Aldo Lipani

Chunks as Arms: Multi-Armed Bandit-Guided Sampling for Long-Context LLM Preference Optimization (2026.acl-long)

Challenge:	Recent research empowers Large Language Models (LLMs) as multi-turn search agents to iteratively retrieve and generate outputs until complex tasks are solved.
Approach:	They propose a distill-based context refiner to dynamically mitigate context interference . they also propose RLs that refine contexts to generate outputs .
Outcome:	The proposed refiner can mitigate context interference in multi-turn search agents.

Copied to clipboard

Shaohua Duan, Pengcheng Huang, Xinze Li, Zhenghao Liu, Xiaoyuan Yi, Yukun Yan, Shuo Wang, Yu Gu, Ge Yu, Maosong Sun

Long Context Modeling with Ranked Memory-Augmented Retrieval (2026.acl-long)

Challenge:	Recent studies have explored fine-tuning Large Language Models with synthetic data to enhance their long-context capabilities.
Approach:	They propose a framework that leverages a Multi-Armed Bandit rollout strategy to identify the most informative chunks from the given long context for sampling high-quality and diverse responses.
Outcome:	The proposed framework achieves 4% improvement on long-context reasoning benchmarks on Llama and Qwen.

Copied to clipboard

Ghadir Alselwi, Hao Xue, Shoaib Jameel, Basem Suleiman, Flora D. Salim, Imran Razzak

A Comprehensive Survey of Process Reward Models: Data Generation, Model Construction, and Usage (2026.acl-long)

Challenge:	Large Language Models (LLMs) face a fundamental limitation in processing long-context scenarios due to quadratic complexity of attention mechanisms and increasing memory demands during generation.
Approach:	They propose a framework that dynamically ranks memory entries based on relevance . ERMAR employs a relevance scoring mechanism and a pointwise re-ranking model for key-value embeddings .
Outcome:	The proposed framework achieves state-of-the-art performance on benchmarks . it uses historical usage patterns and adaptive retrieval to improve performance .

Copied to clipboard

Congmin Zheng, Jiachen Zhu, Zhuoying Ou, Yuxiang Chen, Kangning Zhang, Rong Shan, Zeyu Zheng, Mengyue Yang, Jianghao Lin, Yong Yu, Weinan Zhang

CODERL+: Improving Code Generation via Reinforcement with Execution Semantics Alignment (2026.acl-long)

Challenge:	Large Language Models (LLMs) have advanced reasoning ability, yet conventional alignment remains dominated by outcome reward models that judge only final answers.
Approach:	They summarize applications across math, code, text, multimodal reasoning, robotics, and agents . goal is to clarify design spaces, reveal open challenges, and guide future research toward fine-grained, robust reasoning alignment.
Outcome:	The proposed model enables finer credit assignment, richer diagnostics, and improved robustness.

Copied to clipboard

Xue Jiang, Yihong Dong, Mengyang Liu, Deng Hongyi, Tian Wang, Yongding Tao, Zhi Jin, Wenpin Jiao, Ge Li

Saber: Efficient Sampling with Adaptive Acceleration and Backtracking Enhanced Remasking for Diffusion Language Model in Code Generation (2026.acl-long)

Challenge:	Large Language Models excel at code generation by learning from vast code corpora, but a fundamental semantic gap remains between training on textual patterns and the goal of functional correctness . reinforcement learning with verifiable rewards (RLVR) approaches are inefficient for establishing a well-aligned connection between the textual representation of code and its execution semantics.
Approach:	They propose a novel approach that integrates execution semantics alignment into the RLVR training pipeline for code generation.
Outcome:	The proposed model outperforms baseline training and RLVR and shows strong applicability across RL and LLMs.

Copied to clipboard

Yihong Dong, Zhaoyu Ma, Xue Jiang, Zhiyuan Fan, Jiaru Qian, Yongmin Li, Jianha Xiao, Zhi Jin, Ge Li

Debiased Orthogonal Boundary-Driven Efficient Noise Mitigation (2026.acl-long)

Challenge:	Diffusion language models (DLMs) offer advantages in parallel generation and bidirectional context modeling, but they face a critical trade-off between inference speed and output quality for tasks with strict structural constraints such as code generation.
Approach:	They propose an efficient sampling algorithm that reduces the number of tokens unmasked per step based on the model’s evolving confidence.
Outcome:	The proposed method improves Pass@1 accuracy by 1.9% while achieving 251.4% inference speedup.

Copied to clipboard

Hao Li, Jiayang Gu, Jingkuan Song, An Zhang, Lianli Gao

Collaborative Chain-of-Agents for Parametric-Retrieved Knowledge Synergy (2026.acl-long)

Challenge:	Existing noise mitigation methods face limitations due to their task-specific design, model dependency, and significant computational overhead.
Approach:	They propose a model-agnostic noise label mitigation paradigm that uses an estimator model and a scoring function to assess the noise level of input pairs.
Outcome:	The proposed method is superior to existing methods in training tasks and tasks.

Copied to clipboard

Yi Jiang, Sendong Zhao, Jianbo Li, Haochun Wang, Lizhe Zhang, Yan Liu, Bing Qin

APPSI-139: A Parallel Corpus of English Application Privacy Policy Summarization and Interpretation (2026.acl-long)

Challenge:	Existing RAG methods focus on external retrieval, while ignoring the rich content of the model.
Approach:	They propose a framework that enhances explicit synergy over parametric and retrieved knowledge by integrating external retrieval components into the input context of the LLMs.
Outcome:	The proposed framework enhances explicit synergy over parametric and retrieved knowledge.

Copied to clipboard

Pengyun Zhu, Qiheng Sun, Long Wen, Yanbo Wang, Yang Cao, Junxu Liu, Deyi Xiong, Jinfei Liu, Zhibo Wang, Kui Ren

Compressing then Matching: An Efficient Pre-training Paradigm for Multimodal Embedding (2026.acl-long)

Challenge:	a lack of high-quality English privacy policy corpus optimized for legal clarity and readability is limiting translation of privacy policies . 139 privacy policies are often considered "incomprehensible" due to technical jargon, legal language, and convoluted grammatical structures.
Approach:	They propose a high-quality English privacy policy corpus annotated by domain experts . they propose APPSI-139 to summarize and interpret privacy policies in English .
Outcome:	The proposed framework outperforms large language models in terms of readability and accuracy.

Copied to clipboard

Da Li, Yuxiao Luo, Keping Bi, Jiafeng Guo, Wei Yuan, Biao Yang, Yan Wang, Fan Yang, Tingting Gao, Guorui Zhou

LLM-XTM: Enhancing Cross-Lingual Topic Models with Large Language Models (2026.acl-long)

Challenge:	Recent approaches demonstrate that MLLMs can be adapted into competitive embedding models via large-scale contrastive learning.
Approach:	They propose a compressed pre-training phase which serves as a warm-up stage for contrastive learning.
Outcome:	The proposed model achieves state-of-the-art among MLLMs of comparable size on the MMEB, realizing optimization in both efficiency and effectiveness.

Copied to clipboard

Minh Chu Xuan, Tien-Phat Nguyen, Linh Ngo Van, Dinh Viet Sang, Nguyen Thi Ngoc Diep, Trung Le

Open Your Model’s Eyes: Video and Context-Aware Multimodal Backchannel Prediction (2026.acl-long)

Challenge:	Existing cross-lingual topic models depend on sparse bilingual resources and often yield incoherent or weakly aligned topics.
Approach:	They propose a framework that integrates LLM-guided topic refinement with self-consistency uncertainty quantification to enable black-box, stable, and scalable enhancement of cross-lingual topic models.
Outcome:	Experiments on multilingual corpora show that the proposed framework achieves superior topic coherence and alignment while reducing reliance on bilingual dictionaries and expensive LLM calls.

Copied to clipboard

Min-Jae Kim, Jun-Yeong Moon, Mujeen Sung, Gyeong-Moon Park

Bringing Real-World Relations into Video Generation with Graph-Structured Knowledge (2026.acl-long)

Challenge:	Existing methods for predicting backchannels rely on audio and text . existing methods omit visual cues and conversational contexts for accurate prediction .
Approach:	They propose a framework that leverages visual cues and conversational contexts to enhance backchannel prediction.
Outcome:	The proposed framework outperforms existing methods and simple multimodal baselines in recognizing complex backchannels such as empathy.

Copied to clipboard

Joonhyung Park, Jaeyun Song, Sihwan Park, Eunho Yang

FC-TTS: Style and Timbre Control in Zero-Shot Text-to-Speech with Disentangled Speech Representations (2026.acl-long)

Challenge:	Existing text-to-video models struggle to accurately simulate real-world physics and dynamic entity interactions.
Approach:	They propose a framework that integrates graph-structured temporal knowledge into video latent diffusion models to enhance compositional generation and interaction fidelity.
Outcome:	The proposed framework enhances compositional generation and interaction fidelity by integrating graph-structured temporal knowledge into video latent diffusion models.

Copied to clipboard

Yoonhyung Lee, Hyunsin Park, Jinhwan Park, Jinkyu Lee

SAGE: Sparse Adaptive Guidance for Dependency-Aware Tabular Data Generation (2026.acl-long)

Challenge:	Recent advances in text-to-speech (TTS) have enabled accurate imitation of reference speech in terms of both speaking style and speaker timbre.
Approach:	They propose a zero-shot text-to-speech framework that enables disentangled control of style and timbre by conditioning on two distinct reference utterances.
Outcome:	The proposed framework achieves high-fidelity synthesis and competitive zero-shot naturalness while supporting consistent and independent manipulation of style and timbre.

Copied to clipboard

Shuo Yang, Zheyu Zhang, Bardh Prenkaj, Gjergji Kasneci

Mask-to-Correct+: Leveraging Retriever Diversity for Masking-guided Faithful Fact Correction (2026.acl-long)

Challenge:	Recent approaches to generate tabular data are limited by their static dependences and lack of fidelity.
Approach:	They propose a novel LLM-based generation framework that enforces sparse and dynamic dependency guidance.
Outcome:	The proposed framework boosts F1 scores by 10% and reduces policy violations by one point.

Copied to clipboard

Payel Santra, Lavisha Sharma, Madhusudan Ghosh, Partha Basuchowdhuri

Native Hybrid Attention for Efficient Sequence Modeling (2026.acl-long)

Challenge:	Existing methods for fact correction ignore semantic faithfulness in their process.
Approach:	They propose a supervised learning approach that uses a diversity-aware masking approach to identify erroneous spans of claims and evaluate the faithfulness of corrections using retrieved evidence.
Outcome:	The proposed framework outperforms baseline frameworks on social media datasets, achieving up to 14% improvement in SARI scores, without using gold evidence.

Copied to clipboard

Jusen Du, Jiaxi Hu, Zhang Tao, Weigao Sun, Yu Cheng

UniversalRAG: Retrieval-Augmented Generation over Corpora of Diverse Modalities and Granularities (2026.acl-long)

Challenge:	Experimental results show that NHA surpasses Transformers on recall-intensive tasks.
Approach:	They propose a hybrid architecture of linear and full attention that integrates both into a unified layer design.
Outcome:	The proposed architecture surpasses Transformers and other hybrid baselines on recall-intensive and commonsense reasoning tasks.

Copied to clipboard

Woongyeong Yeo, Kangsan Kim, Soyeong Jeong, Jinheon Baek, Sung Ju Hwang

Crossing the Reward Bridge: Expanding Reinforcement Learning with Verifiable Rewards Across Diverse Domains (2026.acl-long)

Challenge:	Retrieval-Augmented Generation (RAG) has shown substantial promise in improving factual accuracy by grounding model responses with external knowledge relevant to queries.
Approach:	They propose a framework to retrieve and integrate knowledge from heterogeneous sources with diverse modalities and granularities.
Outcome:	The proposed framework shows superiority over existing methods on 10 benchmarks of multiple modalities.

Copied to clipboard

Yi Su, Dian Yu, Linfeng Song, Juntao Li, Haitao Mi, Zhaopeng Tu, Min Zhang, Dong Yu

Challenge:	Reinforcement learning with verifiable rewards (RLVR) has been effective on structured tasks, but its reliance on simple, rule-based verifiers creates a bottleneck.
Approach:	They propose a framework that uses a generative verifier to provide soft, probabilistic rewards.
Outcome:	The proposed framework outperforms existing models up to 10x their size and can be scalable and effective.

From ID to LLM: Rethinking Representation Learning for Recommendation (2026.acl-long)

Copied to clipboard

Song-Li Wu, Zhaocheng Du, Weinan Gan, Jingyi Wang, Xianquan Wang

DMHM: Density-aware Manifold Learning and Hybrid Mahalanobis Energy for LLMs-generated Text Detection (2026.acl-long)

Challenge:	Recent studies indicate a fundamental incompatibility between ID representations and language model (LM) representations as they capture behavioral and semantic spaces respectively.
Approach:	They propose a Profile-then-Embedding framework for recommendation that integrates semantic user and item profiles and a Personalized Embedded stage to encode these profiles into task-aligned recommendation embeddings.
Outcome:	The proposed framework achieves significant gains across three benchmark datasets, including cold-start and long-tail scenarios.

Copied to clipboard

Tianle Liu, Zhiliang Tian, Zhen Huang, Tianlun Liu, Jingyuan Huang, Zhaoning Zhang, Chengcheng Shao, Dongsheng Li

Dynamic Emotion and Personality Profiling for Multimodal Deception Detection (2026.acl-long)

Challenge:	Existing methods for LGT detection assume that it is a single homogeneous distribution.
Approach:	They propose a framework for LGT detection based on density-aware manifold learning and hybrid Mahalanobis energy.
Outcome:	The proposed framework outperforms baselines in detecting LLM-generated text (LGT) it is based on density-aware manifold learning and hybrid Mahalanobis energy .

Copied to clipboard

Li Zheng, Yanyi Luo, Hao Fei, Yuzhe Ding, Yujie Huang, Fei Li, Chong Teng, Donghong Ji

Can We Predict Before Executing Machine Learning Agents? (2026.acl-long)

Challenge:	Existing methods for deception detection lack sample-level dynamic annotations for emotions and personality.
Approach:	They propose a multi-model multi-prompt annotation scheme and a strict label quality evaluation standard for deception, emotion, and personality annotations.
Outcome:	The proposed framework outperforms state-of-the-art models on the MDPE and DDEP datasets.

Copied to clipboard

Jingsheng Zheng, Jintian Zhang, Yujie Luo, Yuren Mao, Yunjun Gao, Lun Du, Huajun Chen, Ningyu Zhang

Challenge:	Existing approaches to scientific discovery rely on expensive physical execution . a Generate-Execute-Feedback paradigm is costly and slow .
Approach:	They propose to internalize execution priors to substitute costly runtime checks with instantaneous predictive reasoning, drawing inspiration from World Models.
Outcome:	The proposed framework achieves 61.5% accuracy and robust confidence calibration when primed with a Verified Data Analysis Report.

IMPACT: Importance-Aware Activation Space Reconstruction (2026.acl-long)

Copied to clipboard

Md Mokarram Chowdhury, Daniel Agyei Asante, Ernie Chang, Yang Li

Challenge:	Large language models (LLMs) achieve strong performance across domains but remain difficult to deploy in resource-constrained environments due to their massive size.
Approach:	They propose an importance-aware activation reconstruction framework that links compression to its effect on model performance.
Outcome:	Experiments show that IMPACT reduces model size by 55.4% while maintaining accuracy comparable to or better than state-of-the-art models.

SOS-LoRA: Static Orthogonal-Subspace Low-Rank Adaptation with Fixed Multi-Scale Scaling (2026.acl-long)

Copied to clipboard

Yupeng Chang, Yuan Wu, Yi Chang

SDiaReward: Modeling and Benchmarking Spoken Dialogue Rewards with Modality and Colloquialness (2026.acl-long)

Challenge:	Low-Rank Adaptation (LoRA) is a parameter-efficient fine-tuning method for large language models.
Approach:	They propose a drop-in extension that reparameterizes a rank-rtot update as a sum of K static low-rank experts.
Outcome:	Experiments on reasoning and knowledge-intensive benchmarks show consistent gains over matched-budget LoRA.

Copied to clipboard

Jingyu Lu, Yuhan Wang, Fan Zhuo, Xize Cheng, Changhao Pan, Xueyi Pu, Yifu Chen, Chenyuhao Wen, Tianle Liang, Zhou Zhao

Looking at Radiology Report Generation through a Causal Lens: A Survey (2026.acl-long)

Challenge:	SDiaReward is an end-to-end spoken dialogue system that integrates paralinguistic nuances and spontaneous nature of human conversation.
Approach:	They propose an end-to-end multi-turn reward model trained on SDiaReward-Dataset . it is a collection of episode-level preference pairs targeting modality and colloquiality gaps .
Outcome:	The proposed model outperforms general-purpose audio LLMs in episode-level evaluation.

Copied to clipboard

Satyam Kumar, Kaustubh Shivshankar Shejole, Pushpak Bhattacharyya

CTTA-T: Continual Test-Time Adaptation for Text Understanding via Teacher-Student with a Domain-aware and Generalized Teacher (2026.acl-long)

Challenge:	Existing surveys on RRG emphasize deep learning while overlooking the critical role of causality.
Approach:	They propose to analyze biases across the RRG pipeline and formalize it as a causal modeling problem and review representative causal techniques from the literature.
Outcome:	The proposed model can mitigate biases and yield fair, reliable systems with clinically meaningful outputs.

Copied to clipboard

Tianlun Liu, Zhiliang Tian, Zhen Huang, Xingzhi Zhou, Wanlong Yu, Tianle Liu, Feng Liu, Dongsheng Li

Adam’s Law: Textual Frequency Law on Large Language Models (2026.acl-long)

Challenge:	Existing models for text understanding fail to adapt to domain shifts in real-world applications . current models do not improve themselves as they are applied to new domains .
Approach:	They propose a continual test-time adaptation framework that adapts to evolving domains . they propose accumulating domains and a refine-then-filter framework to calibrate teacher predictions .
Outcome:	The proposed model excels in a teacher-student framework adaptable to evolving domains.

Copied to clipboard

Hongyuan Lu, Zixuan Li, Zefan Zhang, Bowen Cao, Wai Lam

Disco-RAG: Discourse-Aware Retrieval-Augmented Generation (2026.acl-long)

Challenge:	Textual frequency is a topic of understudied research, but its relevance to Large Language Models is not well understood.
Approach:	They propose a framework to estimate textual data frequency using a paraphraser and a textual distillation method to refine LLMs.
Outcome:	The proposed framework can be used to estimate sentence-level frequency with word-level frequencies.

Copied to clipboard

Dongqi Liu, Hang Ding, Qiming Feng, Xurong Xie, Zhucun Xue, Chengjie Wang, Jian Li, Jiangning Zhang, Yabiao Wang

Stable Language Guidance for Vision–Language–Action Models (2026.acl-long)

Challenge:	Existing RAG strategies treat retrieved passages in a flat and unstructured way, which prevents the model from capturing structural cues and constrains its ability to synthesize knowledge from dispersed evidence across documents.
Approach:	They propose a framework that explicitly injects discourse signals into the generation process.
Outcome:	Experiments on question answering and long-document summarization benchmarks show the efficacy of the proposed framework.

Copied to clipboard

Zhihao Zhan, Yuhao Chen, Jiaying Zhou, Qinhan Lyu, Hao Liu, Keze Wang, Liang Lin, Guangrun Wang

Controlling Multimodal Conversational Agents with Coverage-Enhanced Latent Actions (2026.acl-long)

Challenge:	Existing vision-Language-Action models are notoriously brittle to linguistic perturbations.
Approach:	They propose a probabilistic framework that disentangles physical affordance from semantic execution.
Outcome:	The proposed framework disentangles physical affordance from semantic execution.

Copied to clipboard

Yongqi Li, Hao Lang, Tieyun Qian, Yongbin Li

Guaranteeing Knowledge Integration with Joint Decoding for Retrieval-Augmented Generation (2026.acl-long)

Challenge:	Recent reinforcement learning (RL) has been widely explored for adapting MCAs to various human-AI interaction scenarios.
Approach:	They propose to use a latent action space for reinforcement learning instead of RL to fine-tune MCAs.
Outcome:	The proposed method outperforms baselines on two conversation tasks with a novel cycle consistency loss.

Copied to clipboard

Zhengyi Zhao, Shubo Zhang, Zezhong Wang, Yuxi Zhang, Huimin Wang, Yutian Zhao, Yefeng Zheng, Binyang Li, Kam-Fai Wong, Xian Wu

Are Emotion and Rhetoric Neurons in LLM? Neuron Recognition and Adaptive Masking for Emotion-Rhetoric Prediction Steering (2026.acl-long)

Challenge:	Retrieval-Augmented Generation (RAG) provides access to external knowledge, but current research focuses on retrieval quality and 'integration bottleneck' .
Approach:	They propose a framework that explicitly decouples reasoning from evidence integration by generating an 'Inner-Answer' and a 'Refer-Aswer" they propose 'a joint decoding mechanism that dynamically fuses the logical coherence of the Inner-Andswer with the factual precision of the Refer-Adswer at the token level'
Outcome:	The proposed framework improves accuracy by 12.1% and reduces hallucinations by 16.3% on five QA benchmarks.

Copied to clipboard

Li Zheng, Xin Zhang, Shuyi He, Fei Li, Chong Teng, Jiang-Ming Yang, Donghong Ji, Zhuang Li

Towards Fine-Grained and Multi-Granular Contrastive Language-Speech Pre-training (2026.acl-long)

Challenge:	Existing studies on neurons focus on emotion and rhetoric, neglecting their intrinsic connections.
Approach:	They propose a framework for fine-grained steering of emotion and rhetoric in large language models . they propose 'neuro-based' masking method that integrates multi-dimensional screening .
Outcome:	The proposed method achieves directed induction of non-target sentences and enhancement of emotion tasks via rhetoric neurons.

Copied to clipboard

Yifan Yang, Bing Han, Hui Wang, Wei Wang, Ziyang Ma, Long Zhou, Zengrui Jin, Guanrou Yang, Tianrui Wang, Xu Tan, Xie Chen

Are We Using the Right Benchmark: An Evaluation Framework for Visual Token Compression Methods (2026.acl-long)

Challenge:	Existing models for fine-grained speaking styles are limited in terms of accuracy, coverage, and naturalness.
Approach:	They propose a model that pre-trains with coarse captions and annotates with a pipeline that grounds captions in audio.
Outcome:	The proposed model outperforms existing models with fine-grained style annotations . it integrates global and fine-granular supervision, enabling unified representations based on the proposed model .

Copied to clipboard

Chenfei Liao, Wensong Wang, Zichen Wen, Xu Zheng, Yiyu Wang, Haocong He, Yuanhuiyi Lyu, Lutao Jiang, Xin Zou, Yuqian Fu, Bin Ren, Linfeng Zhang, Xuming Hu

Retrieval as Generation: A Unified Framework with Self-Triggered Information Planning (2026.acl-long)

Challenge:	Recent efforts to accelerate inference in Multimodal Large Language Models have focused on visual token compression.
Approach:	They propose a framework that leverages downsampling as a discriminator to denoise existing benchmarks.
Outcome:	The proposed evaluation framework leverages downsampling as a discriminator to denoise existing benchmarks.

Copied to clipboard

Bo Li, Mingda Wang, Gexiang Fang, Shikun Zhang, Wei Ye

Between a Rock and a Hard Place: The Tension Between Ethical Reasoning and Safety Alignment in LLMs (2026.acl-long)

Challenge:	Existing models that ground retrieval on external evidence are limited in their ability to implement retrieval-augmented generation.
Approach:	They propose a retrieval-augmented generation model that embeds retrieval control directly into generation.
Outcome:	The proposed model surpasses strong RAG baselines and uses substantially fewer parameters.

Copied to clipboard

Shei Pern Chua, Zhen Leng Thai, Kai Jun Teh, Xiao Li, Qibing Ren, Xiaolin Hu

EventWeave: A Dynamic Framework for Capturing Core and Supporting Events in Dialogue Systems (2026.acl-long)

Challenge:	Large Language Models (LLMs) safety alignment predominantly operates on a binary assumption that requests are either safe or unsafe.
Approach:	They propose a methodology that embeds harmful requests within ethical framings to exploit this vulnerability.
Outcome:	The proposed framework achieves high success rates by exploiting model's own ethical reasoning to frame harmful actions as morally necessary compromises.

Copied to clipboard

Zhengyi Zhao, Shubo Zhang, Yiming Du, Bin Liang, Baojun Wang, Zhongyang Li, Binyang Li, Kam-Fai Wong

Deliberative Searcher: Improving LLM Reliability via Reinforcement Learning with Constraints (2026.acl-long)

Challenge:	Existing dialogue systems process conversational turns in isolation, overlooking event structures that guide natural interactions.
Approach:	They propose a framework that explicitly models relationships between conversational events to generate more contextually appropriate dialogue responses.
Outcome:	Experiments on three dialogue datasets show that the proposed approach produces more natural responses while requiring less computational overhead.

Copied to clipboard

Zhenyun Yin, Shujie Wang, Xuhong Wang, Xingjun Ma, Yingchun Wang

Thermometer of Thoughts: Enhancing LLM’s Exploration via Attention Temperature Modulation (2026.acl-long)

Challenge:	Large language models with search capabilities often exhibit miscalibrated confidence, causing incorrect answers with high certainty.
Approach:	They propose a reasoning-primary framework that integrates search operations into chain-of-thought generation while maintaining explicit confidence calibration.
Outcome:	The proposed framework improves accuracy and reliability of large language models with search capabilities.

Copied to clipboard

Zhiyuan Yu, Shijian Xiao, Cam-Tu Nguyen, Zhangyue Yin, Lekai Xing, Wenzhong Li, Sanglu Lu

CodeFlowBench: A Multi-turn, Iterative Benchmark for Complex Code Generation (2026.acl-long)

Challenge:	Recent advances in the reasoning capabilities of large language models have enabled them to tackle complex tasks such as mathematics reasoning.
Approach:	They propose a method that modulates attention temperature dynamically to shift LLM's internal focus during reasoning, enabling a dynamic shift between exploratory and focused modes.
Outcome:	The proposed method improves Pass@10 by 6.78–14.20% and aggregation accuracy by 9.74% across 7 reasoning benchmarks.

Copied to clipboard

Sizhe Wang, Zhengren Wang, Dongsheng Ma, Yongan Yu, Rui Ling, Zhiyu li, Feiyu Xiong, Wentao Zhang

REFLEX: Self-Refining Explainable Fact-Checking via Verdict-Anchored Style Control (2026.acl-long)

Challenge:	Modern software development demands code that is maintainable, testable, and scalable by organizing the implementation into modular components with iterative reuse of existing codes.
Approach:	They propose a benchmark to evaluate LLMs' ability to perform codeflow by reusing existing functions over multiple turns.
Outcome:	The proposed benchmarks show that LLMs perform significantly worse in multi-turn codeflow scenarios and that their performance inversely correlates with dependency complexity.

Copied to clipboard

Chuyi Kong, Wei Gao, Jing Ma, Hongzhan Lin, Yuxi Sun

Illusions of Confidence? Diagnosing LLM Truthfulness via Neighborhood Consistency (2026.acl-long)

Challenge:	Existing methods for automated fact-checking often overlook deceptive misinformation styles in generated explanations.
Approach:	They propose a framework that explicitly controls reasoning style by anchoring explanations to the predicted verdict.
Outcome:	The proposed framework achieves state-of-the-art under LLaMA-series models with 465 samples.

Copied to clipboard

Haoming Xu, Ningyuan Zhao, Yunzhi Yao, Weihong Xu, Hongru Wang, Xinle Deng, Shumin Deng, Jeff Z. Pan, Huajun Chen, Ningyu Zhang

Challenge:	Existing evaluations rely on point-wise confidence, which can mask brittle belief.
Approach:	They propose a measure of belief robustness that evaluates coherence across a conceptual neighborhood.
Outcome:	The proposed model is more resistant to interference than existing models.

Scaling Beyond Context: A Survey of Multimodal Retrieval-Augmented Generation for Document Understanding (2026.acl-long)

Copied to clipboard

Sensen Gao, Shanshan Zhao, Xu Jiang, Lunhao Duan, Yong Xien Chng, Qing-Guo Chen, Weihua Luo, Kaifu Zhang, Jia-Wang Bian, Mingming Gong

Beyond Literal Mapping: Benchmarking and Improving Non-Literal Translation Evaluation (2026.acl-long)

Challenge:	Document understanding is critical for applications from financial analysis to scientific discovery.
Approach:	They propose a taxonomy based on domain, retrieval modality, and granularity and review advances involving graph structures and agentic frameworks.
Outcome:	The proposed model enables holistic retrieval and reasoning across all modalities, unlocking comprehensive document intelligence.

Copied to clipboard

Yanzhi Tian, Cunxiang Wang, Zeming Liu, Heyan Huang, Wenbo Yu, Dawei Song, Jie Tang, Yuhang Guo

G2RPO-A: Guided Group Relative Policy Optimization with Adaptive Guidance (2026.acl-long)

Challenge:	Large Language Models (LLMs) have advanced machine translation (MT) a meta-evaluation dataset focused on non-literal translations is lacking . experimental results show the inaccuracies of traditional MT metrics and the limitations of LLM-as-a-Judge.
Approach:	They propose a meta-evaluation framework that leverages sub-agents to evaluate machine translation metrics.
Outcome:	The proposed framework improves on the knowledge cutoff and score inconsistency problem.

Copied to clipboard

Yongxin Guo, Wenbo Deng, Zhenglin Cheng, Xiaoying Tang

Efficient Low-Resource Language Adaptation via Multi-Source Dynamic Logit Fusion (2026.acl-long)

Challenge:	Recent advances in reasoning-centric large language models (LLMs) have significantly expanded the performance boundaries of LLMs, showcasing the immense potential of reasoning-enhanced models.
Approach:	They propose an adaptive algorithm that injects ground-truth reasoning steps into roll-out trajectories to compensate for SLMs’ inherent weaknesses.
Outcome:	Experiments on mathematical reasoning and code-generation benchmarks confirm that G2RPO-A substantially outperforms vanilla GRPO.

Copied to clipboard

Chen Zhang, Jiuheng Lin, Zhiyuan Liao, Yansong Feng

Open Schrödinger’s Closed Box: Identifying Retrieval Augmented Generation in API-Accessible Large Language Model Services (2026.acl-long)

Challenge:	Proxy Tuning offers a logit-level strategy for introducing scaling effects, but it often fails in LRL settings because the large model’s weak LRL competence might overwhelm the knowledge of specialized smaller models.
Approach:	They propose a logit-based framework that balances LRL competence from a continually pretrained small model, task competence from high-resource language instruction tuning, and the scaling benefits of large models.
Outcome:	Experiments across four model families and eight LRLs show that TriMix outperforms single-model baselines and Proxy Tuning.

Copied to clipboard

Yukun Jiang, Xinyue Shen, Michael Backes, Zheng Li, Yang Zhang

AFT-Tab: Adversarial Fine-Tuning for Tabular Data Synthesis with Long Text Columns (2026.acl-long)

Challenge:	Large language models (LLMs) are powerful at question-answering but prone to hallucinations due to limited domain-specific or up-to-date knowledge.
Approach:	They propose a framework for IDentifying RAG properties in LLM services that integrates LLMs with retrieval systems and adds an external retriever and knowledge database to mitigate hallucinations.
Outcome:	The proposed framework detects RAG-enhanced LLMs with 99.97% accuracy with partial or no optional knowledge and nearly 100% when the LLM and database are known.

Copied to clipboard

Yuhao Zhang, Liang Yan, Shaoming Duan, Xinyu Zha, Jinhang Su, Peiyi Han, Chuanyi Liu

Revisiting a Pain in the Neck: A Semantic Reasoning Benchmark for Language Models (2026.acl-long)

Challenge:	Existing tabular data synthesis methods fail to account for cross-modal heterogeneity of real-world tables, where structured continuous and discrete attributes coexist with unstructured long-text columns.
Approach:	They propose a framework that synergistically trains an LLM-based text generator and a deep-learning-based non-textual generator to quantify cross-modal semantic alignment.
Outcome:	The proposed framework outperforms state-of-the-art frameworks in fidelity, diversity, and task utility.

Copied to clipboard

Yang Liu, Hongming Li, Melissa Xiaohui Qin, Chao Huang, Qiankun Liu

Timely Machine: Awareness of Time Makes Test-Time Scaling Agentic (2026.acl-long)

Challenge:	Semantic phrases (SP) are lexical combinations whose meanings or usages may not be fully derived from their individual components.
Approach:	They propose to consolidate existing multiword expression resources into a unified testbed to assess language models in semantic phrase processing tasks.
Outcome:	The evaluation suite covers idiomatic expressions, noun compounds, and verbal constructions.

Copied to clipboard

Yichuan Ma, Linyang Li, Yongkang Chen, Peiji Li, Xiaozhe Li, Qipeng Guo, Dahua Lin, Kai Chen

Multi-Granularity Semantic Revision for Large Language Model Distillation (2026.acl-long)

Challenge:	Large language models (LLMs) tackle complex reasoning tasks, but test-time scaling is becoming expensive.
Approach:	They propose to redefine test-time as wall-clock time, where models dynamically adjust strategies based on time budgets.
Outcome:	The proposed model improves time budget awareness and boosts performance across Timely-Eval.

Copied to clipboard

Xiaoyu Liu, Yun Zhang, Wei Li, Simiao Li, Xudong Huang, Hanting Chen, Yehui Tang, Jie Hu, Zhiwei Xiong, Yunhe Wang

RADO: Reasoning Audit-Driven Optimization for Rigorous Reasoning in High-Stakes Domains (2026.acl-long)

Challenge:	Existing methods for generating large language models rely on student-generated outputs, which introduce generation errors and misguide the distillation process.
Approach:	They propose a multi-granularity semantic revision method for LLM distillation that corrects errors using teacher-generated tokens and re-generates the sequence to minimize errors.
Outcome:	The proposed method reduces errors and misguides distillation on student models and improves consistency between teacher and student outputs.

Copied to clipboard

Zhijie Tan, Xu Chu, Guanyu Wang, Ziyu Li, Weiping Li, Tong Mo

Instruction Data Selection via Answer Divergence (2026.acl-long)

Challenge:	Current reinforcement learning paradigms rely on outcome-based rewards, overlooking latent logical fallacies in intermediate steps.
Approach:	They propose a specialized audit model augmented with external tools to identify local logical ruptures and calibrate reward signals.
Outcome:	The proposed framework improves accuracy and logical rigor in high-stakes domains.

Copied to clipboard

Bo Li, Mingda Wang, Shikun Zhang, Wei Ye

Challenge:	Existing methods for instruction tuning use data-centric methods, but they do not explicitly reflect what a particular base model is missing.
Approach:	They propose a method for instruction tuning that uses geometric structure of multi-sample outputs to select instruction data.
Outcome:	The proposed approach outperforms strong selectors on six benchmarks spanning reasoning, knowledge, and coding.

Interpretable Safety Alignment via SAE-Constructed Low-Rank Subspace Adaptation (2026.acl-long)

Copied to clipboard

Dianyun Wang, Qingsen Ma, Yuhu Shang, Zhifeng Lu, Zhenbo Xu, Lechen Ning, Huijia Wu, Zhaofeng He

D2Plan: Dual-Agent Dynamic Global Planning for Complex Retrieval-Augmented Reasoning (2026.acl-long)

Challenge:	Prior work has shown that safety behaviors are governed by low-rank structures . Low-Rank Adaptation (LoRA) consistently underperforms full fine-tuning and reinforcement learning on safety benchmarks .
Approach:	They propose a safety alignment system that disentangles safety-relevant directions into monosemantic features and constructs an interpretable safety subspace from SAE directions.
Outcome:	Empirically, the proposed model achieves 99.6% safety rates across multiple model families and scales . low-rank Adaptation consistently underperforms full fine-tuning and reinforcement learning on safety benchmarks compared with previous methods .

Copied to clipboard

Kangcheng Luo, Tinglang Wu, Yansong Feng

Instructions are all you need: Self-supervised Reinforcement Learning for Instruction Following (2026.acl-long)

Challenge:	Recent advances in reinforcement learning (RL) have empowered Large Language Models (LLMs) with the capability to perform autonomous retrieval during reasoning tasks.
Approach:	They propose a "D2Plan" paradigm for retrieval-augmented reasoning that integrates a 'Reasoner' and a'Purifier'
Outcome:	Experiments show that the proposed paradigm improves on QA benchmarks.

Copied to clipboard

Qingyu Ren, Qianyu He, Powei Chang, Jie Zeng, Zeye Sun, Fei Yu, Jiaqing Liang, Yanghua Xiao

Analytical FFN-to-MoE Restructuring via Activation Pattern Analysis (2026.acl-long)

Challenge:	Existing reinforcement learning approaches suffer from dependency on external supervision and sparse reward signals from multi-constraint tasks.
Approach:	They propose a self-supervised reinforcement learning framework that eliminates dependency on external supervision by deriving reward signals directly from instructions and generating pseudo-labels for reward model training.
Outcome:	The proposed framework achieves strong improvements across 3 in-domain and 5 out-of-domain datasets while maintaining computational efficiency.

Copied to clipboard

Zehua Pei, Hui-Ling Zhen, Lancheng Zou, Xianzhi Yu, Wulong Liu, Sinno Jialin Pan, Mingxuan Yuan, Bei Yu

Towards Order Fairness: Mitigating LLMs Order Sensitivity through Dual Group Advantage Optimization (2026.acl-long)

Challenge:	Large language models (LLMs) are fast but require expensive pre-training . a new approach to scale large language models into MoEs reduces inference costs .
Approach:	They propose an analytical post-training framework that rapidly restructures FFNs into sparse MoE architectures using only a small calibration dataset.
Outcome:	The proposed framework outperforms existing methods on a small calibration dataset.

Copied to clipboard

Xu Chu, Guanyu Wang, Zhijie Tan, Xinrong Chen, Ziyu Li, Tong Mo, Weiping Li

Revitalizing Black-Box Interpretability: Actionable Interpretability for LLMs via Proxy Models (2026.acl-long)

Challenge:	Recent studies attempt to obtain optimal or suboptimal arrangements based on statistical results or using dataset-based search, but these methods increase inference overhead while leaving the model’s inherent order bias unresolved.
Approach:	They propose Dual Group Advantage Optimization (DGAO) which aims to improve model accuracy and order stability simultaneously.
Outcome:	The proposed method improves model accuracy and order stability while penalizing order-sensitive or incorrect responses.

Copied to clipboard

Junhao Liu, Haonan Yu, Zhenyu Yan, Xin Zhang

MedMCP-Calc: Benchmarking LLMs for Realistic Medical Calculator Scenarios via MCP Integration (2026.acl-long)

Challenge:	Applying model-agnostic explanations to Large Language Models is hindered by prohibitive computational costs rendering them dormant for real-world applications.
Approach:	They propose a budget-friendly proxy framework that leverages efficient models to approximate the decision boundaries of expensive Large Language Models.
Outcome:	The proposed framework achieves over 90% fidelity with only 9.5% of the oracle’s cost and is open-source to facilitate future research.

Copied to clipboard

Yakun Zhu, Yutong Huang, Shengqian Qin, Zhongzhen Huang, Shaoting Zhang, Xiaofan Zhang

Challenge:	Existing benchmarks focus on static single-step calculations with explicit instructions.
Approach:	They propose a benchmark for evaluating medical calculators in realistic scenarios . they use 118 scenario tasks across 4 clinical domains to evaluate medical calculator performance .
Outcome:	The first benchmark for evaluating medical calculators in realistic scenarios is released . it features 118 scenario tasks across 4 clinical domains and is based on a model context protocol integration.

Uncovering Temporal Framing in the News (2026.acl-long)

Copied to clipboard

Tarek Mahmoud, Veronika Solopova, Premtim Sahitaj, Ariana Sahitaj, Max Upravitelev, Mervat Abassy, Hana Fatima Shaikh, Neda Foroutan, Vera Schmitt, Preslav Nakov

Simple-VGC: Enhancing Visual Grounding in Multimodal Reasoning via Adaptive Tool Composition (2026.acl-long)

Challenge:	Temporal language is used to structure meaning rather than report chronology in news discourse . a recent study focused on temporal expression extraction and temporal reasoning .
Approach:	They propose a taxonomy of eight temporal frames grounded in prior work on time and framing . they analyze frame prevalence, co-occurrence patterns, and lexical cues from a news corpus .
Outcome:	The proposed taxonomy outperforms zero-shot models at the sentence level . it shows that temporal framing is learnable at the sentences level compared to other methods .

Copied to clipboard

Ye Wang, Qianglong Chen, Siyuan Wang, Zejun Li, Shijie Guo, Zhirui Zhang, Zhongyu Wei

Enhancing Lexical Relation Mining with Structured Sememe Knowledge (2026.acl-long)

Challenge:	Existing multimodal large language models suffer from systematic failures in basic visual understanding.
Approach:	They propose a tool-augmented reasoning framework with three targeted compensation strategies to address these problems.
Outcome:	The proposed framework improves visual grounding by re-injecting the original image to mitigate visual forgetting, the authors show . the proposed framework also improves the accuracy of the visual inputs, the researchers show - and the results are promising .

Copied to clipboard

Hansi Wang, Qiliang Liang, Yue Wang, Yang Liu

Look Less, Reason More: Rollout-Guided Adaptive Pixel-Space Reasoning (2026.acl-long)

Challenge:	Existing top-performing methods for Lexical Relation Mining rely on pre-trained language models yet fail to distinguish nuanced lexical relations.
Approach:	They propose a framework to leverage structured sememe knowledge to enhance LRC and LE.
Outcome:	The proposed method outperforms existing methods on benchmarks and outperformed the LLMs.

Copied to clipboard

Xuchen Li, Xuzhao Li, Jiahui Gao, Renjie Pi, Shiyu Hu, Wentao Zhang

Conditional Memory via Scalable Lookup: A New Axis of Sparsity for Large Language Models (2026.acl-long)

Challenge:	Recent work has shown promise by incorporating pixel-level visual information into the reasoning process, enabling VLMs to access high-resolution visual details during their thought process.
Approach:	They propose a framework that dynamically determines necessary pixel-level operations based on the input query.
Outcome:	The proposed model achieves 73.4% accuracy on HR-Bench 4K while maintaining a tool usage ratio of only 20.1%, improving accuracy and reducing tool usage by 66.5% compared to the previous methods.

Copied to clipboard

Xin Cheng, Wangding Zeng, Damai Dai, Qinyu Chen, Bingxuan Wang, Zhenda Xie, Kezhao Huang, Xingkai Yu, Zhewen Hao, Han Zhang, Yu-Kun Li, Huishuai Zhang, Dongyan Zhao, Wenfeng Liang

PII-Bench: Evaluating Query-Aware Privacy Protection Systems (2026.acl-long)

Challenge:	Mixture-of-Experts (MoE) scales capacity via conditional computation, but lacks knowledge lookup primitive.
Approach:	They propose a conditional memory instantiated via Deep Sparse Embedding (DSE) they propose 'u-shaped scaling law' that identifies optimal balance between MoE experts and DSE memory .
Outcome:	The proposed model outperforms an iso-parameter and isoFLOPs MoE baseline across knowledge and reasoning benchmarks and is infrastructure-efficient.

Copied to clipboard

Hao Shen, Zhouhong Gu, Haokai Hong, Weili Han, Hongfeng Chai

GenProve: Learning to Generate Text with Fine-Grained Provenance (2026.acl-long)

Challenge:	Existing models do not detect PII in user prompts, despite their convenience . current models show significant limitations in determining PI I query relevance .
Approach:	They propose a query-unrelated PII masking strategy and propose PIi-Bench . they propose 'quick-and-easy' PI I masking with a user query and context description .
Outcome:	The proposed model performs well in basic PII detection, but shows significant limitations in query relevance.

Copied to clipboard

Jingxuan Wei, Xingyue Wang, Yanghaoyu Liao, Jie Dong, Yuchen Liu, Caijun Jia, Bihui Yu, Junnan Zhu

Rethinking Meeting Effectiveness: A Benchmark and Framework for Temporal Fine-grained Automatic Meeting Effectiveness Evaluation (2026.acl-long)

Challenge:	Existing methods for large language models (LLMs) are coarse-grained and fail to distinguish between direct quotes and complex reasoning.
Approach:	They propose a framework that combines supervised fine-tuning and group relative policy optimization to generate fluent answers while simultaneously producing sentence-level provenance triples.
Outcome:	The proposed framework outperforms 14 strong large language models in joint evaluation.

Copied to clipboard

Yihang Li, Chenhui Chu

Challenge:	Current methods for evaluating meeting effectiveness rely on post-hoc surveys that yield a single coarse-grained score for an entire meeting.
Approach:	They propose a new paradigm for evaluating meeting effectiveness centered on novel criteria and temporal fine-grained approach . they introduce a meta-evaluation dataset containing 2,459 human-annotated segments from 130 AMI Corpus meetings.
Outcome:	The proposed framework can be used to evaluate meeting effectiveness across different meeting types and from business scenarios to unstructured discussions.

AgentOCR: Reimagining Agent History via Optical Self-Compression (2026.acl-long)

Copied to clipboard

Lang Feng, Fuchao Yang, Feng Chen, Xin Cheng, Haiyang Xu, Zhenglin Wan, Ming Yan, Bo An

MCP-Flow: Facilitating LLM Agents to Master Real-World, Diverse and Scaling MCP Tools (2026.acl-long)

Challenge:	Recent advances in large language models (LLMs) have enabled agentic systems trained with reinforcement learning over multi-turn interaction, but practical deployment is bottlenecked by rapidly growing textual histories that inflate token and memory costs.
Approach:	They propose a framework that represents the accumulated observation-action history as a compact rendered image.
Outcome:	The proposed framework preserves over 95% of text-based agent performance while significantly reducing token consumption (>50%), yielding consistent token and memory efficiency.

Copied to clipboard

WenHao Wang, Peizhi Niu, Zhao Xu, Zhaoyu Chen, Jian Du, Yaxin Du, Xianghe Pang, Keduan Huang, Yanfeng Wang, Qiang Yan, Siheng Chen

Everyone is unique: Towards Behaviorally Heterogeneous Negotiation Dialogue Systems for Debt Collection (2026.acl-long)

Challenge:	Existing research on Large Language Models (LLMs) relies on few servers and lacks training support.
Approach:	They propose a web-agent-driven pipeline for large-scale server discovery, data synthesis, and model training that collects and filters data from 1166 servers and 11536 tools.
Outcome:	Empirical evidence shows that MCP-Flow generates higher quality instruction-function call pairs and higher agentic task performance than previous work.

Copied to clipboard

Yuhang Yang, Kai Tang, Chao Ye, Haobo Wang, Qiqi Luo, Jin Guang Zheng, Zhixin Zhang

Beyond the Crowd: LLM-Augmented Community Notes for Governing Health Misinformation (2026.acl-long)

Challenge:	Existing models that assume users to be static, rational agents with fixed preferences fail to capture rich behavioral heterogeneity in real-world debt collection scenarios.
Approach:	They propose a public persona-enriched debt collection benchmark that highlights behavioral heterogeneity in negotiation.
Outcome:	The proposed benchmark outperforms existing models in realistic scenarios using 16 state-of-the-art LLMs.

Copied to clipboard

Jiaying Wu, Zihang Fu, Haonan Wang, Fanxiao Li, Jiafeng Guo, Preslav Nakov, Min-Yen Kan

BaseCal: Unsupervised Confidence Calibration via Base Model Signals (2026.acl-long)

Challenge:	X (formerly Twitter) users can flag misleading posts, attach contextual notes, and rate the notes’ helpfulness, but there is a significant latency in Community Notes, which is unable to provide accurate notes.
Approach:	They propose a framework that augments Community Notes for faster and more reliable health misinformation governance.
Outcome:	The proposed framework outperforms human contributors in correctness, helpfulness, and evidence utility in health misinformation surges.

Copied to clipboard

Hexiang Tan, Wanli Yang, Junwei Zhang, Xin Chen, Rui Tang, Du Su, Jingang Wang, Yuanzhuo Wang, Fei Sun, Xueqi Cheng

Beyond the Final Actor: Modeling the Dual Roles of Creator and Editor for Fine-Grained LLM-Generated Text Detection (2026.acl-long)

Challenge:	Post-trained LLMs typically compromise reliability with severe overconfidence, resulting in inaccurate responses.
Approach:	They propose a solution that feeds PoLLMs into the base LLM to get confidence.
Outcome:	The proposed solution reduces expected calibration error (ECE) by 42.90% compared to the best unsupervised baselines.

Copied to clipboard

Yang Li, Qiang Sheng, Zhengjia Wang, Yehan Yang, Danding Wang, Juan Cao

PILOT: Planning via Internalized Latent Optimization Trajectories for Large Language Models (2026.acl-long)

Challenge:	Existing methods to detect large language models (LLMs) use binary or ternary classifications, which can only distinguish pure human/LLM text or collaborative text at best.
Approach:	They propose a fine-grained method that characterizes distinct signatures of creator and editor by using Rhetorical Structure Theory to construct a logic graph for creator's foundation and extracting Elementary Discourse Unit (EDU)-level features for the editor's style.
Outcome:	The proposed method outperforms 12 baselines in identifying fine-grained types with low false alarms, offering a policy-aligned solution for LLM regulation.

Copied to clipboard

Haoyu Zheng, Yun Zhu, Yuqian Yuan, Bo Yuan, Wenqiao Zhang, Siliang Tang, Jun Xiao

Miner: Mining Intrinsic Mastery for Data-Efficient RL in Large Reasoning Models (2026.acl-long)

Challenge:	Large Language Models lack the capacity to formulate global strategies due to latency and availability constraints.
Approach:	They propose a framework to internalize the strategic oversight of large models into intrinsic Latent Guidance by synthesizing a query-conditioned Latent Guide.
Outcome:	The proposed framework outperforms strong baselines on mathematical and coding benchmarks with negligible inference latency.

Copied to clipboard

Shuyang Jiang, Yuhao Wang, Ya Zhang, Yanfeng Wang, Yu Wang

CEDAR: A Chinese Evaluation Dataset for Computational Argumentation (2026.acl-long)

Challenge:	Current critic-free RL methods for large reasoning models suffer from severe inefficiency when training on positive homogeneous prompts.
Approach:	They propose a method that repurposes the policy’s intrinsic uncertainty as a self-supervised reward signal, with no external supervision, auxiliary models, or additional inference cost.
Outcome:	Evaluated across six reasoning benchmarks on Qwen3-4B and Qwend3-8B base models, the proposed method achieves state-of-the-art performance among the other four methods.

Copied to clipboard

Tian Lan, Jiang Li, Rong Yan, Feilong Bao, Weihua Wang, Guanglai Gao, Xiangdong Su

MSEarth: A Multimodal Benchmark for Earth Science Phenomenon Discovery with MLLMs (2026.acl-long)

Challenge:	Existing debate datasets neglect important labels for argument mining, generation, and evaluation.
Approach:	They propose a Chinese Evaluation Dataset for Computational Argumentation that includes key arguments and key rhetorical figures, debater roles, modal words, debate results and transcripts.
Outcome:	The proposed dataset covers 600 debates about 318 topics from Chinese debate competitions.

Copied to clipboard

Xiangyu Zhao, Wanghan Xu, Bo Liu, Yuhao Zhou, Fenghua Ling, Ben Fei, Xiaoyu Yue, Lei Bai, Wenlong Zhang, Xiao-Ming Wu

Reasoning Structure Matters for Safety Alignment of Reasoning Models (2026.acl-long)

Challenge:	Existing datasets often rely on synthetic data or figure-caption pairs, failing to capture the depth and complexity of geoscientific reasoning.
Approach:	They propose a multimodal scientific dataset and benchmark curated from open-access publications.
Outcome:	MSEarth features over 289K figures with captions enriched by contextual discussions and reasoning from original papers.

Copied to clipboard

Yeonjun In, Wonjoong Kim, Sangwu Park, Chanyoung Park

Investigating the Representation of Backchannels and Fillers in Fine-tuned Language Models (2026.acl-long)

Challenge:	Large reasoning models (LRMs) achieve strong performance on complex reasoning tasks but often generate harmful responses to malicious user queries.
Approach:	They propose a method that alters the reasoning structure of large reasoning models to achieve effective safety alignment by supervised fine-tuning.
Outcome:	The proposed method is practical and generalizable, requiring no complex training or reward design.

Copied to clipboard

Yu Wang, Leyi Lao, Langchu Huang, Gabriel Skantze, Yang Xu, Hendrik Buschmeier

Retrievals Can Be Detrimental: Unveiling the Backdoor Vulnerability of Retrieval-Augmented Diffusion Models (2026.acl-long)

Challenge:	Backchannels and fillers are important linguistic expressions in dialogue, but often ignored in modern transformer-based language models.
Approach:	They use clustering analysis to learn backchannels and fillers in dialogues in English and Japanese and use natural language generation metrics to confirm this.
Outcome:	The proposed models can learn representations of backchannels and fillers using three fine-tuning strategies.

Copied to clipboard

Hao Fang, Xiaohang Sui, Hongyao Yu, Kuofeng Gao, Jiawei Kong, Sijin Yu, Bin Chen, Shu-Tao Xia

A Theoretically Grounded Approach to Summarizing Conversation Dynamics for Forecasting the Derailment of Online Conversations (2026.acl-long)

Challenge:	Retrieval-augmented diffusion models (RDMs) have been developed to enhance performance with reduced parameters.
Approach:	They propose to integrate retrieval-augmented diffusion models with Retrieval-augmented generation (RAG) that enhances performance with reduced parameters.
Outcome:	The proposed framework achieves outstanding attack effects while maintaining benign utility.

Copied to clipboard

Yingxue Fu, Anaïs Ollagnier

DPDV: Dual-Pathway and Dual-View Representation Learning for Bridging Information Asymmetry in Text-Video Retrieval (2026.acl-long)

Challenge:	Recent work on conversation derailment prediction relies on linguistic features rooted in linguistic and social theories.
Approach:	They propose a system that predicts from the start of a conversation whether it will derail into toxic exchanges.
Outcome:	The proposed system achieves 10% performance increase over baseline and 6.47% increase on benchmark dataset.

Copied to clipboard

Zequn Xie, Xin Liu, Fangming Feng, Boyun Zhang, Tao Jin

Who Wrote This Line? Evaluating the Detection of LLM-Generated Classical Chinese Poetry (2026.acl-long)

Challenge:	Existing methods for text-based person anomaly search fail to address the pose-semantic gap . asymmetric cross-modal information poses a challenge to accurately establishing retrieval relationships .
Approach:	They propose a video retrieval framework that partitions visual features into two categories based on relevance to the text query and performs effective interaction.
Outcome:	The proposed framework achieves leading retrieval performance on five benchmark datasets.

Copied to clipboard

Jiang Li, Tian Lan, Shanshan Wang, Dongxing Zhang, Dianqing Lin, Guanglai Gao, Derek F. Wong, Xiangdong Su

Challenge:	a recent study shows that large language models can generate text, but they can also fabricate large amounts of false or misleading content.
Approach:	They propose a benchmark to detect LLM-generated classical Chinese poetry . they compare 12 different AI detectors to find out whether a poem is authored by AI .
Outcome:	The proposed benchmark compared 12 AI detectors with a dataset of 30,664 Chinese poems . the results highlight the limitations of current Chinese text detectors .

Representation Interventions Enable Lifelong Knowledge Memory Control in LLMs (2026.acl-long)

Copied to clipboard

Xuyuan Liu, Shengyu Chen, Xinshuai Dong, Yanchi Liu, Xujiang Zhao, Haoyu Wang, Yujun Yan, Haifeng Chen, Zhengzhang Chen

SLoRA: Balancing Plasticity and Forgetting in Large Language Models for Continual Learning (2026.acl-long)

Challenge:	Large language models (LLMs) produce outdated or inaccurate content. Updating their knowledge efficiently and accurately without costly retraining is a major challenge.
Approach:	They propose a robust and scalable method that treats knowledge control as interventions within the model’s representation space.
Outcome:	The proposed method achieves fine-grained control over complex, unstructured knowledge while maintaining general utility with frozen base weights.

Copied to clipboard

Lina Yang, Yusheng Liao, Yanfeng Wang, Yu Wang

G-Cap: A Game Character Caption Generator (2026.acl-long)

Challenge:	Large language models (LLMs) have achieved remarkable success across diverse tasks through large-scale pretraining.
Approach:	They propose a framework that filters noisy components from LoRA updates via subspace similarity with the base model.
Outcome:	The proposed framework improves accuracy by 12%, reduces forgetting by 29%, and filters out over 30% of LoRA parameters identified as noisy.

Copied to clipboard

Yang Yang, Feng Hu, Haiming Zhang, XU Cheng, Gui Zheng, Liang Yao, Wenqi Ren

Common to Whom? Regional Cultural Commonsense and LLM Bias in India (2026.acl-long)

Challenge:	Existing studies on Large Vision-Language Models (LVLMs) primarily focus on real-world scenarios, leaving surreal, highly stylized, and semantically hybrid virtual-world situations significantly underexplored.
Approach:	They propose to use a manually annotated benchmark to evaluate LVLMs' ability to perceive and describe game character from the virtual-world.
Outcome:	The proposed task evaluates LVLMs’ ability to perceive and describe game character from the virtual-world.

Copied to clipboard

Sangmitra Madhusudan, Trush Shashank More, Steph Buongiorno, Renata Dividino, Jad Kabbara, Ali Emami

The Stackelberg Speaker: Optimizing Persuasive Communication in Social Deduction Games (2026.acl-long)

Challenge:	Existing cultural commonsense benchmarks treat nations as monolithic, assuming uniform practices within national boundaries.
Approach:	They evaluate eight state-of-the-art LLMs and find two critical gaps . commonsense knowledge is fundamentally long-tailed, with most facts rare in training data .
Outcome:	The proposed model achieves only 13.4%–20.9% accuracy on region-specific questions and exhibits geographic bias over-selecting Central and North India as the "default" while under-representing East and West.

Copied to clipboard

Zhang Zheng, Deheng Ye, Peilin Zhao, Hao Wang

GuideDog: A Real-World Egocentric Multimodal Dataset for Blind and Low-Vision Accessibility-Aware Guidance (2026.acl-long)

Challenge:	Existing approaches focus on information processing and strategy selection, overlooking the significance of persuasive communication in social deduction games.
Approach:	They propose a reinforcement learning framework that trains agents to optimize influential utterances for persuasive impact by formalizing turn-based dialogue as a Stackelberg competition .
Outcome:	The proposed framework outperforms baselines across four social deduction benchmarks and shows that it is effective in persuasive communication.

Copied to clipboard

Junhyeok Kim, Jaewoo Park, Junhee Park, Sangeyl Lee, Jiwan Chung, Jisung Kim, Ji Hoon Joung, Youngjae Yu

LogicPoison: Logical Attacks on Graph Retrieval-Augmented Generation (2026.acl-long)

Challenge:	Recent advances in multimodal large language models (MLLMs) offer new opportunities for higher-level scene understanding, but they require labor-intensive, expert annotation.
Approach:	They propose a dataset that combines 2K human-verified images with 22K image-description pairs to provide a more accurate representation of pedestrian scenes.
Outcome:	The proposed dataset improves scalability while maintaining quality.

Copied to clipboard

Yilin Xiao, Jin Chen, Qinggang Zhang, Yujing Zhang, Chuang Zhou, Longhao Yang, Lingfei Ren, Xin Yang, Xiao Huang

Par-ITA: Benchmarking Seq2Seq and LLMs on a Human-Supervised Parallel Corpus for Italian Hyperpartisan Neutralization (2026.acl-long)

Challenge:	Graph-based Retrieval-Augmented Generation (GraphRAG) enhances the reasoning capabilities of Large Language Models (LLMs) however, traditional RAG attacks are difficult to pose an effective threat to GraphRAg systems.
Approach:	They propose a novel attack framework that targets logical reasoning rather than injecting false contents into GraphRAG systems by grounding their responses in structured knowledge graphs.
Outcome:	The proposed framework outperforms state-of-the-art attacks on GraphRAG systems in both effectiveness and stealth.

Copied to clipboard

Michele Joshua Maggini, Søren Fomsgaard, Michele Maestroni, Gaël Dias, Pablo Gamallo

Incentivizing Parametric Knowledge via Reinforcement Learning with Verifiable Rewards for Cross-Cultural Entity Translation (2026.acl-long)

Challenge:	a new study examines the role of hyperpartisan content in online polarization in the social web.
Approach:	They propose a human-supervised parallel corpus for italian hyperpartisan neutralization of 2,475 paragraph pairs.
Outcome:	The proposed dataset is the first human-supervised parallel corpus for italian hyperpartisan neutralization of 2,475 paragraph pairs.

Copied to clipboard

Jiang Zhou, Xiaohu Zhao, Xinwei Wu, Tianyu Dong, Hao Wang, Yangyang Liu, Heng Liu, Linlong Xu, Longyue Wang, Weihua Luo, Deyi Xiong

Writing-RL: Advancing Long-form Writing via Adaptive Curriculum Reinforcement Learning (2026.acl-long)

Challenge:	Current systems often fall short of this goal in settings where translation hinges on culturally grounded entities such as books, films, places, songs and idioms.
Approach:	They propose a framework that anchors supervision on a verifiable, entity-level reward signal and incorporates lightweight structural gates to stabilize optimization.
Outcome:	The proposed framework improves on XC-Translate and shows that it can learn a robust reasoning process rather than imitating reference translations.

Copied to clipboard

Xuanyu Lei, Chenliang Li, Yuning Wu, Kaiming Liu, Weizhou Shen, Peng Li, Ming Yan, Fei Huang, Ya-Qin Zhang, Yang Liu

Challenge:	Recent advances in Large Language Models (LLMs) have enabled strong performance in long-form writing, but current training paradigms remain limited.
Approach:	They propose an Adaptive Curriculum Reinforcement Learning framework to advance long-form writing capabilities beyond SFT.
Outcome:	Experiments on 7B-scale writer models show that Writing-RL improves long-form writing performance over strong SFT baselines.

Spec-o3: A Tool-Augmented Vision-Language Agent for Rare Celestial Object Candidate Vetting via Automated Spectral Inspection (2026.acl-long)

Copied to clipboard

Minghui Jia, Qichao Zhang, Ali Luo, Linjing Li, Shuo Ye, Hailing Lu, Wen Hou, Dongbin Zhao

ROSE: An Intent-Centered Evaluation Metric for NL2SQL (2026.acl-long)

Challenge:	Spec-o3 is a tool-augmented vision-language agent that performs astronomer-aligned spectral inspection.
Approach:	They propose a tool-augmented vision-language agent that performs astronomer-aligned spectral inspection via interleaved multimodal chain-of-thought reasoning.
Outcome:	Spec-o3 outperforms traditional visual inspection methods on rare-object inspection tasks.

Copied to clipboard

Wenqi Pei, Shizheng Hou, Boyan Li, Chen Han, Zhichao Shi, Yuyu Luo

Compete to Complete: Co-opetition Adversarial Learning for Retrieval-Augmented Generation (2026.acl-long)

Challenge:	Existing evaluation metrics for evaluating the effectiveness of Natural Language to SQL (NL2SQL) solutions are becoming unreliable due to its sensitiveness to syntactic variation and inconsistent consistency with ground-truth SQL.
Approach:	They propose an intent-centered metric that focuses on whether the predicted SQL answers the question, rather than consistency with the ground-truth SQL.
Outcome:	The proposed metric outperforms the next-best metric by nearly 24% on the expert-aligned validation set ROSE-VEC.

Copied to clipboard

Xin Liu, Yu-An Liu, Ruqing Zhang, Yixing Fan, Lixin Su, Jiafeng Guo, Xueqi Cheng

Demystifying Uncertainty in LLMs: Active Calibration between Concepts and Human Evaluations (2026.acl-long)

Challenge:	Existing approaches to reduce hallucination in large language models lack a robust mechanism for generating a generative model.
Approach:	They propose a framework that formulates retriever–generator training in RAG as a minimax game.
Outcome:	The proposed framework improves retrieval-augmented generation performance on seven benchmark datasets.

Copied to clipboard

Pengqi Li, Lizhong Ding, Zhehao Zhou, Chunhui Zhang, Jiarun Fu, Hao Li, Ye Yuan, Guoren Wang

LLM-VA: Resolving the Jailbreak-Overrefusal Trade-off via Vector Alignment (2026.acl-long)

Challenge:	Existing static strategies for mitigating hallucinations do not explicitly model the information gain from interacting with the external environment.
Approach:	They propose a calibration-driven interactive learning strategy that selects clarification queries by optimizing calibration error.
Outcome:	The proposed method provides theoretical guarantees and empirical gains for reliability.

Copied to clipboard

Haonan Zhang, Dongxia Wang, Yi Liu, Kexin Chen, Wenhai Wang

TiKMiX: Efficient Semi-Dynamic Data Mixture via Data Influence for LLM Pre-training (2026.acl-long)

Challenge:	Existing vector steering methods adjust the magnitude of answer vectors, but this creates a fundamental trade-off—reducing jailbreak increases over-refusal.
Approach:	They propose a method which aligns va with vb through closed-form weight updates, making the model’s willingness to respond causally dependent on its safety assessment.
Outcome:	Experiments on 12 LLMs show that the proposed method achieves 11.45% higher F1 than the best baseline while preserving 95.92% utility.

Copied to clipboard

Yifan Wang, null Binbinliu, Fengze Liu, Yuanfan Guo, Jiyao Deng, Xuecheng Wu, Weidong Zhou, Xiaohuan Zhou, Taifeng Wang

Revisit What You See: Revealing Visual Semantics in Vision Tokens to Guide LVLM Decoding (2026.acl-long)

Challenge:	Static data mixing strategies in large language models are often suboptimal as they fail to adapt to the model’s evolving learning states.
Approach:	They propose a semi-dynamic data mixing framework that uses a key observation of influence ranking invariance to reduce computational overhead by 80% .
Outcome:	The proposed method reduces computational overhead by 80% and achieves an average performance gain of 2% across nine downstream benchmarks, effectively mitigating data under-digestion.

Copied to clipboard

Beomsik Cho, Jaehyung Kim

FlexGuard: Continuous Risk Scoring for Strictness-Adaptive LLM Content Moderation (2026.acl-long)

Challenge:	Large Vision–Language Models (LVLMs) integrate visual perception with language understanding, but how vision information contributes to the model’s decoding process remains under-explored.
Approach:	They propose a simple training-free decoding method that guides text generation in Large Vision–Language Models by Referencing Vision Tokens.
Outcome:	The proposed method leverages the semantic information embedded within vision tokens by projecting it into the text token distribution.

Copied to clipboard

Zhihao Ding, Jinming Li, Ze Lu, Jieming Shi

Modeling Human-Like Cognition for Stance Detection: Integrating Intuitive Judgment and Analytical Reasoning (2026.acl-long)

Challenge:	Existing guardrail models for content moderation assume a fixed definition of harmfulness, but enforced strictness varies across platforms and evolves over time, resulting in brittle moderators.
Approach:	They propose a strictness-adaptive LLM moderation benchmark that enables controlled evaluation under multiple strictness regimes.
Outcome:	The proposed moderator performs better under one regime and under another, and is more robust under varying strictness.

Copied to clipboard

Zhaodan Zhang, Jin Zhang, Jiafeng Guo, Xueqi Cheng

FinSight: Towards Real-World Financial Deep Research (2026.acl-long)

Challenge:	Recent advances in large language models (LLMs) have revolutionized stance detection, enabling complex reasoning strategies such as chain-of-thought prompting.
Approach:	They propose Cognitive-Driven Stance Detection (CDSD) that integrates fast intuitive judgment and analytical reasoning enhanced by three key modules: attention-based cognitive alignment to compare system focus, uncertainty-aware belief update using Bayesian inference, and self-doubt-triggered counterfactual reasoning for re-evaluation under low consistency or high uncertainty.
Outcome:	The proposed method outperforms state-of-the-art methods on SEM16, P-Stance, and VAST.

Copied to clipboard

Jiajie Jin, Yuyao Zhang, Yimeng Xu, Hongjin Qian, Yutao Zhu, Zhicheng Dou

Dual-Cluster Memory Agent: Resolving Multi-Paradigm Ambiguity in Optimization Problem Solving (2026.acl-long)

Challenge:	FinSight is the first multi-agent framework for automating end-to-end professional, multimodal financial reports.
Approach:	They propose a code agent with variable memory architecture that unifies data, tools, and agents into a programmable variable space.
Outcome:	The proposed framework outperforms leading deep research systems in factual accuracy, analytical depth, and presentation quality.

Copied to clipboard

Xinyu Zhang, Yuchen Wan, Boxuan Zhang, Zesheng Yang, Lingling Zhang, Bifan Wei, Jun Liu

AfriqueLLM: How Data Mixing and Model Architecture Impact Continued Pre-training for African Languages (2026.acl-long)

Challenge:	Large Language Models (LLMs) struggle with structural ambiguity in optimization problems, where a single problem admits multiple related but conflicting modeling paradigms, hindering effective solution generation.
Approach:	They propose to use memory to leverage historical solutions in a training-free manner to enhance performance by leveraging generalizable guidance knowledge.
Outcome:	The proposed agent achieves an average performance improvement of 11%-21% over previous agents.

Copied to clipboard

Hao Yu, Tianyi Xu, Michael A. Hedderich, Wassim Hamidouche, Syed Waqas Zamir, David Ifeoluwa Adelani

What Does LLM Refinement Actually Improve? A Systematic Study on Document-Level Literary Translation (2026.acl-long)

Challenge:	Continued pretraining (CPT) is a practical route to language adaptation, but improvements on demanding capabilities such as mathematical reasoning are limited.
Approach:	They propose to use CPT to adapt large language models to African languages . they use math, code, and synthetic translated data to analyze their models .
Outcome:	The proposed models improve on multilingual benchmarks and document-level translation.

Copied to clipboard

Shaomu Tan, Dawei Zhu, Ke Tran, Michael Denkowski, Sony Trenous, Leonardo F. R. Ribeiro, Bill Byrne, Felix Hieber

OctoBench: Benchmarking Scaffold-Aware Instruction Following in Repository-Grounded Agentic Coding (2026.acl-long)

Challenge:	Large language models (LLMs) have made document-level machine translation increasingly practical, enabled by long-context modeling and strong generation quality.
Approach:	They propose to use document-level MT followed by segment-level refinement to find the strongest and most stable improvements across six LLMs and seven language pairs.
Outcome:	The proposed method outperforms error-specific prompting and evaluate-then-refine schemes in document-level translation.

Copied to clipboard

Deming Ding, Shichun Liu, Enhui Yang, Jiahang Lin, Ziying Chen, Shihan Dou, Honglin Guo, Weiyu Cheng, Pengyu Zhao, Chengjun Xiao, Qunhong Zeng, Qi Zhang, Xuanjing Huang, Qidi Xu, Tao Gui

Lying with Truths: Open-Channel Multi-Agent Collusion for Belief Manipulation via Generative Montage (2026.acl-long)

Challenge:	coding scaffolds that follow heterogeneous instructions remain under-examined in software engineering . coding models are capable software agents, but their ability to follow constraints remains under-explored .
Approach:	They introduce OctoBench, which benchmarks scaffold-aware instruction following in agentic coding.
Outcome:	The proposed benchmark aims to accelerate the development of more scaffold-aware agents.

Copied to clipboard

Jinwei Hu, Xinmiao Huang, Youcheng Sun, Yi Dong, Xiaowei Huang

EXCEEDS: Extracting Complex Events via Nugget-based Grid Modeling in Scientific Domain (2026.acl-long)

Challenge:	Large language models (LLMs) evolve to autonomous agents synthesizing real-time information, but their reasoning capabilities introduce an unexpected attack surface.
Approach:	They propose a framework that constructs deceptive narratives through adversarial debate and coordinated posting of evidence fragments, causing victims to internalize and propagate fabricated conclusions.
Outcome:	The proposed framework constructs deceptive narratives through adversarial debate and coordinated posting of evidence fragments, causing victims to internalize and propagate fabricated conclusions.

Copied to clipboard

Yi-Fan Lu, Xian-Ling Mao, Bo Wang, Xiao Liu, Heyan Huang

Heterogeneous Adaptive Policy Optimization: Tailoring Optimization to Every Token’s Nature (2026.acl-long)

Challenge:	Extensive event extraction research has been conducted in many domains, including news, finance, and biology.
Approach:	They propose an end-to-end scientific event extraction framework for encoding nuggets into a grid matrix and simplifying complex event extraction as a nuggot-based grid modeling task.
Outcome:	The proposed framework performs well in scientific domain, demonstrating state-of-the-art performance.

Copied to clipboard

Zheng Liu, Mengjie Liu, Siwei Wen, Mengzhang Cai, Bin Cui, Conghui He, Wentao Zhang

Mirroring Users: Towards Building Preference-aligned User Simulator with User Feedback in Recommendation (2026.acl-long)

Challenge:	Existing methods that use entropy as a discrete filter or post-hoc regulator are limited in their ability to optimize for reasoning tasks.
Approach:	They propose a token-aware algorithm that continuously adapts optimization dynamics based on token-level entropy throughout the entire training process.
Outcome:	Extensive experiments on mathematical reasoning, code, and logic tasks across multiple models demonstrate HAPO’s consistent superiority over DAPO.

Copied to clipboard

Tianjun Wei, Huizhong Guo, Yingpeng Du, Zhu Sun, Huang Chen, Dongxia Wang, Jie Zhang

Editing the Moving World: Model Editing for Video LLMs (2026.acl-long)

Challenge:	Large Language Models lack specific task alignment and large-scale simulations are challenging due to their ambiguity, noise and massive volume.
Approach:	They propose a framework that leverages user feedback in RSs with advanced LLM capabilities to generate high-quality simulation data.
Outcome:	The proposed framework boosts the alignment with human preferences and in-domain reasoning capabilities of the fine-tuned LLMs.

Copied to clipboard

Qian Zhang, Xinye Li, Xiaokai Wu, Junhao Xu, Zhanyue Qin, Qingbin Liu, Junxian Cai, Xi Chen, Bolin Zhang, Zhiying Tu, Dianhui Chu, Xiaoyan Yu, Dianbo Sui

Bridging the Sensory Gap: Visual Injection for Taxonomy Completion (2026.acl-long)

Challenge:	Existing models for knowledge editing focus on knowledge-level or static visual domains, overlooking dynamic semantics.
Approach:	They propose a benchmark for modeling large language models using six representative models . they analyze the strengths and limitations of existing models and identify new directions .
Outcome:	The proposed benchmark extends existing models from static modalities to dynamic video scenarios.

Copied to clipboard

Yuhang Niu, Hongyuan Xu, Ciyi Liu, Bofan Wei, Jiaqi Ye, Yanlong Wen, Xiaojie Yuan

OCP: Outlier-Centric Probing for Dynamic Structured Pruning of LLMs (2026.acl-long)

Challenge:	Existing text-only methods suffer from a "Sensory Gap" in integrating new concepts into existing hierarchies.
Approach:	They propose a framework leveraging Visual Injection for Taxonomy Completion that maps synthesized images into intrinsic pseudo-tokens and decouples magnitude from selection to prevent visual signals from being drowned out.
Outcome:	Experiments on three datasets show that VITC achieves state-of-the-art performance . it delivers an average absolute gain of over 19% in Hit@1.

Copied to clipboard

Yang Ji, Ying Sun

Structured Episodic Event Memory (2026.acl-long)

Challenge:	Existing structured pruning methods fail to identify outlier-triggering tokens and uniform layer-wise sparsity misaligns with heterogeneous outlier distributions.
Approach:	They propose a framework that prioritizes capturing outlier-triggering tokens rather than reconstructing full hidden distributions.
Outcome:	Experiments on LLaMA2, LLama3 and OPT show that the proposed framework outperforms state-of-the-art methods and achieves 25% perplexity reduction at 1.6 speedup.

Copied to clipboard

Zhengxuan Lu, Dongfang Li, Yukun Shi, Beilun Wang, Longyue Wang, Baotian Hu

MobileWorld: Benchmarking Autonomous Mobile Agents in Agent-User Interactive and MCP-Augmented Environments (2026.acl-long)

Challenge:	Current approaches to memory in Large Language Models (LLMs) rely on static Retrieval-Augmented Generation (RAG) this lacks the cognitive organization necessary to model the dynamic and associative nature of long-term interaction.
Approach:	They propose a hierarchical framework that transforms interaction streams into structured Episodic Event Frames (EEFs) anchored by precise provenance pointers.
Outcome:	The proposed framework outperforms baseline approaches on LoCoMo and LongMemEval benchmarks.

Copied to clipboard

Quyu Kong, Xu Zhang, Zhenyu Yang, Nolan Gao, Chen Liu, Panrong Tong, Chenglin Cai, Hanzhang Zhou, Jianan Zhang, Liangyu Chen, Zhidan Liu, Steven Hoi, Yue Wang

Measuring Human Contribution in AI-Assisted Content Generation (2026.acl-long)

Challenge:	AndroidWorld is the dominant mobile GUI agent evaluation benchmark, but its success rates are low . despite reproducible emulator environment, it lacks key application categories such as e-commerce and enterprise communication.
Approach:	They propose a benchmark for mobile GUI agents that reflects real-world usage through long-horizon, cross-application workflows.
Outcome:	The proposed framework achieves over 90% success rates, while AndroidWorld is the dominant benchmark.

Copied to clipboard

Yueqi Xie, Tao Qi, Jingwei Yi, Xiyuan Yang, Ryan Whalen, Junming Huang, Qian Ding, Yu Xie, Xing Xie, Fangzhao Wu

AutoTaskEval: Towards Domain-Specific and Fine-Grained Evaluation for LLMs (2026.acl-long)

Challenge:	generative AI has created a new way to generate content with humans . varying degrees of human contribution in content generation poses significant challenges for the delineation of originality .
Approach:	They propose a framework to measure human contribution in AI-assisted content generation by calculating mutual information between human input and AI-aided output relative to self-information of AI-assist output.
Outcome:	The proposed measure discriminates between varying degrees of human contribution across multiple creative domains and is validated in real-world applications.

Copied to clipboard

Qingqing Lyu, Linjuan Wu, Yongliang Shen, Hengwei Liu, Hao Li, Shengpei Jiang, Yin Zhang, Weiming Lu

Challenge:	Existing automated approaches operate within fixed task schemas and often fail to autonomously discover new evaluation dimensions.
Approach:	They propose an automated framework that constructs domain-specific benchmarks directly from unstructured corpora using Bloom’s Taxonomy.
Outcome:	The proposed framework uncovers a broader and more fine-grained task space than expert-curated benchmarks while producing high-quality instances that preserve established model-level evaluation trends.

ACE-Router: Generalizing History-Aware Routing from MCP Tools to the Agent Web (2026.acl-long)

Copied to clipboard

Zhiyuan Yao, Zishan Xu, Yifu Guo, Zhiguang Han, Cheng Yang, Shuo Zhang, Weinan Zhang, Xingshan Zeng, Weiwen Liu

Challenge:	Existing routers that use hardcoded tools are limited by scalability and generality bottlenecks.
Approach:	They propose a pipeline for training history-aware routers to empower precise navigation in large-scale ecosystems.
Outcome:	The proposed pipeline can train routers with dynamic context understanding to create the plug-and-play Light Routing Agent.

CiteGuard: Faithful Citation Attribution for LLMs via Retrieval-Augmented Validation (2026.acl-long)

Copied to clipboard

Yee Man Choi, Xuehang Guo, Yi R. Fung, Qingyun Wang

From Insight to Action: A Novel Framework for Interpretability-Guided Data Selection in Large Language Models (2026.acl-long)

Challenge:	Large Language Models (LLMs) have emerged as powerful assistants for scientific writing, but reliability of LLM alone is in doubt.
Approach:	They propose a retrieval-aware agent framework to provide more faithful grounding for citation validation.
Outcome:	The proposed framework improves over the baseline and achieves 68.1% accuracy on the CiteME benchmark, approaching human performance.

Copied to clipboard

Ling Shi, Xinwei Wu, Xiaohu Zhao, Hao Wang, Heng Liu, Yangyang Liu, Linlong Xu, Longyue Wang, Deyi Xiong, Weihua Luo

Unlocking the Potential of Diffusion Language Models through Template Infilling (2026.acl-long)

Challenge:	Recent research in mechanistic interpretability has revealed that Large Language models contain disentangled, human-understandable components.
Approach:	They propose a framework that first identifies causal task features through frequency recall and interventional filtering, then selects “Feature-Resonant Data” that maximally activates task features for fine-tuning.
Outcome:	The proposed framework outperforms existing models on mathematical reasoning, summarization, and translation tasks while using only 50% of the data.

Copied to clipboard

Junhoo Lee, Seungyeon Kim, Nojun Kwak

UniSpec: Training-Free Speculative Decoding for Robust LLM Acceleration Across Languages and Hardware (2026.acl-long)

Challenge:	Existing methods rely on prefix-based prompting, resulting in a lack of stability and a large computational time.
Approach:	They propose a conditioning methodology tailored for Diffusion Language Models that distributes structural anchors across the target response, establishing a global template before infilling masked segments.
Outcome:	The proposed method improves on mathematical reasoning, code generation, and trip planning benchmarks while maintaining speed and robustness.

Copied to clipboard

Truong Dinh Do, Nguyen-Khang Le, Le-Minh Nguyen

Logical Consistency as a Bridge: Improving LLM Hallucination Detection via Label Constraint Modeling between Responses and Self-Judgments (2026.acl-long)

Challenge:	Existing methods for speculative decoding ignore device-specific verification costs and lack of mechanisms to assess draft token quality.
Approach:	They propose a training-free, lossless speculative decoding framework that enables robust, plug-and-play LLM acceleration across diverse hardware configurations and languages.
Outcome:	The proposed framework outperforms existing training-free methods while maintaining identical output quality across different hardware environments.

Copied to clipboard

Hao Mi, Qiang Sheng, Shaofei Wang, Beizhe Hu, Yifan Sun, Zhengjia Wang, Hengqi Zeng, Yang Li, Danding Wang, Juan Cao

UNIKIE-BENCH: Benchmarking Large Multimodal Models for Key Information Extraction in Visual Documents (2026.acl-long)

Challenge:	Existing methods for hallucination detection focus on implicit neural uncertainty or explicit symbolic reasoning, ignoring factual hallucinosities.
Approach:	They propose a framework that bridges neural features and symbolic judgments for hallucination detection by leveraging a "meta-judgment" process to map symbolic labels back into the feature space.
Outcome:	Extensive experiments on 4 public datasets, across 4 LLMs, against 8 baselines demonstrate the superiority of LaaB.

Copied to clipboard

Yifan Ji, Zhipeng Xu, Zhenghao Liu, Zulong Chen, Qian Zhang, Zhibo Yang, Junyang Lin, Yu Gu, Ge Yu, Maosong Sun

FS-Researcher: Test-Time Scaling for Long-Horizon Research Tasks with File-System-Based Agents (2026.acl-long)

Challenge:	Recent Large Multimodal Models (LMMs) have shown promising potential for performing end-to-end KIE directly from document images.
Approach:	They propose a benchmark to evaluate the performance of Large Multimodal Models (LMMs) using a constrained-category KIE track and an open-categorical KIE Track.
Outcome:	Experiments on 15 state-of-the-art LMMs show performance degradation under diverse schema definitions, long-tail key fields, and complex layouts, along with pronounced performance disparities across different document types and scenarios.

Copied to clipboard

Chiwei Zhu, Benfeng Xu, Mingxuan Du, Shaohan Wang, Xiaorui Wang, Zhendong Mao, Yongdong Zhang

Annotating Dimensions of Social Perception in Text: A Sentence-Level Dataset of Warmth and Competence (2026.acl-long)

Challenge:	Long trajectories in deep research often exceed model context limits, compressing token budgets for both evidence collection and report writing.
Approach:	They propose a file-system-based framework that scales deep research beyond context window . a Context Builder agent acts as a librarian and a Report Writer agent composes the final report .
Outcome:	Experiments on two open-ended benchmarks show that FS-Researcher achieves state-of-the-art report quality across different backbone models.

Copied to clipboard

Mutaz Ayesh, Saif M. Mohammad, Nedjma Ousidhoum

SCOUT: Selective Coupling via Optimal Unbalanced Transport for Interpretable Text Classification (2026.acl-long)

Challenge:	Warmth (W) and *Competence (C) are central dimensions along which people evaluate individuals and social groups.
Approach:	They propose a first sentence-level dataset annotated for warmth and competence . they analyze sentences that express attitudes and opinions about individuals or social groups .
Outcome:	The first sentence-level dataset annotated for warmth and competence is presented in this paper.

Copied to clipboard

Junhao Jia, Hanwen Zheng, Yueyi Wu, Huangwei Chen, Haishuai Wang, Jiajun Bu, Lei Wu

LLMs Enable Bag-of-Texts Representations for Short-Text Clustering (2026.acl-long)

Challenge:	Standard interpretable models often rely on scalar similarities that obscure the true evidentiary basis of a prediction.
Approach:	They propose a new paradigm that grounds prototype reasoning in the selective correspondence of discriminative fragments.
Outcome:	The proposed model outperforms rationale extraction and post-hoc attribution methods on seven benchmarks.

Copied to clipboard

I-Fan Lin, Faegheh Hasibi, Suzan Verberne

TriEx: A Game-based Tri-View Framework for Explaining Internal Reasoning in Multi-Agent LLMs (2026.acl-long)

Challenge:	Existing methods for short text clustering require labeling and no embeddings optimization.
Approach:	They propose a training-free method for unsupervised short text clustering that relies less on careful selection of embedders than other methods.
Outcome:	The proposed method achieves comparable or superior results to state-of-the-art methods, but without embeddings optimization or prior knowledge of clusters or labels.

Copied to clipboard

Ziyi Wang, Chen Zhang, Wenjun Peng, Qi Wu, Xinyu Wang

What’s Left Unsaid? Detecting and Correcting Misleading Omissions in Multimodal News Previews (2026.acl-long)

Challenge:	Existing explainability methods for large language models have been limited in capturing interaction-dependent belief dynamics and multi-agent reasoning.
Approach:	They propose a tri-view explainability framework that instruments sequential decision making with aligned artifacts.
Outcome:	The proposed framework enables analysis of explanation faithfulness, belief dynamics, and evaluator reliability, revealing systematic mismatches between what agents say, what they believe, and what they do.

Copied to clipboard

Fanxiao Li, Jiaying Wu, Tingchao Fu, Dayang Li, Herun Wan, Wei Zhou, Min-Yen Kan

ContextLens: Modeling Imperfect Privacy and Safety Context for Legal Compliance (2026.acl-long)

Challenge:	Existing efforts to detect factually incorrect content are omitted by creators who subtly reshape impressions by omitting crucial background context.
Approach:	They propose a multi-stage pipeline that simulates preview-based and context-based understanding and a OMGuard pipeline that combines interpretation-aware fine-tuning and rationale-guided misleading content correction.
Outcome:	The proposed framework lifts an 8B model’s detection accuracy to the level of a 235B LVLM while delivering stronger end-to-end correction.

Copied to clipboard

Haoran Li, Yulin Chen, Huihao Jing, Wenbin Hu, Tsz Ho Li, Chanhou Lou, Hong Ting Tsang, Sirui Han, Yangqiu Song

StealthGraph: Exposing Domain-Specific Risks in LLMs through Knowledge-Graph-Guided Harmful Prompt Generation (2026.acl-long)

Challenge:	Existing approaches to contextualize safety and privacy assessments assume the availability of complete and clear context, whereas real-world contexts tend to be ambiguous and incomplete.
Approach:	They propose a semi-rule-based framework that leverages large language models to ground the input context in the legal domain and explicitly identify both known and unknown factors for legal compliance.
Outcome:	The proposed framework can significantly improve existing baselines without training and can identify the ambiguous and missing factors.

Copied to clipboard

Huawei Zheng, Xinqi Jiang, Sen Yang, Shouling Ji, Yingcai Wu, Dazhen Deng

Efficient Multi-Agent System Training with Data Influence-Oriented Tree Search (2026.acl-long)

Challenge:	Domain-specific datasets of harmful prompts are scarce and often rely on manual construction. Existing efforts to improve domain knowledge and reduce harmful prompt generation are lacking.
Approach:	They propose a framework that transforms domain knowledge into actionable constraints and increases the implicitness of generated harmful prompts.
Outcome:	The proposed framework yields high-quality datasets combining strong domain relevance with implicitness, enabling more realistic red-teaming and advancing LLM safety research.

Copied to clipboard

Wentao Shi, Zichun Yu, Fuli Feng, Xiangnan He, Chenyan Xiong

Integrating Data Validation with Large Language Models for Regulation-Guided Tabular Anomaly Detection (2026.acl-long)

Challenge:	Large Language Model (LLM) based multi-agent systems (MAS) have high potential for tackling complex tasks through collaborative intelligence.
Approach:	They propose a framework that incorporates influence scores to guide tree search and data selection in data synthesis.
Outcome:	The proposed framework incorporates influence scores to guide tree search and data selection in data synthesis.

Copied to clipboard

Haoliang Huang, Zihuang Cai, Zhuo Tang, Yifan Liu, Chen Tian, Kenli Li, Changjian Chen

Efficient Self-Evaluation for Diffusion Language Models via Sequence Regeneration (2026.acl-long)

Challenge:	Existing tabular anomaly detection methods focus on detecting anomalies based on data distribution without considering regulatory compliance.
Approach:	They propose a task that leverages regulations to detect anomalies in tabular data . they also develop three new datasets to address this task .
Outcome:	The proposed method outperforms baselines on three new datasets.

Copied to clipboard

Linhao Zhong, Linyu Wu, Wen Wang, Yuling Xi, Chenchen Jing, Jiaheng Zhang, Hao Chen, Chunhua Shen

DeReA: Improving Idiom Translation with Detect-Retrieve-Arbitrate Reasoning (2026.acl-long)

Challenge:	Non-sequential and bidirectional nature of diffusion large language models makes direct likelihood-based self-evaluation challenging.
Approach:	They propose a self-evaluation confidence quantification method for diffusion large language models that quantifies confidence by computing the probability of regenerating tokens in the entire generated sequence, given the full context.
Outcome:	The proposed method is correlated with semantic coherence and answer accuracy.

Copied to clipboard

Rongqing Jiang, Xuebo Liu, Shengxin Liu, Yutong Wang, Min Zhang, Shimin Tao, Daimeng Wei, Min Zhang

Faithful Serum: Mitigating the Faithfulness Gap in Textual Explanations of LLM Decisions via Attribution Guidance (2026.acl-long)

Challenge:	Existing approaches to idiom translation are limited by the constraints of static parametric memory and retrieval noise . idiomatic expressions are non-compositional units where figurative meanings diverge from literal interpretations .
Approach:	They propose a detect-retrieve-arbitrate framework that detects idiomatic spans by reasoning over semantic conflicts between literal and contextual meanings.
Outcome:	The proposed framework improves GPT-5-mini and Emerging Slang datasets on various model scales.

Copied to clipboard

Bar Alon, Itamar Zimerman, Lior Wolf

Visually-Guided Policy Optimization for Multimodal Reasoning (2026.acl-long)

Challenge:	Prior work has focused on generating convincing rationales that appear to be subjectively faithful, but it remains unclear whether these explanations are epistemic faithful.
Approach:	They propose a method that enhances epistemic faithfulness by guiding explanation generation through attention-level interventions, informed by token-level heatmaps.
Outcome:	The proposed method significantly improves epistemic faithfulness across multiple models, benchmarks, and prompts.

Copied to clipboard

Zengbin Wang, Feng Xiong, Liang Lin, Xuecai Hu, Yong Wang, Yanlin Wang, Man Zhang, Xiangxiang Chu

Don’t Be Misled by Style: A Style-Adaptive Reranker for Capturing Effective Knowledge in Retrieval-Augmented Generation (2026.acl-long)

Challenge:	Existing RLVRs lack visual faithfulness due to text-dominated reasoning . a novel framework to reinforce visual focus during policy optimization is proposed .
Approach:	They propose a framework to reinforce visual focus during policy optimization using visual attention compensation mechanism.
Outcome:	The proposed framework exhibits better visual activation and superior performance in multimodal reasoning and visual-dependent tasks.

Copied to clipboard

Ruwen Zhang, Bo Liu, Zhang Sheng Xiang, Yida Chen, Hantao Zhao, Ding Ding, Jiahui Jin, Jiuxin Cao

Text2Tabular – Reconstructing Tabular Research Data from Scientific Publications (2026.acl-long)

Challenge:	Existing rerankers are mainly trained on well-edited texts, but stylistic features can be misled by reranked models.
Approach:	They propose a style-augmented multi-task framework that prioritizes effective knowledge over stylistic perturbations by using an LLM to derive passage-level supervision on whether a passage helps or harms answer correctness.
Outcome:	Extensive experiments show that SARK improves generation performance across multiple LLMs under mixed-style conditions.

Copied to clipboard

Jonas Gottal, Florian Matthes

From "Aha Moments" to Controllable Thinking: Toward Meta-Cognitive Reasoning in LRMs via Decoupled Reasoning and Control (2026.acl-long)

Challenge:	Text2Tabular reconstructs research datasets from scientific literature using advanced natural language processing and statistical modeling.
Approach:	Text2Tabular reconstructs research datasets from scientific publications using natural language processing and statistical modeling.
Outcome:	Text2Tabular reconstructs scientific literature-based datasets using natural language processing and statistical modeling.

Copied to clipboard

Rui Ha, Rui Pu, Chaozhuo Li, Li Sun, Sen Su

Calibrating Inference Time Alignment with Sequence-level Risk Accumulation (2026.acl-long)

Challenge:	Large Reasoning Models exhibit step-by-step reasoning, reflection, and backtracking, but these behaviors are often unregulated, leading to overthinking.
Approach:	They propose a meta-cognitive reasoning framework that decouples reasoning from control to enable independent optimization of control strategies.
Outcome:	Experiments show that the proposed model improves efficiency and accuracy across reasoning benchmarks.

Copied to clipboard

Shanwen Tan, Ziyang Dong, Wei Ju, Yiwei Fu, Hao Wu, Kun Wang, Yifan Wang, Ziyue Qiao

Diff4TST: Masked Diffusion Language Model for Text Style Transfer (2026.acl-long)

Challenge:	Existing approaches to decode large language models (LLMs) often over-reject benign information, limiting their generalizability in real-world scenarios where harmful and benign information coexist.
Approach:	They propose a framework to regulate decoding alignments for Large Language Models (LLMs) they employ a reward-guided branch decoding paradigm to incorporate safety awareness during generation.
Outcome:	The proposed framework achieves superior performance on four attack benchmarks and two neutral datasets.

Copied to clipboard

Xinchen Ma, Gaole He, Yunshi Lan, Weining Qian

In-depth Research Impact Summarization through Fine-Grained Temporal Citation Analysis (2026.acl-long)

Challenge:	Existing methods for text style transfer rely on task-specific training and expensive training stages.
Approach:	They propose a diffusion-based language model that formulates text style transfer as an explicit copy-and-edit process.
Outcome:	The proposed model improves style accuracy and controllability while maintaining strong content preservation and fluency.

Copied to clipboard

Hiba Arnaout, Noy Sternlicht, Tom Hope, Iryna Gurevych

QuantileMark: A Message-Symmetric Multi-bit Watermark for LLMs (2026.acl-long)

Challenge:	citation counts are a shallow view that fails to capture how a paper has influenced subsequent work.
Approach:	They propose a task to generate nuanced, expressive, and time-aware impact summaries . they analyze fine-grained confirmatory and correction citation intents to generate summary .
Outcome:	The proposed task shows moderate to strong human correlation on subjective metrics such as insightfulness.

Copied to clipboard

Junlin Zhu, Baizhou Huang, Xiaojun Wan

Standard-to-Dialect Transfer Trends Differ across Text and Speech: A Case Study on Intent and Topic Classification in German Dialects (2026.acl-long)

Challenge:	a number of large language models (LLMs) require multi-bit watermarking to ensure provenance.
Approach:	They propose a multi-bit watermark that embeds messages within a continuous cumulative probability interval.
Outcome:	The proposed watermark breaks message symmetry in low-entropy decoding, showing it can be used for verification and quality verification.

Copied to clipboard

Verena Blaschke, Miriam Winkler, Barbara Plank

Don’t Click That: Teaching Web Agents to Resist Deceptive Interfaces (2026.acl-long)

Challenge:	Research on cross-dialectal transfer from a standard to a non-standard dialect variety has typically focused on text data.
Approach:	They compare standard-to-dialect transfer in three settings: text models, speech models, and cascaded systems where speech first gets automatically transcribed and then further processed by a text model.
Outcome:	The proposed model performs best on German dialect data while the text-only model perform best on the standard data.

Copied to clipboard

Yilin Zhang, Yingkai Hua, Chunyu Wei, Xin Wang, Yueguo Chen

Challenge:	Existing approaches to deception detection and defenses are inadequate . Existing methods do not integrate with agent decision-making .
Approach:	They propose a framework that integrates hybrid-reward learning with asymmetric penalties and experience summarization to distill failure patterns into transferable guidance.
Outcome:	The proposed framework reduces deception susceptibility by 53.8% while maintaining task performance, establishing an effective foundation for robust web agent deployment.

Empirical Analysis of Decoding Biases in Masked Diffusion Models (2026.acl-long)

Copied to clipboard

Pengcheng Huang, Tianming Liu, Zhenghao Liu, Yukun Yan, Shuo Wang, Tong Xiao, Zulong Chen, Maosong Sun

LAFaCT: Attribution-based Localization and Focused Sequential Analysis of Fact-Critical Tokens for Hallucination Detection (2026.acl-long)

Challenge:	Existing MDMs employ uncertainty-based decoding strategies that limit their reasoning ability and ultimately degrade generation quality.
Approach:	They propose a framework that regularizes uncertainty-based decoding by incorporating two complementary priors to shape global decoding trajectories and promote content informativeness.
Outcome:	The proposed framework outperforms existing decoding strategies by more than 7% while achieving comparable performance to autoregressive models of similar parameter scales.

Copied to clipboard

Xin Wang, Jiahao Li, Licheng Zhang, Zhendong Mao

DPC: Training-Free Text-to-SQL Candidate Selection via Dual-Paradigm Consistency (2026.acl-long)

Challenge:	Large Language Models suffer from hallucinations, severely undermining their reliability.
Approach:	They propose a framework that localizes fact-critical tokens and performs sequential analysis on their hidden states.
Outcome:	The proposed framework localizes fact-critical tokens using Factual Criticality . it then performs a focused sequential analysis on their hidden states .

Copied to clipboard

Boyan Li, Ou Ocean Kun Hei, Yue Yu, Yuyu Luo

Metaphor Reasoning is Meta-reasoning (2026.acl-long)

Challenge:	Existing methods for generating SQL queries lack the ability to self-evaluate correctness without an execution oracle.
Approach:	They propose a framework that reformulates SQL selection from a probabilistic guessing task on hidden data into a deterministic verification task on visible data.
Outcome:	Experiments on BIRD and Spider show that the proposed method outperforms baselines.

Copied to clipboard

Qianyu He, Junting Lu, Yikai Zhang, Siyu Yuan, Xiaojun Meng, Jiansheng Wei, Jiaqing Liang, Yanghua Xiao

Beyond Hard Masks: Progressive Token Evolution for Diffusion Language Models (2026.acl-long)

Challenge:	Existing work on metaphor reasoning's impact on reasoning abilities is limited.
Approach:	They propose a system for synthesizing metaphorical riddles that satisfy five quality dimensions: diverse, balanced, reasoning-oriented, challenging, and verifiable.
Outcome:	The proposed system improves reasoning abilities across six domains using only thousands of metaphorical riddles.

Copied to clipboard

Linhao Zhong, Linyu Wu, Bozhen Fang, Tianjian Feng, Chenchen Jing, Wen Wang, Jiaheng Zhang, Hao Chen, Chunhua Shen

MAS-Bench: A Unified Benchmark for Shortcut-Augmented Hybrid Mobile GUI Agents (2026.acl-long)

Challenge:	Existing Diffusion Language Models rely on hard binary masking and discrete token assignments, which hinder the revision of early decisions.
Approach:	They propose a diffusion-based language modeling approach that replaces hard binary masks with evolving soft token distributions.
Outcome:	The proposed approach outperforms existing DLMs on multiple benchmarks.

Copied to clipboard

Pengxiang Zhao, Guangyi Liu, Yaozhen Liang, Weiqing He, Zhengxi Lu, WenHao Wang, Yuehao Huang, Yuxiang Chai, Zhaolu Kang, Yaxuan Guo, Hao Wang, Kexin Zhang, Liang Liu, Yong Liu

HoWToBench: Holistic Evaluation for LLM’s Capability in Human-level Writing using Tree of Writing (2026.acl-long)

Challenge:	Shortcuts such as APIs and deep-links have emerged as efficient complements to flexible GUI operations, but systematic evaluation of GUI–shortcut hybrid agents remains underexplored.
Approach:	They propose a benchmark that evaluates GUI-shortcut hybrid agents with a specific focus on the mobile domain.
Outcome:	MAS-Bench evaluates agent's ability to generate shortcuts by discovering and creating reusable, low-cost workflows.

Copied to clipboard

Andrew Zhuoer Feng, Cunxiang Wang, Yu Luo, Lin Fan, Irene Zhou, Zikang Wang, Xiaotao Gu, Jie Tang, Hongning Wang, Minlie Huang

Challenge:	Evaluating the writing capabilities of large language models remains a significant challenge due to the multidimensional nature of writing skills and the limitations of existing metrics.
Approach:	They propose to model the aggregation weights of sub-features in a tree-structured workflow and propose a Chinese writing benchmark that mitigates biases.
Outcome:	The proposed tree-of-writing (ToW) measures the writing capabilities of large language models (LLMs) in Chinese and shows that it mitigates biases and achieves a 0.93 Pearson correlation with human judgments.

Hierarchical Reinforcement Learning with Augmented Step-Level Transitions for LLM Agents (2026.acl-long)

Copied to clipboard

Shuai Zhen, Yanhua Yu, Ruopei Guo, Nan Cheng, Yang Deng

Challenge:	Existing Large language model agents rely on increasingly long interaction histories, resulting in high computational cost and limited scalability.
Approach:	They propose a hierarchical reinforcement learning framework that enables step-level learning by conditioning only on single-step transitions rather than full interaction histories.
Outcome:	The proposed framework outperforms baselines on ScienceWorld and ALFWorld benchmarks in terms of performance and generalization while reducing token usage.

Two Streams, One Sarcasm: Orthogonal Expert Tuning for Holistic Multimodal Sarcasm Understanding (2026.acl-long)

Copied to clipboard

Diandian Guo, Cong Cao, Fangfang Yuan, Pin Xu, Cheng Hu, Zhicheng Zhang, Yu Liu, Yanbing Liu

SeLaR: Selective Latent Reasoning in Large Language Models (2026.acl-long)

Challenge:	Existing benchmarks for multimodal satirical cognition hinder evaluation of multimodal Sarcasm Understanding . lack of a unified benchmark for holistic satire cognition hampers evaluation of MSU .
Approach:	They propose a framework to decouple experts into orthogonal shared perception and private execution streams to physically block gradient interference between tasks.
Outcome:	The proposed framework achieves superior performance on DocMSU-PLUS.

Copied to clipboard

Renyu Fu, Guibo Luo

STRIDE-ED: A Strategy-Grounded Stepwise Reasoning Framework for Empathetic Dialogue Systems (2026.acl-long)

Challenge:	Recent latent reasoning approaches replace discrete tokens with soft embeddings or hidden states, but they often suffer from two issues: (1) global activation injects perturbations into high-confidence steps, impairing reasoning stability; and (2) soft embeds quickly collapse toward the highest-probability token, limiting exploration of alternative trajectories.
Approach:	They propose a lightweight and training-free framework that replaces discrete tokens with soft embeddings or hidden states to address these challenges.
Outcome:	Experiments on five reasoning benchmarks show that SeLaR outperforms standard CoT and state-of-the-art training-free methods.

Copied to clipboard

Hongru Ji, Yuyin Fan, Meng Zhao, Xianghua Li, Lianwei Wu, Chao Gao

GRAPHIA: Harnessing Social Graph Data to Enhance LLM-Based Social Simulation (2026.acl-long)

Challenge:	Empathetic dialogue requires not only recognizing a user’s emotional state but also making strategy-aware, context-sensitive decisions throughout response generation.
Approach:	They propose a STRategy-grounded, interpretable, and DEep reasoning framework that models Empathetic Dialogue through structured, strategy-conditioned reasoning.
Outcome:	The proposed framework outperforms existing methods on automatic metrics and human evaluations.

Copied to clipboard

Jiarui Ji, Zehua Zhang, Zhewei Wei, Bin Tong, Guan Wang, Bo Zheng

SharedRequest: Privacy-Preserving Model-Agnostic Inference for Large Language Models (2026.acl-long)

Challenge:	Social graphs provide high-quality supervision signals that encode local interactions and global network structure, yet they remain underutilized for LLM training.
Approach:	They propose a general LLM-based social graph simulation framework that leverages graph data as supervision for LLM training.
Outcome:	The proposed framework improves micro-level alignment by 6.1% on three real-world networks compared to the strongest baseline.

Copied to clipboard

Peihua Mai, Xuanrong Gao, Youlong Ding, Xianglong Du, Wei Liu, Yan Pang

Efficient Learned Data Compression via Dual-Stream Feature Decoupling (2026.acl-long)

Challenge:	Existing privacy-preserving inference methods sacrifice utility or efficiency, authors say . current approaches suffer a trilemma between privacy, utility, and efficiency, they say .
Approach:	They propose a model-agnostic framework for privacy-preserving LLM inference that reformulates privacy protection at the batch level rather than the individual-prompt level.
Outcome:	The proposed model-agnostic framework achieves 20% higher utility than previous models . it reduces query cost by up to 5 compared to non-batched inference .

Copied to clipboard

Huidong Ma, Xinyan Shi, Sun Hui, Xiaofei Yue, Xiaoguang Liu, Gang Wang, Wentong Cai

Planning-Guided Tutoring with Assessment-Driven Memory for Pedagogical LLM Tutors (2026.acl-long)

Challenge:	Learned data compression has achieved superior compression ratios, but balancing precise probability modeling with system efficiency remains challenging.
Approach:	They propose a Dual-Stream Multi-Scale Decoupler that disentangles local and global contexts to replace deep serial processing with shallow parallel streams.
Outcome:	The proposed method achieves state-of-the-art performance in both compression ratio and throughput while maintaining the lowest latency and memory usage.

Copied to clipboard

Zechen Li, Qiannan Zhu, Mei Wang, Jia Li, Hua Huang

MixKVQ: Query-Aware Mixed-Precision KV Cache Quantization for Long-Context Reasoning (2026.acl-long)

Challenge:	Existing approaches to simulate tutor behaviors or preferences fail to sustain high-quality pedagogical conversations that provide explicit stepwise scaffolding and adapt to learners’ evolving cognitive states.
Approach:	They propose a planning-guided tutoring framework with an assessment-driven memory for multi-turn math dialogue tutoring.
Outcome:	Experiments on multi-turn math tutoring benchmarks show that ScaffoldLM significantly improves pedagogical tutoring quality over strong baselines.

Copied to clipboard

Tao Zhang, Ziqian Zeng, Hao Peng, Huiping Zhuang, Cen Chen

When KV Cache Reuse Fails in Multi-Agent Systems: Cross-Candidate Interaction is Crucial for LLM Judges (2026.acl-long)

Challenge:	Existing low-bit quantization methods often exhibit severe performance degradation on complex reasoning tasks.
Approach:	They propose a plug-and-play method that uses a key channel's intrinsic quantization difficulty and relevance to the query to identify and preserve critical key channels that need higher precision.
Outcome:	Experiments on complex reasoning datasets show that the proposed method outperforms low-bit methods at a substantially reduced memory footprint.

Copied to clipboard

Sichu Liang, Zhenglin Wang, null Chujiajia, Pengfei Xia, Hui Zang, Deyu Zhou

Challenge:	Multi-agent LLMs generate multiple candidate responses that are aggregated by an LLM judge.
Approach:	They propose to advocate KV cache reuse across partially shared contexts and report substantial speedups for generation agents.
Outcome:	The proposed reuse strategies weaken cross-candidate attention, especially for later candidate blocks, and highlight judge-centric inference as a distinct regime that requires dedicated, risk-aware system design.

Reducing Token Redundancy in LVLMs: A Systematic Review of Token Pruning Methods (2026.acl-long)

Copied to clipboard

Hanzhang Yuan, Mengxuan Hu, Wenhao Zhang, Tianlong Wang, Zhongliang Zhou, Jiasen Lu, Sheng Li

STEM: Structure-Tracing Evidence Mining for Knowledge Graphs-Driven Retrieval-Augmented Generation (2026.acl-long)

Challenge:	Large Vision-Language Models (LVLMs) excel at visual understanding but face severe computational bottlenecks when processing high-resolution images and long videos due to massive visual token counts.
Approach:	They propose a taxonomy categorizing methods into vision-side, LLM-side and hybrid paradigms and analyze token selection mechanisms and pruning strategy.
Outcome:	The proposed method selectively removes less informative tokens while maintaining performance.

Copied to clipboard

Peng Yu, En Xu, Bin Chen, Haibiao Chen, Yinfei Xu

Evo-Attacker: Memory-Augmented Reinforcement Learning for Long-Horizon Tool Attacks on LLM-MAS (2026.acl-long)

Challenge:	Existing reasoning path retrieval methods lack a global structural perspective.
Approach:	They propose a framework that reframes multi-hop reasoning as a schema-guided graph search task.
Outcome:	The proposed framework improves accuracy and evidence completeness of multi-hop reasoning graph retrieval.

Copied to clipboard

Bingyu Yan, Xiaoming Zhang, JinYu Hou, Chaozhuo Li, Ziyi Zhou, Yiming Hei, Litian Zhang

Beyond Timestamps: Bridging Forward and Backward Reasoning in Temporal Numerical and Relational Understanding (2026.acl-long)

Challenge:	Existing tool attacks are limited by domain specificity or fixed and static templates.
Approach:	They propose an attack-based memory-augmented reinforcement learning process that constructs a dynamic attack memory and employs deliberative reasoning to retrieve adversarial patterns.
Outcome:	Evo-Attacker outperforms baselines in the long-horizon credit assignment challenge.

Copied to clipboard

Xinying Qian, Ying Zhang, Xuhui Sui, Yu Zhao, Baohang Zhou, Jeff Z. Pan

Reasoning Over Space: Enabling Geographic Reasoning for LLM-Based Generative Next POI Recommendation (2026.acl-long)

Challenge:	Existing benchmarks for Temporal Numerical and Relational reasoning rely on single-task evaluation paradigms.
Approach:	They propose a benchmark to evaluate Temporal Numerical and Relational reasoning . they propose QA and verification, and a Consistency Rate to quantify robustness .
Outcome:	The proposed framework evaluates both Temporal Numerical and Relational reasoning . it measures the alignment between QA and FV and the Consistency Rate measures robustness across these directions.

Copied to clipboard

Dongyi Lv, Qiuyu Ding, Heng-Da Xu, Zhaoxu Sun, Zhi Wang, Feng Xiong, Mu Xu

ET-Agent: Incentivizing Effective Tool-Integrated Reasoning Agent via Behavior Calibration (2026.acl-long)

Challenge:	Existing LLM-based recommenders lack explicit modeling of geographic signals . without explicit modeling geographic signals, recommenders struggle to capture core mobility patterns .
Approach:	They propose a framework that utilizes geography as a decision variable within the reasoning process.
Outcome:	The proposed framework achieves over 10% relative gains in hit rate over strongest LLM-based baselines and improves cross-city transfer.

Copied to clipboard

Yifei Chen, Guanting Dong, Zhicheng Dou

Spectral Disentanglement: Rank-Aware Task Adaptation for Rehearsal-free Continual Learning in LLMs (2026.acl-long)

Challenge:	Existing training frameworks for Large Language Models (LLMs) focus on answers’ accuracy, overlooking specific alignment for behavior patterns.
Approach:	They propose a training framework for calibrating agent’s tool-use behavior through two synergistic perspectives: self-evolving data flywheel and behavior calibration training.
Outcome:	The proposed framework improves the accuracy, efficiency, reasoning conciseness, and tool execution accuracy of large language models.

Copied to clipboard

Huanxuan Liao, Shizhu He, Yupu Hao, Yequan Wang, Wenhao Teng, Xiangwen Liao, Jun Zhao, Kang Liu

DeepPlanning: Benchmarking Long-Horizon Agentic Planning with Verifiable Constraints (2026.acl-long)

Challenge:	Continual Learning (CL) for Large Language Models faces a fundamental Stability-Plasticity Dilemma . Rank-Blindness enforces a single rank constraint across diverse tasks, leading to catastrophic forgetting of earlier tasks and underfitting on complex new ones.
Approach:	They propose a rank-spectrum-based rehearsal-free framework that explicitly disentangles knowledge into two orthogonal subspaces.
Outcome:	The proposed framework achieves a superior stability-plasticity balance compared to single-rank baselines.

Copied to clipboard

Yinger Zhang, Shutong Jiang, Renhao Li, Jianhong Tu, Yang Su, Lianghao Deng, Xudong Guo, ChenXu Lv, Junyang Lin

Arg-LLaDA: Argument Summarization via Large Language Diffusion Models and Sufficiency-Aware Refinement (2026.acl-long)

Challenge:	Existing LLM planning benchmarks emphasize local, step-level reasoning rather than global constrained optimization.
Approach:	They propose a benchmark for practical long-horizon agent planning that uses local constrained reasoning and global constrained optimization.
Outcome:	The proposed benchmarks show that even frontier agentic LLMs struggle with these problems.

Copied to clipboard

Hao Li, Yizheng Sun, Viktor Schlegel, Kailai Yang, Riza Batista-Navarro, Goran Nenadic

AgencyBench: Benchmarking the Frontiers of Autonomous Agents in 1M-Token Real-World Contexts (2026.acl-long)

Challenge:	Existing approaches to argument summarization rely on single-pass generation, offering limited support for factual correction or structural refinement.
Approach:	They propose a large language diffusion framework that iteratively improves argument summarization by sufficiency-guided remasking and regeneration.
Outcome:	Empirical results show that Arg-LLaDA surpasses state-of-the-art baselines in 7 out of 10 evaluation metrics.

Copied to clipboard

Keyu Li, Junhao Shi, Yang Xiao, Mohan Jiang, Jie Sun, Yunze Wu, Dayuan Fu, Shijie Xia, Xiaojie Cai, Tianze Xu, Weiye Si, Wenjie Li, Dequan Wang, Pengfei Liu

Challenge:	Existing benchmarks focus on single agentic capability, failing to capture long-horizon real-world scenarios.
Approach:	They propose a benchmark that evaluates 6 agentic capabilities across 32 real-world scenarios.
Outcome:	Experiments show that closed-source models outperform open-source model (48.4% vs 32.1%) integrating models with advanced scaffolds to form autonomous agents is a paradigm shift.

All Languages Matter: Understanding and Mitigating Language Bias in Multilingual RAG (2026.acl-long)

Copied to clipboard

Dan Wang, Guozhao Mo, Yafei Shi, Cheng Zhang, Bo Zheng, Boxi Cao, Xuanang Chen, Yaojie Lu, Hongyu Lin, Ben He, Xianpei Han, Le Sun

Beyond Accuracy: Unveiling Inefficiency Patterns in Tool-Integrated Reasoning (2026.acl-long)

Challenge:	Existing mRAG systems suffer from a language bias during reranking, systematically favoring English and the query’s native language.
Approach:	They propose a language-agnostic utility-driven reranker alignment technique to mitigate language bias during re-ranking.
Outcome:	The proposed approach mitigates language bias and consistently improves mRAG performance across languages.

Copied to clipboard

Qisheng Su, Shiting Huang, Zhen Fang, Ziyan Chen, Zehui Chen, Feng Zhao

Safety-Utility Conflicts Are Not Global: Surgical Alignment via Head-Level Diagnosis (2026.acl-long)

Challenge:	Tool-Integrated Reasoning (TIR) is a tool that can be used to solve complex tasks.
Approach:	They propose a hardware-aware TIR-efficiency metric that unifies internal reasoning and external tool-use costs while explicitly accounting for non-reusable KV-Cache and long-tool-response scenarios.
Outcome:	The proposed metric explains wall-clock latency significantly better than token-count metric in a simulated high-concurrency industrial setting.

Copied to clipboard

Wang Cai, Yilin Wen, Jinchang Hou, Du Su, Guoqiu Wang, Zhonghou Lv, Chenfu Bao, Yunfang Wu

VideoPro: Adaptive Program Reasoning for Long Video Understanding (2026.acl-long)

Challenge:	Existing mitigation strategies rely on global gradient geometry to resolve alignment conflicts . however, they overlook Modular Heterogeneity within Transformers, resulting in suboptimal trade-offs . Conflict-Aware Sparse Tuning (CAST) combines head-level diagnosis with sparse fine-tuning .
Approach:	They propose a framework that integrates head-level diagnosis with sparse fine-tuning to address this limitation.
Outcome:	The proposed framework integrates head-level diagnosis with sparse fine-tuning to reduce alignment conflicts in LLMs.

Copied to clipboard

Chenglin Li, Feng Han, Yikun Wang, Ruilin Li, Shuai Dong, Haowen Hou, Haitao Li, Qianglong Chen, Feng Tao, Jingqi Tong, Yin Zhang, Jiaqi Wang

Parity-Aware Byte-Pair Encoding: Improving Cross-lingual Fairness in Tokenization (2026.acl-long)

Challenge:	Existing methods for understanding long videos are limited due to the sparsity of visual evidence relevant to a given query.
Approach:	They propose a framework that enables VideoLLMs to reason over long videos and refine their predictions through executable programs.
Outcome:	The proposed framework outperforms existing methods across long-video understanding benchmarks.

Copied to clipboard

Negar Foroutan, Clara Meister, Debjit Paul, Joel Niklaus, Sina Ahmadi, Antoine Bosselut, Rico Sennrich

Boundary-Guided Policy Optimization for Memory-efficient RL of Diffusion Large Language Models (2026.acl-long)

Challenge:	Tokenization is the first step of most NLP pipelines.
Approach:	They propose a parity-aware byte pair encoder that maximizes the compression gain of the currently worst-compressed language for cross-lingual parity.
Outcome:	a new algorithm reduces tokenization inequality by 89% compared to classical BPE . the proposed algorithm is based on a fair-max rule that maximizes the compression gain of the currently worst-compressed language .

Copied to clipboard

Nianyi Lin, Jiajie Zhang, Lei Hou, Juanzi Li

ChartVerse: Scaling Chart Reasoning via Reliable Programmatic Synthesis from Scratch (2026.acl-long)

Challenge:	Existing methods for reinforcement learning (RL) require a large sample size to be implemented.
Approach:	They propose a memory-efficient RL algorithm that maximizes a lower bound of the ELBO-based objective.
Outcome:	Experiments show that BGPO outperforms previous RL algorithms for diffusion large language models in math problem solving, code generation, and planning tasks.

Copied to clipboard

Zheng Liu, Honglin Lin, Xiaoyang Wang, Xin Gao, Yu Li, Mengzhang Cai, Yun Zhu, Zhanping Zhong, Qizhi Pei, Zhuoshi Pan, Xiaoran Shang, Conghui He, Bin Cui, Wentao Zhang, Lijun Wu

NeoAMT: Neologism-Aware Agentic Machine Translation with Reinforcement Learning (2026.acl-long)

Challenge:	Existing open-source vision language models lack high-quality training data for chart reasoning . current models are simplistic and repetitive, while associated QA pairs are prone to hallucinations .
Approach:	They propose a framework to synthesize complex charts and reliable reasoning data from scratch.
Outcome:	Experimental results show that ChartVerse-8B surpasses existing models in QA and difficulty . lack of high-quality training data hampers development of open-source models .

Copied to clipboard

Zhongtao Miao, Kaiyan Zhao, Masaaki Nagata, Yoshimasa Tsuruoka

arXiv2Table: Toward Realistic Benchmarking and Evaluation for LLM-Based Literature-Review Table Generation (2026.acl-long)

Challenge:	Neologism-aware machine translation aims to translate source sentences containing neologismes into target languages.
Approach:	They propose an agentic framework for neologism-aware machine translation equipped with a Wiktionary-based search toolkit.
Outcome:	The proposed framework is based on a Wiktionary-based search toolkit and a dedicated dataset for neologism-aware machine translation.

Copied to clipboard

Weiqi Wang, Jiefu Ou, Yangqiu Song, Benjamin Van Durme, Daniel Khashabi

Responsible Evaluation of AI for Mental Health (2026.acl-long)

Challenge:	Literature review tables are essential for summarizing and comparing collections of scientific papers.
Approach:	They propose to generate a database of literature review tables from a pool of papers and to model retrieval noise via semantically related but out-of-scope distractor papers verified by human annotators.
Outcome:	The proposed method improves over strong baselines while the absolute scores remain modest, underscoring the task’s difficulty.

Copied to clipboard

Hiba Arnaout, Anmol Goel, H. Andrew Schwartz, Steffen T. Eberhardt, Dana Atzil-Slonim, Gavin Doherty, Brian Schwartz, Wolfgang Lutz, Tim Althoff, Munmun De Choudhury, Hamidreza Jamalabadi, Raj Sanjay Shah, Flor Miriam Plaza-del-Arco, Dirk Hovy, Maria Liakata, Iryna Gurevych

JPU: Bridging Jailbreak Defense and Unlearning via On-Policy Path Rectification (2026.acl-long)

Challenge:	Existing approaches to evaluating AI tools in this domain remain fragmented and inconsistent.
Approach:	They propose a taxonomy of AI mental health support types that integrates clinical soundness, social context, and equity to provide a structured basis for evaluation.
Outcome:	The proposed framework integrates clinical soundness, social context, and equity, providing a structured basis for evaluation.

Copied to clipboard

Xi Wang, Songlei Jian, Shasha Li, Xiaopeng Li, Zhaoye Li, Bin Ji, Baosheng Wang, Jie Yu

SpeechLLM-as-Judges: Towards General and Interpretable Speech Quality Evaluation (2026.acl-long)

Challenge:	Large Language Models (LLMs) are vulnerable to diverse jailbreak attacks despite extensive safety alignment .
Approach:	They propose a method to rectify dynamic jailbreak paths towards safety anchors by dynamically mining on-policy adversarial samples to expose vulnerabilities and identify jailbreak path.
Outcome:	The proposed model significantly improves jailbreak resistance against dynamic attacks while maintaining its utility.

Copied to clipboard

Hui Wang, Jinghua Zhao, Yifan Yang, Shujie Liu, Junyang Chen, Yanzhe Zhang, Shiwan Zhao, Jinyu Li, Jiaming Zhou, Haoqin Sun, Yan Lu, Yong Qin

LePREC: Reasoning as Classification over Structured Factors for Assessing Relevance of Legal Issues (2026.acl-long)

Challenge:	Existing methods for evaluating the perceptual quality of synthetic speech are limited due to the complexity of perceptual quality factors and the diversity of speech generation tasks.
Approach:	They propose a new paradigm for enabling large language models to conduct structured speech quality evaluation using a large-scale dataset.
Outcome:	The proposed model performs well across tasks and languages.

Copied to clipboard

Fanyu Wang, Xiaoxi Kang, Paul Burgess, Aashish Srivastava, Chetan Arora, Adnan Trakic, Lay-Ki Soon, Md Khalid Hossain, Lizhen Qu

S2O: Early Stopping for Sparse Attention via Online Permutation (2026.acl-long)

Challenge:	Large language models (LLMs) have impressive reasoning capabilities, but their precision remains inadequate.
Approach:	They propose a framework that integrates neural generation with statistical reasoning to improve the accuracy of large language models.
Outcome:	The proposed framework achieves interpretability through transparent feature weighting while maintaining data efficiency through correlation-based statistical classification.

Copied to clipboard

Yu Zhang, Songwei Liu, Chenqian Yan, null Linsheng, Beichen Ning, Fangmin Chen, Xing Wang

Challenge:	Existing block-granularity sparsification can reduce latency, but coarse blocks impose an intrinsic sparsity ceiling.
Approach:	They propose a method that performs early stopping for sparse attention via online permutation.
Outcome:	The proposed approach reduces the complexity of the model and its performance.

DART: Disambiguation-Aware Reasoning for Video-guided Machine Translation (2026.acl-long)

Copied to clipboard

Boyu Guan, Chuang Han, Yang Zhao, Chengqing Zong

HCSpec: Two-Tier Horizontal Cascade Speculative Decoding for High-Efficiency Large Language Model Inference (2026.acl-long)

Challenge:	Video-guided Machine Translation (VMT) uses short video clips to enhance translation quality, but many samples are text-sufficient.
Approach:	They propose a framework that integrates multimodal large language models’ multimodal reasoning into video-guided machine translation by using a pipeline for constructing training data based on multimodal relevance to translation.
Outcome:	The proposed framework improves multimodal information utilization in video-guided machine translation, yielding gains in translation quality and computational efficiency.

Copied to clipboard

Yizhou Zhang, Siming Chen, Hao Ye, Erhu Feng

Revisiting the Reliability of Language Models in Instruction-Following (2026.acl-long)

Challenge:	Existing approaches to decode large language models adopt a homogeneous architecture . autoregressive decoding is a bottleneck because tokens must be generated sequentially .
Approach:	They propose a framework that organizes heterogeneous position-specialized draft modules into a horizontal cascade.
Outcome:	The proposed framework outperforms the current state-of-the-art (EAGLE3) and achieves 3.72x acceleration over vanilla decoding.

Copied to clipboard

Jianshuo Dong, Yutong Zhang, Liu Yan, Zhenyu Zhong, Tao Wei, Chao Zhang, Han Qiu

EfficientLLM: Unified Pruning-Aware Pretraining for Auto-Designed Compact Language Models (2026.acl-long)

Challenge:	Several benchmarks have been proposed to measure instruction-following accuracy, but these scores do not translate to reliable services in real-world use.
Approach:	They propose a new metric reliable@k and develop an automated pipeline to generate cousin prompts.
Outcome:	The proposed model can be instantiated with cousin prompts and generates high-quality cousin prompt data.

Copied to clipboard

Xingrun Xing, Zheng Liu, Shitao Xiao, Boyan Gao, Yiming Liang, Haokun Lin, Xianlin Zeng, Guoqi Li, Jiajun Zhang

SPPO: Sequence-Level PPO for Long-Horizon Reasoning Tasks (2026.acl-long)

Challenge:	Large language models (LLMs) driven by scaling laws can be developed in large model sizes.
Approach:	They propose a pruning-aware pretraining approach that decouples LLM pruning from direct pretraining.
Outcome:	The proposed model outperforms pretraining models with 100M 1B parameters in commen sense benchmarks.

Copied to clipboard

Tianyi Wang, Yixia Li, Long Li, Yibiao Chen, Shaohan Huang, Yun Chen, Peng Li, Yang Liu, Guanhua Chen

Enhancing the Transferability of Jailbreak Attacks on Large Language Models via Exploiting Reparameterization Invariance (2026.acl-long)

Challenge:	Proximal Policy Optimization (PPO) is central to aligning Large Language Models with verifiable rewards.
Approach:	They propose a scalable algorithm that harmonizes sample efficiency with stability of outcome-based updates.
Outcome:	The proposed algorithm outperforms standard PPO and matches the performance of computation-heavy group-based methods.

Copied to clipboard

Ao Wang, Xinghao Yang, Yongshun Gong, Wei Liu, Bao-di Liu, Weifeng Liu

GIFT: Guided Fine-Tuning and Transfer for Enhancing Instruction-Tuned Language Models (2026.acl-long)

Challenge:	Existing token-level attacks have shown efficacy on open-source models but suffer from poor cross-model transferability.
Approach:	They propose a framework to improve cross-model transferability by modifying model parameters and generating update directions according to differences in output distributions rather than parameter-space distances.
Outcome:	The proposed framework improves cross-model transferability and success rates on open-source models.

Copied to clipboard

Zhiwen Ruan, Yichao Du, Jianjie Zheng, Longyue Wang, Yun Chen, Peng Li, Jinsong Su, Yang Liu, Guanhua Chen

AMATA: Adaptive Multi-Agent Trajectory Alignment for Knowledge-Intensive Question Answering (2026.acl-long)

Challenge:	Existing adapter-based transfer methods treat instruction-tuned models as passive targets . direct fine-tuning can disrupt this delicate balance and lead to instability or performance degradation.
Approach:	They propose a framework that incorporates instruction-level guidance into task adaptation.
Outcome:	The proposed framework outperforms direct fine-tuning and representative transfer-based baselines while maintaining robust generalization and favorable test-time scaling behavior.

Copied to clipboard

Taolin Zhang, Dongyang Li, Chen Chen, Qizhou Chen, Jiuheng Wan, Xiaofeng He, Chengyu Wang, Richang Hong

Question Answering

Challenge:	Empirical results show that AMATA outperforms baseline approaches, knowledge-augmented frameworks, and LLMs on knowledge-intensive QA benchmarks.
Approach:	They propose an Adaptive Multi-Agent Trajectory Alignment framework that integrates external knowledge to improve response interpretability and factual grounding.
Outcome:	The proposed framework outperforms baseline approaches, knowledge-augmented frameworks, and LLM-based trajectory systems on five established knowledge-intensive QA benchmarks.

ChessArena: A Chess Testbed for Evaluating Strategic Reasoning Capabilities of Large Language Models (2026.acl-long)

Copied to clipboard

Jincheng Liu, Sijun He, Jingjing Wu, Xiangsen Wang, Yang Chen, Zhaoqi Kuang, Siqi Bao, Yuan Yao

From Completion to Editing: Unlocking Context-Aware Code Infilling via Search-and-Replace Instruction Tuning (2026.acl-long)

Challenge:	Existing evaluation frameworks focus on isolated question-answering tasks that may not capture the essential aspects of strategic reasoning.
Approach:	They evaluate 13 large language models across over 800 games in chess . they use a chessian-based framework to test strategic reasoning and pattern recognition .
Outcome:	The proposed framework improves performance and basic understanding of large language models.

Copied to clipboard

Jiajun Zhang, Zeyu Cui, Jiaxi Yang, Lei Zhang, Yuheng Jing, Zeyao Ma, Tianyi Bai, Zilei Wang, Qiang Liu, Liang Wang, Binyuan Hui, Junyang Lin

What Makes an Ideal Quote? Recommending “Unexpected yet Rational” Quotations via Novelty (2026.acl-long)

Challenge:	Fill-in-the-Middle (FIM) models suffer from performance degradation and prohibitive latency.
Approach:	They propose a search-and-replace infilling framework that integrates agentic verification and editing into a single-pass inference process.
Outcome:	The proposed framework harmonizes completion tasks with the instruction-following priors of Chat LLMs, extending the paradigm from static infilling to dynamic context-aware editing.

Copied to clipboard

Powei Chang, Jin Xiao, Guanglei Yue, Qianyu He, Yanghua Xiao, Deqing Yang, Jiaqing Liang

Challenge:	Prior systems focus on topical relevance and overlook what makes quotes memorable.
Approach:	They propose a system that maps quotations and contexts into deep-meaning labels for label-enhanced retrieval.
Outcome:	The proposed system can recommend quotations that are contextually novel while semantically coherent.

Do Not Step Into the Same River Twice: Learning to Reason from Trial and Error (2026.acl-long)

Copied to clipboard

Chenming Tang, Hsiu-Yuan Huang, Weijie Liu, Clive Bai, Saiyong Yang, Yunfang Wu

iTAG: Inverse Design for Natural Text Generation with Accurate Causal Graph Annotations (2026.acl-long)

Challenge:	Existing approaches to RLVR train LMs based on their own on-policy responses and are constrained by the initial capability of LM.
Approach:	They propose an approach that hints LMs with their self-made mistakes without external guidance.
Outcome:	The proposed approach outperforms the normal group relative policy optimization and requires no external guidance.

Copied to clipboard

Wenshuo Wang, Boyu Cao, Nan Zhuang, Wei Li

Dynamic Long Context Reasoning over Compressed Memory via End-to-End Reinforcement Learning (2026.acl-long)

Challenge:	Lack of causally annotated text data for use as ground truth hinders causal discovery . early template-based generation methods sacrifice text naturalness in exchange for high annotation costs .
Approach:	They propose a method which performs real-world concept assignment to nodes before converting causal graphs into text.
Outcome:	The proposed method shows high annotation accuracy and naturalness across extensive tests.

Copied to clipboard

Zhuoen Chen, Dongfang Li, Meishan Zhang, Baotian Hu, Min Zhang

From Word to World: Can Large Language Models be Implicit Text-based World Models? (2026.acl-long)

Challenge:	Large Language Models face quadratic computational costs, information forgetting, and context fragmentation . current methodologies diverge into three paradigms, each facing inherent trade-offs between Equal contribution.
Approach:	They propose a framework that enables efficient long-context inference via chunk-wise compression and selective memory recall.
Outcome:	The proposed framework reduces peak GPU memory usage and speeds up inference on multi-hop reasoning benchmarks.

Copied to clipboard

Yixia Li, Hongru Wang, Jiahao Qiu, Zhenfei Yin, Dongdong Zhang, Cheng Qian, Zeping Li, Xiaoteng Ma, Guanhua Chen, Heng Ji

GAMBIT: A Gamified Jailbreak Framework for Multimodal Large Language Models (2026.acl-long)

Challenge:	Agentic learning increasingly hinges on interaction, yet real-world experience is expensive, limited, and often irreversible at inference time.
Approach:	They propose a framework that reframes language modeling as next-state prediction under interaction.
Outcome:	The proposed framework evaluates world models in text-based environments . it shows that sufficiently trained models capture coherent environment dynamics .

Copied to clipboard

Xiangdong Hu, Yangyang Jiang, Qin Hu, Xiaojun Jia

Large Language Model-Enhanced Multi-Armed Bandits (2026.acl-long)

Challenge:	Existing attacks focus on increasing the complexity of the modified visual task and do not explicitly leverage the model’s own reasoning incentives.
Approach:	They propose a framework that decomposes and reassembles harmful visual semantics and constructs a gamified scene that drives the model to explore, reconstruct intent and answer as part of winning the game.
Outcome:	Experiments on reasoning and non-reasoning MLLMs show that the proposed framework outperforms baseline models on both vision and text.

Copied to clipboard

Jiahang Sun, Zhiyong Wang, Runhan Yang, Chenjun Xiao, John C.s. Lui, Zhongxiang Dai

Flattery in Motion: Benchmarking and Analyzing Sycophancy in Video-LLMs (2026.acl-long)

Challenge:	Large language models (LLMs) have been used to sequential decision-making tasks like multi-armed bandits where an LLM is tasked with selecting arms in each iteration is often suboptimal.
Approach:	They propose to combine MAB and LLMs to leverage the in-context learning capability of LLM for reward prediction.
Outcome:	The proposed approach outperforms LLM-based direct arm selection on synthetic tasks where only preference feedback between arm pairs is available.

Copied to clipboard

Wenrui Zhou, Mohamed Hendy, Shu Yang, Qingsong Yang, Zikun Guo, Yuyu Luo, Lijie Hu, Di Wang

Mem2ActBench: A Benchmark for Evaluating Long-Term Memory Utilization in Task-Oriented Autonomous Agents (2026.acl-long)

Challenge:	Current sycophancy research has largely overlooked its specific manifestations in the video-language domain.
Approach:	They propose a video-LLM sycophancy benchmarking and evaluation to evaluate scophancies in video-LLMs.
Outcome:	The proposed benchmark evaluates sycophantic behavior in state-of-the-art Video-LLMs across diverse question formats, prompt biases, and visual reasoning tasks.

Copied to clipboard

Yiting Shen, Kun Li, Wei Zhou, Songlin Hu

Challenge:	Existing benchmarks focus on direct queries for a factual answer, but fail to evaluate the more crucial capability of actively applying memory to execute tasks.
Approach:	They propose a benchmark to evaluate whether agents can proactively leverage long-term memory to execute tool-based actions by selecting appropriate tools and grounding their parameters.
Outcome:	The proposed benchmarks show that 91.3% of tasks are memory-dependent . the benchmarks simulate persistent assistant usage, where users mention the same topic across long, interrupted interactions and expect previously established preferences and task states to be implicitly applied.

ARK: Answer-Centric Retriever Tuning via KG-augmented Curriculum Learning (2026.acl-long)

Copied to clipboard

Jiawei Zhou, Hang Ding, Haiyun Jiang

On the Emotion Understanding of Synthesized Speech (2026.acl-long)

Challenge:	Retrieval-Augmented Generation (RAG) is a powerful framework for knowledge-intensive tasks, but its effectiveness in long-context scenarios is often bottlenecked by the retriever’s inability to distinguish sparse yet crucial evidence.
Approach:	They propose a framework that fine-tunes the retriever for Answer Alignment by identifying high-quality positive chunks by evaluating their sufficiency to generate the correct answer.
Outcome:	The proposed framework improves 14.5% over the base model and maintains strong efficiency for long-context RAG.

Copied to clipboard

Yuan Ge, Haishu Zhao, AoKai Hao, Junxiang Zhang, Bei Li, Xiaoqian Liu, Chenglong Wang, Jianjin Wang, Bingsen Zhou, Bingyu Liu, JingBo Zhu, Zhengtao Yu, Tong Xiao

Taming "Zombie" Agents: A Markov State-Aware Framework for Resilient Multi-Agent Evolution (2026.acl-long)

Challenge:	Existing models for emotion understanding do not capture fundamental features of synthesized speech.
Approach:	They evaluate emotion recognition models on synthesized speech using SER models and generative models.
Outcome:	The proposed model can't generalize to synthesized speech because of speech token prediction . generative models tend to infer emotion from textual semantics while ignoring paralinguistic cues.

Copied to clipboard

Taolin Zhang, Pukun Zhao, Qizhou Chen, Jiuheng Wan, Chen Chen, Xiaofeng He, Chengyu Wang, Richang Hong

A Survey of Large Language Model-Based Search Agents (2026.acl-long)

Challenge:	Existing approaches to improve efficiency of multi-agent systems rely on aggressive graph topology evolution . however, such hard pruning overlooks the potential for "zombie" agents to recover and contribute in subsequent discussion rounds.
Approach:	They propose a Markov state-aware framework for resilient multi-agent evolution that manages agent collaboration through soft state transitions.
Outcome:	The proposed framework outperforms baselines and significantly reduces token consumption through state-aware agent scheduling.

Copied to clipboard

Yunjia Xi, Jianghao Lin, Yongzhao Xiao, Zheli Zhou, Rong Shan, Te Gao, Jiachen Zhu, Weiwen Liu, Yong Yu, Weinan Zhang

RSDA: Restoring Stale Data Affinity via Dynamic Renovation Strategy for Mitigating Data Scarcity (2026.acl-long)

Challenge:	Large Language Models (LLMs) have revolutionized web search, but their integration is static and cannot handle complex contexts.
Approach:	They analyze existing research and analyze existing work from the perspectives of architecture, optimization, application, and evaluation.
Outcome:	The proposed models can comprehend user intentions and context and execute multi-turn retrieval with dynamic planning, extending search capabilities far beyond the web.

Copied to clipboard

Yidan Liang, Jia Zhu, Weijie Shi, Hanghui Guo, Yue Cui, Jiawei Shen, Guoqing Ma, Jingjiang Liu, Qingyu Niu, Yilin Wang, Shimin Di, Jiajie Xu

The Reasoning Trap: How Enhancing LLM Reasoning Amplifies Tool Hallucination (2026.acl-long)

Challenge:	High-quality data is the cornerstone of advancing large language models, but the supply of premium data is nearing depletion, while vast stale corpora remain underutilized.
Approach:	They propose a framework to restore stale data affinity by quantifying the latent value of samples and employing a dynamic renovation strategy selection mechanism to determine the optimal component-level strategy.
Outcome:	The proposed framework achieves performance improvements using less than 10% of the data volume, underscoring that the latent potential of stale corpora remains largely untapped.

Copied to clipboard

Chenlong Yin, Zeyang Sha, Shiwen Cui, Changhua Meng, Zechao Li

Protecting multimodal large language models against misleading visualizations (2026.acl-long)

Challenge:	Recent studies suggest that strengthening reasoning often coincides with increased hallucination . however, no prior work has examined whether reasoning enhancement itself causes tool hallucinism .
Approach:	They propose a diagnostic benchmark measuring tool hallucination in two failure modes . they demonstrate a causal relationship between enhancing reasoning and tool hallubulation .
Outcome:	The proposed benchmark measures tool hallucination in two failure modes: no tool available, and (ii) only distractor tools available.

Copied to clipboard

Jonathan Tonglet, Tinne Tuytelaars, Marie-Francine Moens, Iryna Gurevych

Dual Activation-Weight Sparsity: A Training-Free Framework for Efficient Large Language Model Compression (2026.acl-long)

Challenge:	MLLMs are robust to misleading visualizations, i.e., charts that distort the underlying data, leading readers to draw inaccurate conclusions.
Approach:	They propose to use table-based QA and redrawing the visualization to improve QA performance on misleading visualizations.
Outcome:	The proposed methods improve MLLM question-answering accuracy on misleading visualizations without compromising accuracy on non-misleading ones.

Copied to clipboard

Luoyang Sun, Guangyan Li, Cheng Deng, Haifeng Zhang, Jian Zhao, Yongqiang Tang, Wensheng Zhang, Jun Wang

Repeated Sequences Reveal Gaps between Large Language Models and Natural Language (2026.acl-long)

Challenge:	Large language models (LLMs) excel at natural language tasks but face deployment bottlenecks due to computational demands.
Approach:	They propose a training-free framework that exploits activation and weight sparsity . they use a three-tier routing strategy that uses magnitude-based pruning .
Outcome:	Experiments on Llama and Mistral models show that DAWS outperforms activation-weight sparsity pruning methods.

Copied to clipboard

Kumiko Tanaka-Ishii

AgentCoMa: A Compositional Benchmark Mixing Commonsense and Mathematical Reasoning in Real-World Scenarios (2026.acl-long)

Challenge:	Existing evaluation methods provide limited insight into the long-range organization of generated text.
Approach:	They propose a framework for evaluation based on repeatedsubsequences . they compare their distribution across scales and their results to Rényi entropies .
Outcome:	The proposed framework relates distribution of results to higher-order Rényi entropies on human-written and length-matched GPT-generated texts.

Copied to clipboard

Lisa Alazraki, Lihu Chen, Ana Brassard, Joe Stacey, Hossein A. Rahmani, Marek Rei

HERMES: KV Cache as Hierarchical Memory for Efficient Streaming Video Understanding (2026.acl-long)

Challenge:	brittleness of Large Language Models in reasoningintensive tasks is a problem . current compositional benchmarks focus on either commonsense or math reasoning .
Approach:	They propose a "Commonsense and Math" benchmark where each compositional task requires a commonsense reasoning step and a math reasoning step.
Outcome:	The proposed benchmarks show that LLMs can solve both steps in isolation, but their accuracy drops by nearly 30% when the two steps are combined.

Copied to clipboard

Haowei Zhang, Shudong Yang, Jinlan Fu, See-Kiong Ng, Xipeng Qiu

Reasoning with OmniThought: A Large CoT Dataset with Verbosity and Cognitive Difficulty Annotations (2026.acl-long)

Challenge:	Existing models struggle to maintain stable understanding performance and low GPU memory overhead.
Approach:	They propose a training-free architecture for real-time and accurate understanding of video streams . HERMES reuses a compact KV cache, enabling efficient streaming understanding .
Outcome:	The proposed architecture achieves 10 faster TTFT compared to prior SOTA.

Copied to clipboard

Wenrui Cai, Chengyu Wang, Junbing Yan, Jun Huang, Xiangzhong Fang

InferenceDynamics: Adaptive LLM Routing through Structured Capability and Knowledge Profiling (2026.acl-long)

Challenge:	Existing resources often fail to provide extensive reasoning problems with coherent CoT processes distilled from multiple teacher models.
Approach:	They propose a large-scale dataset featuring 2 million CoT processes generated by multiple powerful LRMs.
Outcome:	The proposed dataset features 2 million CoT processes and is validated by multiple powerful LRMs.

Copied to clipboard

Haochen Shi, Tianshi Zheng, Weiqi Wang, Baixuan Xu, Chunyang Li, Chunkit Chan, Tao Fan, Yangqiu Song

ReportLogic: Evaluating Logical Quality in Deep Research Reports (2026.acl-long)

Challenge:	Large Language Model (LLM) routing is a pivotal technique for navigating a diverse landscape of LLMs.
Approach:	They propose a flexible and scalable multi-dimensional routing framework that models the capability and knowledge of models.
Outcome:	The proposed framework can be used to generalize and identify top-performing models for group-level routing using modern benchmarks including MMLU-Pro, GPQA, BigGenBench, and LiveBench.

Copied to clipboard

Jujia Zhao, Zhaoxin Huan, Zihan Wang, Xiaolu Zhang, Jun Zhou, Suzan Verberne, Zhaochun Ren

Detecting RAG Extraction Attack via Dual-Path Runtime Integrity Game (2026.acl-long)

Challenge:	Existing evaluation frameworks that evaluate large language models for Deep Research largely ignore this requirement.
Approach:	They propose a benchmark that quantifies report-level logical quality through a reader-centric lens of auditability.
Outcome:	The proposed model quantifies logical quality through a reader-centric lens of auditability.

Copied to clipboard

Yuanbo Xie, Yingjie Zhang, Yulin Li, Shouyou Song, Xiaokun Chen, Zhihan Liu, Liya Su, Tingwen Liu

Beyond Monolingual Assumptions: A Survey on Code-Switched NLP in the Era of Large Language Models across Modalities (2026.acl-long)

Challenge:	Retrieval-Augmented Generation (RAG) systems augment large language models with external knowledge, but introduce a critical security vulnerability: Knowledge Base Leakage.
Approach:	They propose a runtime defense mechanism inspired by stack canaries in software security . canaryRAG embeds carefully designed canary tokens into retrieved chunks and reformulates RAG extraction defense as a dual-path runtime integrity game.
Outcome:	The proposed system can detect and prevent RAG Knowledge Base Leakage in real time . it can be integrated into arbitrary RAG pipelines without retraining or structural modifications .

Copied to clipboard

Rajvee Sheth, Samridhi Raj Sinha, Mahavir Patil, Himanshu Beniwal, Mayank Singh

Look Light, Think Heavy: What Multimodal Chain-of-Thought Reasoning Can and Cannot Do (2026.acl-long)

Challenge:	Amidst the rapid advances of large language models, most LLMs struggle with mixed-language inputs, limited Code-switching datasets, and evaluation biases.
Approach:	They propose a roadmap for inclusive datasets, fair evaluation, and linguistically grounded models to achieve truly multilingual intelligence.
Outcome:	The proposed frameworks are based on 327 studies spanning five research areas, 15+ NLP tasks, 30+ datasets, and 80+ languages.

Copied to clipboard

Zhuoran Jin, Kejian Zhu, Hongbang Yuan, Yupu Hao, Pengfei Cao, Yubo Chen, Kang Liu, Jun Zhao

ARCQuant: Boosting NVFP4 Quantization with Augmented Residual Channels for LLMs (2026.acl-long)

Challenge:	Existing open-source models often yield only marginal overall improvements, possibly due to an overemphasis on mathematical reasoning at the expense of broader capabilities.
Approach:	They evaluate 12 multimodal tasks using 14 non-reasoning models and 8 reasoning models.
Outcome:	The proposed method is effective in multimodal reasoning tasks, the authors show . they show that it lacks the ability to maintain deep visual introspection throughout the reasoning process.

Copied to clipboard

Haoqian Meng, Yilun Luo, Yafei Zhao, Wenyuan Liu, Peng Zhang, Xindian Ma

Revisiting Model Interpolation for Efficient Reasoning (2026.acl-long)

Challenge:	NVFP4 supports fine-grained block isolation, 4-bit quantization errors and mixed-precision approaches . ARCQuant boosts NVFO4 performance via Augmented Residual Channels .
Approach:	They propose a framework that boosts NVFP4 performance via Augmented Residual Channels.
Outcome:	ARCQuant boosts NVFP4 performance via Augmented Residual Channels . the proposed framework achieves state-of-the-art accuracy comparable to full-precision baselines compared to FP16 .

Copied to clipboard

Taiqiang Wu, Runming Yang, Tao Liu, Jiahao Wang, Ngai Wong

MirageBackdoor: A Stealthy Attack that Induces Think-Well-Answer-Wrong Reasoning (2026.acl-long)

Challenge:	Existing models that interpolate weights of two specialized models can be abused for efficient reasoning.
Approach:	They propose to merge two specialized models and create a model that combines efficiency and efficiency.
Outcome:	The proposed method outperforms existing models on efficiency and effectiveness.

Copied to clipboard

Yizhe Zeng, Wei Zhang, Yunpeng Li, Juxin Xiao, Xiao Wang, Yuling Liu

MulVul: Retrieval-augmented Multi-Agent Code Vulnerability Detection via Cross-Model Prompt Evolution (2026.acl-long)

Challenge:	Existing CoT backdoor attacks manipulate intermediate reasoning steps to steer the model toward incorrect answers, but these corrupted reasoning traces are readily detected by prevalent process-monitoring defenses.
Approach:	They propose a backdoor attack that exploits the model's post-output space to preserve clean CoTs while selectively steering the final answer toward a specific target.
Outcome:	Experiments show that MirageBD achieves over 90% success rate across four datasets and five models with a poison ratio of only 5%.

Copied to clipboard

Zihan Wu, Jie Xu, Yun Peng, Chun Yong Chong, Xiaohua Jia

LiGen: Active Lipid Generation via a Molecular Language Model (2026.acl-long)

Challenge:	Large Language Models (LLMs) struggle to automate real-world vulnerability detection due to the heterogeneity of vulnerability patterns and manual prompt engineering for massive weakness categories is unscalable.
Approach:	They propose a retrieval-augmented multi-agent framework for precise and broad-coverage vulnerability detection using a coarse-to-fine strategy.
Outcome:	The proposed framework outperforms the baseline model on 130 CWE types and achieves 34.79% Macro-F1 performance.

Copied to clipboard

Ying Zhan, Xiuqi Tang, Yan Zhang, Xiao Tan, Dian Shen, Zhou Yu, Beilun Wang

Gained in Translation: Privileged Pairwise Judges Enhance Multilingual Reasoning (2026.acl-long)

Challenge:	Lipid nanoparticles (LNPs) can deliver cargos to tumor and immune cells . traditional approaches rely on experimental screening and expert judgment .
Approach:	They propose a method to generate lipid molecules efficiently and actively using deep learning.
Outcome:	The proposed method outperforms baseline methods on multiple cell lines and achieves a 30% improvement over the current methods.

Copied to clipboard

Lintang Sutawika, Gokul Swamy, Steven Wu, Graham Neubig

Ted-Tok: Maintaining an Evolving Vocabulary for Lifelong Learning (2026.acl-long)

Challenge:	Current reasoning large language models (RLMs) are trained on data that is primarily in English, resulting in lower performance when asked the same question in a non-English language.
Approach:	They propose a framework for enhancing multilingual reasoning without any data in the target language(s).
Outcome:	The proposed framework outperforms Qwen2.5-7B-Instruct on 4 math and non-math tasks with less than 1/8 of the training data (125).

Copied to clipboard

Jiameng Huang, Zhi Zhang, Zhenyu He, Jiacheng Sun, Di He

When Correct Beliefs Collapse: Epistemic Resilience of LLMs under Clinical Pressure (2026.acl-long)

Challenge:	a static tokenizer fragments newly emerging lexical items as language evolves . as language grows, a dynamic tokenizer reduces compression efficiency and performance .
Approach:	They propose a Temporal Drift Tokenizer that maintains an evolving vocabulary that adapts to emerging linguistic patterns over time.
Outcome:	The proposed tokenizer maintains an evolving vocabulary that adapts to emerging linguistic patterns over time.

Copied to clipboard

Boyu Xiao, Xiuqi Tian, Xuwen Song, Haochun Wang, Guanchun Song, Sendong Zhao, Bing Qin

Beyond Transcripts: A Renewed Perspective on Audio Chaptering (2026.acl-long)

Challenge:	Existing models exhibit severe multi-turn sycophancy in clinical dialogue . high initial diagnostic capability does not imply high belief stability .
Approach:	They propose a stress test framework that evaluates belief stability under escalating pressure.
Outcome:	The proposed stress test framework reduces the risk of multi-turn sycophancy in clinical dialogue . it eliminates belief change and improves robustness in training time .

Copied to clipboard

Fabian Retkowski, Maike Züfle, Thai Binh Nguyen, Jan Niehues, Alexander Waibel

SEA-BED: How Do Embedding Models Represent Southeast Asian Languages? (2026.acl-long)

Challenge:	despite its relevance, research on audio chaptering remains limited and predominantly textbased . authors: audio chapterers can't be used linearly because they skim, scrub timelines, jump to relevant moments . acoustic features and learning representations are not used for audio chapterer evaluation .
Approach:	They propose to use audio-only architecture to automatically segment audio into coherent sections . they compare audio-based models with acoustic features and a novel audio-oriented architecture .
Outcome:	The proposed audio-only architecture outperforms text-based approaches on acoustic features and LLMs.

Copied to clipboard

Wuttikorn Ponwitayarat, Peerat Limkonchotiwat, Raymond Ng, Jann Railey Montalan, Thura Aung, Jian Gang Ngui, Yosephine Susanto, William Chandra Tjhi, Panuthep Tasawong, Erik Cambria, Ekapol Chuangsuwanich, Sarana Nutanong

Is this chart lying to me? Automating the detection of misleading visualizations (2026.acl-long)

Challenge:	SEA-BED examines how multilingual text embeddings perform across tasks and languages . performance gaps arise from data coverage, training objectives, and architectural design, authors say .
Approach:	They propose a large-scale benchmark covering 10 SEA languages and diverse embedding tasks.
Outcome:	The proposed model performs poorly across languages and tasks, but language-task analyses reveal inconsistencies . the results suggest that performance gaps arise from limitations in data coverage, training objectives, and architectural design.

Copied to clipboard

Jonathan Tonglet, Jan Zimny, Tinne Tuytelaars, Iryna Gurevych

From Laboratory to Real-World Applications: Benchmarking Agentic Code Reasoning at the Repository Level (2026.acl-long)

Challenge:	Prior work has shown that both humans and MLLMs are frequently deceived by misleading visualizations.
Approach:	They propose a benchmark of 2,604 real-world visualizations annotated with 12 types of misleaders.
Outcome:	The proposed framework can detect misleading visualizations and identify specific design rules they violate . the proposed framework is based on a synthetic dataset of 81,814 visualizations .

Copied to clipboard

Jia Li, Yuxin Su, Michael R. Lyu

Privacy Collapse: Benign Fine-Tuning Can Break Contextual Privacy in Language Models (2026.acl-long)

Challenge:	Existing benchmarks for repository-level reasoning are inconsistent . repoReason is a white-box diagnostic benchmark centered on abductive assertion verification .
Approach:	They propose a white-box diagnostic benchmark centered on abductive assertion verification.
Outcome:	The proposed framework eliminates memorization while maintaining authentic logical depth . it also regenerates ground-truth states and quantifyes reasoning via three orthogonal metrics .

Copied to clipboard

Anmol Goel, Cornelius Emde, Seong Joon Oh, Sangdoo Yun, Martin Gubri

Dissecting Failure Dynamics in Large Language Model Reasoning (2026.acl-long)

Challenge:	Fine-tuning of frontier models can lead to privacy collapse, causing optimisation for helpfulness, exposure to user information, and debugging code printing internal variables.
Approach:	They propose to fine-tune frontier models to adapt to specific domains and align with organizational workflows and user preferences.
Outcome:	The proposed model fails to perform on safety and utility benchmarks while exhibiting severe privacy vulnerabilities.

Copied to clipboard

Wei Zhu, Jian Zhang, Lixing Yu, Kun Yue, Zhiwen Tang

QBridge: Bridging Natural Language and SQL via Gold Query Rewriting with Agentic Refinement (2026.acl-long)

Challenge:	Large Language Models achieve strong performance through extended inference-time deliberation, yet how their reasoning failures arise remains poorly understood.
Approach:	They propose a framework that probes and redirects critical transitions using uncertainty signals.
Outcome:	Empirical evaluations show that GUARD improves reasoning performance . GUard probes critical transitions and redirects them using uncertainty signals .

Copied to clipboard

Zhensheng Luo, Sai Wu, Yuan Qiu, Chang Yao, Gang Chen, Xiu Tang

SCVQ: Sparse-Compensated Vector Quantization for Large Language Models (2026.acl-long)

Challenge:	Natural language to SQL (NL2SQL) is an intuitive interface for querying structured data . but real user questions are noisy, ambiguous, and weakly grounded to database semantics.
Approach:	They propose an agentic feedback-driven NL2SQL framework that bridges natural language and SQL via Gold Query.
Outcome:	The proposed framework outperforms strong prompting and agentic baselines on spider, BIRD, and three robustness variants on NL2SQL.

Copied to clipboard

Zixuan Zhou, Yujun Diao, Zicheng Kong, Dehua Ma, Zhenbo Xu, Pei Pei Li, Zhaofeng He

Shanks: Simultaneous Hearing and Thinking for Spoken Language Models (2026.acl-long)

Challenge:	Existing vector quantization methods incur inference overhead due to massive codebook storage and intensive index lookups.
Approach:	They propose a framework for vector quantization that incorporates a salience-aware weighted K-means clustering scheme with symmetry constraints to reduce codebook size and indexing costs.
Outcome:	The proposed framework achieves a perplexity of 5.78 on WikiText-2 for LLaMA-2-7B at 2-bit quantization while delivering a 1.4 speedup over existing baselines.

Copied to clipboard

Cheng-Han Chiang, Xiaofei Wang, Linjie Li, Chung-Ching Lin, Kevin Lin, Shujie Liu, Zhendong Wang, Zhengyuan Yang, Hung-yi Lee, Lijuan Wang

Efficient KL Divergence Estimation via Truncated Top-K Integration for Large Language Models (2026.acl-long)

Challenge:	Existing large language models and spoken language models (SLMs) begin thinking and taking actions only after the user has finished their turn.
Approach:	They propose a general inference framework that enables SLMs to generate unspoken chain-of-thought reasoning while listening to user input.
Outcome:	The proposed framework enhances real-time user–SLM interaction in two scenarios.

Copied to clipboard

Xinyuan Wang, Luozhijie Jin, Bo Wang, Yuan Li, Zhangyue Yin, Xipeng Qiu

Visual Attention Reasoning via Hierarchical Search and Self-Verification (2026.acl-long)

Challenge:	Existing methods for estimating KL divergence using only top-k tokens suffer from high variance or systematic bias.
Approach:	They propose a top-k Importance-weighted KL Estimator that exploits the Zipfian structure of language model distributions by integrating only the top-K tokens.
Outcome:	The proposed estimator outperforms existing estimators on multiple benchmarks while exhibiting lower variance.

Copied to clipboard

Wei Cai, Jian Zhao, Yuchen Yuan, Tianle Zhang, Ming Zhu, Haichuan Tang, Xuelong Li

VFA: Empowering Multilingual MLLMs via Vision-Free Adaptation (2026.acl-long)

Challenge:	Multimodal Large Language Models (MLLMs) often hallucinate due to fragile, linear reasoning and weak visual grounding.
Approach:	They propose a framework that reformulates reasoning as a hierarchical search with self-verification and replaces linear Chain-of-Thought with a tree-search policy capable of backtracking to correct logical errors.
Outcome:	The proposed framework outperforms state-of-the-art methods on hallucination and safety benchmarks.

Copied to clipboard

Yixia Li, Yaqing Shi, Zhiwen Ruan, Dongdong Zhang, Lingjie Jiang, Shaohan Huang, Yun Chen, Guanhua Chen, Furu Wei

LearnerCoMPASS: Intelligent Tutoring System with Dynamic Cognitive Diagnosis and Multi-Model Path Planning (2026.acl-long)

Challenge:	Multimodal large language models have advanced rapidly, yet most remain English-centric . scaling multilingual multimodal instruction tuning is limited by the scarcity and high cost of non-English image–text supervision.
Approach:	They propose a framework that decouples multilingual language enhancement from visual alignment by composing complementary task vectors over a shared LLM backbone.
Outcome:	The proposed framework achieves competitive performance with a fully multimodally trained model using less than 2% of the text data.

Copied to clipboard

Ziji Sheng, Guiyao Tie, Weidong Wang, Pan Zhou, Daizong Liu

Investigating More Explainable and Partition-Free Compositionality Estimation for LLMs: A Rule-Generation Perspective (2026.acl-long)

Challenge:	Existing adaptive learning systems struggle to achieve deep personalization, dynamic adaptability, and content trustworthiness.
Approach:	They propose a framework that integrates large language models into adaptive learning systems . they propose 'cognitive multi-model planning adapted system' to enable deep personalization .
Outcome:	The proposed framework outperforms state-of-the-art learning paths and improves trustworthiness.

Copied to clipboard

Ziyao Xu, Cong Wang, Houfeng Wang

LVLMs and Humans Ground Differently in Referential Communication (2026.acl-long)

Challenge:	Compositional generalization tests focus on output results without considering sample compositionality, resulting in explainability defects.
Approach:	They propose a rule-generation perspective for compositionality estimation for LLMs that requires LLM to generate a program as rules for dataset mapping and provides estimates of compositionality using complexity-based theory.
Outcome:	The proposed model provides estimates of the compositionality of LLMs using complexity-based theory on a string-to-grid task.

Copied to clipboard

Peter Zeng, Weiling Li, Amie J. Paige, Zhengxiang Wang, Panagiotis Kaliosis, Dimitris Samaras, Gregory J. Zelinsky, Susan Brennan, Owen Rambow

Challenge:	generative AI agents cannot model common ground in a way that enables smooth communication . a recent study examined whether large language models and large vision language models engage in grounding as human discourse partners do .
Approach:	They propose to use referential communication to model common ground between a pair of directors and a picture matching system.
Outcome:	The proposed experiment shows that generative AI agents cannot model common ground . human conversation relies on common ground accrued and updated by interacting partners .

Evaluating the Expressive Appropriateness of Speech in Rich Contexts (2026.acl-long)

Copied to clipboard

Tianrui Wang, Ziyang Ma, Yizhou Peng, Haoyu Wang, Zhikang Niu, Zikang Huang, Yihao Wu, Yi-Wen Chao, Yu Jiang, Yuheng Lu, Guanrou Yang, Xuanchen Li, Hexin Liu, Chunyu Qiang, Cheng Gong, Yifan Yang, Tianchi Liu, Junyu Wang, Nana Hou, Meng Ge, Fuming You, Yang Wei, Zhongqian Sun, Hu Haifeng, Xiaobao Wang, Eng Siong Chng, Xie Chen, Longbiao Wang, Jianwu Dang

UniMoE-Audio: Unified Speech and Music Generation with Dynamic-Capacity Mixture-of-Experts (2026.acl-long)

Challenge:	Existing methods for evaluating expressive speech focus on word accuracy, naturalness, signal quality, or emotional intensity at the utterance level.
Approach:	They propose a framework for Evaluating Expressive Appropriateness in speech that assesses whether a speech sample aligns with the underlying communicative intent implied by its discourse-level narrative context.
Outcome:	The proposed framework outperforms existing speech evaluation and analysis systems on a human-annotated test set.

Copied to clipboard

Zhenyu Liu, Yunxin li, Xuanyu Zhang, Qixun Teng, Shenyuan Jiang, Xinyu Chen, Haoyuan Shi, Haolan Chen, Fanbo Meng, Mingjun Zhao, Yu Xu, Yancheng He, Baotian Hu, Haizhou Li, Min Zhang

Are LLMs Reliable Rankers? Rank Manipulation via Two-Stage Token Optimization (2026.acl-long)

Challenge:	Recent advances in unified multimodal models indicate a clear trend towards comprehensive content generation.
Approach:	They propose a unified speech and music generation model built upon a novel framework . they propose specialized MoE architectures and curated training strategies to tackle data imbalances .
Outcome:	The proposed model achieves state-of-the-art performance on major speech and music generation benchmarks.

Copied to clipboard

Tiancheng Xing, Jerry Li, Yixuan Du, Xiyang Hu

KCVR: Knowledge-Centric Video Reconstruction for Structured Pedagogical Summarization via Dynamic Graph Planning (2026.acl-long)

Challenge:	Large language models (LLMs) are increasingly used as rerankers, but their ranking behavior can be steered by small, natural-sounding prompts.
Approach:	They propose a token optimization method that crafts concise textual perturbations to consistently promote a target item in LLM-generated rankings.
Outcome:	The proposed method outperforms state-of-the-art base-lines and is hard to detect.

Copied to clipboard

Jingjiang Liu, Jia Zhu, Hanghui Guo, Weijie Shi, Yue Cui, Xiaokang Jin, Yilin Wang, Qingyu Niu, Jiawei Shen, Guoqing Ma, Yidan Liang, Shimin Di, Jiajie Xu

RepoGenesis: Benchmarking End-to-End Microservice Generation from Readme to Repository (2026.acl-long)

Challenge:	Existing summarization methods compress content for gist browsing, but they break prerequisite logic in instructional videos.
Approach:	They propose a framework that decouples epistemic planning from content generation.
Outcome:	The proposed framework outperforms strong end-to-end baselines on Knowledge Progression Consistency and Learning Objective Coverage.

Copied to clipboard

Zhiyuan Peng, Xin Yin, Pu Zhao, Fangkai Yang, Lu Wang, Ran Jia, Xu Chen, Qingwei Lin, Saravan Rajmohan, Dongmei Zhang

Challenge:	Existing benchmarks focus on isolated function/class-level generation, neglecting complete microservice repository generation.
Approach:	They propose a multilingual benchmark for repository-level end-to-end web microservice generation that reflects real-world development workflows.
Outcome:	The benchmark compared 106 repositories across 18 domains and 11 frameworks and 1,258 API endpoints and 2,335 test cases.

CityCube: Benchmarking Cross-view Spatial Reasoning on Vision-Language Models in Urban Environments (2026.acl-long)

Copied to clipboard

Haotian Xu, Yue Hu, Zhengqiu Zhu, Chen Gao, Ziyou Wang, Junreng Rao, Wenhao Lu, Weishi Li, Quanjun Yin, Yong Li

ContrastKV: Robust KV Cache Eviction via Contrastive Signal Fusion for Multi-Query Generalization (2026.acl-long)

Challenge:	Existing benchmarks focus on indoor or street settings, overlooking challenges of open-ended urban spaces.
Approach:	They propose a benchmark to probe cross-view spatial reasoning capabilities of current VLMs in urban settings.
Outcome:	The citycube benchmark examines the performance of current vision-language models in urban environments.

Copied to clipboard

Xingchi Chen, Peiyuan Zong, Ziqiang Gao, Qing Li, Yong Jiang, Fa Zhu, Hui Li

Teach a Reward Model to Correct Itself: Reward Guided Adversarial Failure Discovery for Robust Reward Modeling (2026.acl-long)

Challenge:	Existing query-agnostic approaches rely on a single proxy query, leading to fragile eviction decisions under high evict ratios.
Approach:	They propose a query-agnostic KV cache eviction algorithm that exploits complementary semantic and non-semantic signals.
Outcome:	Experiments show that the proposed algorithm outperforms state-of-the-art methods while retaining up to 92% accuracy with only 20% of the KV cache budget.

Copied to clipboard

Pankayaraj Pathmanathan, Furong Huang

Hierarchical Acoustic-Semantic Modeling: Modality Separation and Semantic Coherence for Full-Duplex SLMs (2026.acl-long)

Challenge:	Existing failure discovery methods rely on prior knowledge of preference attributes . Existing methods do not scale to new models or data.
Approach:	They propose a preference distribution agnostic procedure that uses the reward model itself to guide controlled decoding toward mis specified responses while preserving the underlying preference class.
Outcome:	The proposed procedure improves robustness without degrading reward quality across models.

Copied to clipboard

Zhenyu Liu, Xuanyu Zhang, Yunxin li, Qixun Teng, Shenyuan Jiang, Haolan Chen, Mingjun Zhao, Fanbo Meng, Yu Xu, Yancheng He, Baotian Hu, Haizhou Li, Min Zhang

VisPCO: Visual Token Pruning Configuration Optimization via Budget-Aware Pareto-Frontier Learning for Vision-Language Models (2026.acl-long)

Challenge:	despite significant progress, full-duplex SLMs are constrained by severe modality interference, authors say . modality interferes with acoustic and semantic modeling, making them unintelligent and unnatural . authors propose a hierarchical parameter separation strategy that decouples conflicting modalities in deep layers .
Approach:	They propose a hierarchical parameter separation strategy that decouples conflicting modalities in deep layers while preserving cross-modality coherence via a dedicated semantic alignment channel.
Outcome:	The proposed method significantly advances the state of the art on full-duplex benchmarks . it decouples conflicting modalities in deep layers while preserving cross-modality coherence .

Copied to clipboard

Huawei Ji, Yuanhao Sun, Yuan Jin, Cheng Deng, Jiaxin Ding, Luoyi Fu, Xinbing Wang

CoSToM: Causal-oriented Steering for Intrinsic Theory-of-Mind Alignment in Large Language Models (2026.acl-long)

Challenge:	Existing methods for visual token pruning rely on predefined configurations without determining whether they achieve optimal performance.
Approach:	They propose a framework that formulates visual token pruning as a Pareto configuration optimization problem to automatically identify optimal configurations.
Outcome:	The proposed framework approximates the empirical Pareto frontier obtained through grid search and generalizes well across pruning methods and VLM architectures.

Copied to clipboard

Mengfan Li, Xuanhua Shi, Yang Deng

ConlangCrafter: Constructing Languages with a Multi-Hop LLM Pipeline (2026.acl-long)

Challenge:	Large language models lack intrinsic cognition and cannot generalize to complex task-specific scenarios.
Approach:	They propose a framework that transitions from mechanistic interpretation to active intervention to map internal distributions of ToM features and implement it via targeted activation steering within ToM-critical layers.
Outcome:	The proposed framework significantly enhances human-like social reasoning capabilities and dialogue quality.

Copied to clipboard

Morris Alper, Moran Yanuka, Raja Giryes, Gasper Begus

Challenge:	Constructed languages (conlangs) have played diverse roles in art, philosophy, and international communication. foundation models have revolutionized creative generation in text, images, and beyond.
Approach:	They propose a multi-hop pipeline that decomposes language design into modular stages . they use LLMs' metalinguistic reasoning capabilities to encourage diversity .
Outcome:	The proposed pipeline decomposes language design into modular stages . it leverages LLMs’ metalinguistic reasoning capabilities to encourage diversity and self-refinement feedback to encourage consistency and typological diversity.

XY-Tokenizer: Mitigating the Semantic-Acoustic Conflict in Low-Bitrate Speech Codecs (2026.acl-long)

Copied to clipboard

Yitian Gong, Luozhijie Jin, Kuangwei Chen, Dong Zhang, Ruifan Deng, Xiaogui Yang, Xin Zhang, Zhaoye Fei, Qinyuan Cheng, Shimin Li, Xipeng Qiu

Empathy in Diversity: Personalized Depression and Anxiety Therapy via Dialogue State Tracking and Patient-Aware Planning (2026.acl-long)

Challenge:	Existing speech codecs struggle to balance these objectives at low bitrates . XY-Tokenizer achieves stronger semantic alignment than representative semantic-distillation codec .
Approach:	They propose a low-bitrate speech codec that aligns discrete speech representations with text while preserving fine-grained acoustic details for reconstruction.
Outcome:	The proposed codec outperforms existing low-bitrate speech codecs in speech understanding and generation tasks.

Copied to clipboard

Xinwei Yang, Junyi Fan, Yuqing Liu, Jiaxuan Wang, Jiashuai Zhang, Hongru Liang, Wenqiang Lei, Yao Song

VAPO: End-to-end Slide-Enhanced Speech Recognition with Omni-modal Large Language Models (2026.acl-long)

Challenge:	Recent efforts have turned to large language models (LLMs) as therapeutic agents for psychological therapy tasks, yet robustness across diverse patients remains underexplored.
Approach:	They propose a realistic role-play protocol for evaluating therapeutic dialogue agents and a de-identified, expert-annotated corpus of therapist–patient dialogues.
Outcome:	The proposed framework outperforms baselines on therapeutic outcomes and dialogue quality while improving conversational efficiency.

Copied to clipboard

Rui Hu, Delai Qiu, Yining Wang, Shengping Liu, Jitao Sang

"Newspaper Eat" Means "Not Tasty": A Taxonomy and Benchmark for Coded Language in Real-World Chinese Online Reviews (2026.acl-long)

Challenge:	Current Automatic Speech Recognition models, such as Whisper, have demonstrated impressive performance in general domains, but their accuracy often deteriorates significantly in specialized scenarios.
Approach:	They propose a visually-anchored policy optimization approach to decouple visual perception from auditory processing to optimize the model's inference process.
Outcome:	The proposed model eliminates visual interference and achieves state-of-the-art performance on SlideASR-Bench and public datasets.

Copied to clipboard

Ruyuan Wan, Changye Li, Ting-Hao Kenneth Huang

Data Mixing Agent: Learning to Re-weight Domains for Continual Pre-training (2026.acl-long)

Challenge:	Current language models handle coded language poorly, with limited real-world datasets and clear taxonomies.
Approach:	They propose a taxonomy that captures common encoding strategies including phonetic, orthographic, and cross-lingual substitutions.
Outcome:	The proposed model fails to detect or understand coded language in Chinese reviews . negative reviews can expose users to social pressure, retaliation, or reduced visibility .

Copied to clipboard

Kailai Yang, Xiao Liu, Lei Ji, Hao Li, Xiao Liang, Zhiwei Liu, Yeyun Gong, Peng Cheng, Mao Yang

Experience-driven Multi-turn Reinforcement Learning for GUI Agents (2026.acl-long)

Challenge:	Existing methods for reweighting data mixtures rely on manual designation with certain heuristics based on intuition or empirical results.
Approach:	They propose a model-based framework that learns to re-weight domains by reinforcement learning on large quantities of data mixing trajectories with corresponding feedback from an evaluation environment.
Outcome:	The proposed framework outperforms baselines in achieving balanced performance across source and target fields and domain spaces without retraining.

Copied to clipboard

Zhengxi Lu, Jiabo Ye, Fei Tang, Yongliang Shen, Haiyang Xu, Ziwei Zheng, Weiming Lu, Ming Yan, Fei Huang, Jun Xiao, Yueting Zhuang

Challenge:	GUI agents have demonstrated remarkable progress in automating complex user interface interactions . training such agents for long-horizon tasks remains challenging due to limited rewards and prohibitive costs.
Approach:	They propose a method that leverages expert trajectories as environment experiences for on-policy multi-turn training.
Outcome:	The proposed method achieves significant gains over the base model with 1K public trajectories as RL experiences . it achieves competitive performance against strong baselines such as UI-TARS-7B and GPT-4o .

UMMF: Protecting Copyright of Large Vision-Language Models through Unlearning-based Multimodal Memorization Fingerprint (2026.acl-long)

Copied to clipboard

Xiaofan Zheng, Xinghao Wang, Xiaojun Wan

Automatic Correction of Writing Anomalies in Hausa Texts (2026.acl-long)

Challenge:	Existing methods for fingerprinting large vision-Language Models rely on explicit triggers, which have limitations in terms of stealthiness and robustness.
Approach:	They propose to use model fingerprints to verify the ownership of large vision-Language Models (LVLMs) they use implicit model fingerprinting techniques that leverage neighboring samples as implicit model .
Outcome:	The proposed fingerprinting technique is superior to existing methods, but has limitations in terms of stealthiness and robustness.

Copied to clipboard

Ahmad Mustapha Wali, Sergiu Nisioi

Challenge:	Hausa texts are often characterized by writing anomalies such as incorrect character substitutions and spacing errors, which hinder natural language processing (NLP) applications.
Approach:	They propose to fine tune transformer-based Hausa-based models to correct writing anomalies by introducing synthetically generated noise to mimic real-world errors.
Outcome:	The proposed model improves Hausa text quality and improves other low-resource languages.

OS-Sentinel: Towards Safety-Enhanced Mobile GUI Agents via Hybrid Validation in Realistic Workflows (2026.acl-long)

Copied to clipboard

Qiushi Sun, Mukai Li, Zhoumianze Liu, Zhihui Xie, Fangzhi Xu, Zhangyue Yin, Kanzhi Cheng, Zehao Li, Zichen Ding, Qi Liu, Zhiyong Wu, Zhuosheng Zhang, Ben Kao, Lingpeng Kong

Challenge:	Existing methods for detecting unsafe mobile GUI agents are underexplored.
Approach:	They propose a mobile agent safety detection framework that integrates a formal verifier and a VLM-based contextual judge to detect system-level violations.
Outcome:	The proposed framework achieves 10%–30% improvements over existing approaches across multiple metrics.

PAM: Enhancing General Alignment of Large Reasoning Models through Priority-Aware Metacognition (2026.acl-long)

Copied to clipboard

Zhihao Xu, Fuzhen Yang, Liang Lin, Xiting Wang

One Battle After Another: Probing LLMs’ Limits on Multi-Turn Instruction Following with a Benchmark Evolving Framework (2026.acl-long)

Challenge:	Existing studies indicate that System-2 thinking alone does not transfer to the general alignment domain.
Approach:	They propose to use priority-aware metacognition to help LRMs understand human preferences and monitor and regulate their thinking process.
Outcome:	The proposed model improves general alignment performance by 10 points on helpfulness and harmless benchmarks.

Copied to clipboard

Qi Jia, Ye Shen, Xiujie Song, Kaiwei Zhang, Shibo Wang, Dun Pei, Xiangyang Zhu, Guangtao Zhai

AGSC: Adaptive Granularity and Semantic Clustering for Uncertainty Quantification in Long-text Generation (2026.acl-long)

Challenge:	Existing benchmarks for instruction-following in multi-topic dialogues are limited to a fixed number of turns, susceptible to saturation and failing to account for users’ interactive experience.
Approach:	They propose a framework featuring a three-layer tracking mechanism and a query synthesis agent to mimic sequential user behaviors.
Outcome:	The proposed framework outperforms existing benchmarks in the evaluation of instruction following in multi-topic dialogues and demonstrates deficiencies in failure recovery and fine-grained instruction following.

Copied to clipboard

Guanran Luo, Wentao Qiu, Wanru Zhao, Wenhan Lv, Zhongquan Jian, Meihong Wang, Qingqiang Wu

Tracing the Roots: A Multi-Agent Framework for Uncovering Data Lineage in Post-Training LLMs (2026.acl-long)

Challenge:	Existing methods for aggregating large-form outputs overlook the nuance of neutral information and suffer from the high computational cost of fine-grained decomposition.
Approach:	They propose a UQ framework that uses NLI neutral probabilities as triggers to distinguish irrelevance from uncertainty, reducing computation costs.
Outcome:	Experiments on BIO and LongFact show that the proposed framework reduces inference time by 60% compared to full atomic decomposition.

Copied to clipboard

Yu Li, Xiaoran Shang, Qizhi Pei, Yun Zhu, Xin Gao, Honglin Lin, Zhanping Zhong, Zhuoshi Pan, Zheng Liu, Xiaoyang Wang, Conghui He, Dahua Lin, Feng Zhao, Lijun Wu

Understanding or Memorizing? A Case Study of German Definite Articles in Language Models (2026.acl-long)

Challenge:	High-quality post-training data is the primary engine driving LLM capabilities . datasets are often treated as isolated artifacts, overlooking their true developmental context .
Approach:	They propose a framework to reconstruct the evolutionary graph of dataset development using data lineage.
Outcome:	The proposed framework characterizes domain-specific structural patterns in Math-oriented datasets and general-domain corpora.

Copied to clipboard

Jonathan Drechsel, Erisa Bytyqi, Steffen Herbold

Alloc-MoE: Budget-Aware Expert Activation Allocation for Efficient Mixture-of-Experts Inference (2026.acl-long)

Challenge:	grammatical agreement is unclear whether language models rely on memorization or generalization . morphologically rich languages such as German have syncretic forms that are syncretically arranged .
Approach:	They use a GRADIEND-based interpretability method to learn parameter update directions for gender-case specific article transitions.
Outcome:	Using GRADIEND, we find that updates learned for gender-case specific article transitions affect unrelated gender- case settings .

Copied to clipboard

Baihui Liu, Kaiyuan Tian, Wei Wang, Zhaoning Zhang, Linbo Qiao, Dongsheng Li

CodeEvo: Interaction-Driven Synthesis of Code-centric Data through Hybrid and Iterative Feedback (2026.acl-long)

Challenge:	Existing approaches that reduce expert activations lead to severe model performance degradation.
Approach:	They propose a framework that optimizes budget allocation coordinately at layer and token levels to minimize model performance degradation.
Outcome:	The proposed framework achieves 1.15 prefill and 1.34 decode speedups on DeepSeek-V2-Lite at half of the original budget.

Copied to clipboard

Qiushi Sun, Jingyang Gong, Lei Li, Qipeng Guo, Fei Yuan

Challenge:	Existing methods for generating instruction-code pairs rely on rigid heuristics and are labor-intensive.
Approach:	They propose a dual-agent architecture that integrates a Coder and a Reviewer to orchestrate the generation trajectory.
Outcome:	The proposed architecture outperforms baselines on a large-scale dataset of instruction-code pairs with stepped difficulty levels.

Benchmarking and Learning Real-World Customer Service Dialogue (2026.acl-long)

Copied to clipboard

Tianhong Gao, Jundong Shen, Jiapeng Wang, Bei Shi, Ying Ju, Junfeng Yao, Huiyu Yu

AdaJudge: Adaptive Multi-Perspective Judging for Reward Modeling (2026.acl-long)

Challenge:	Existing benchmarks and training pipelines for industrial intelligent customer service (ICS) focus on task completion and tool correctness.
Approach:	They propose a benchmark-to-optimization loop that bridges offline gains to deployment . they propose OlaMind, which distills reusable reasoning patterns from expert dialogues .
Outcome:	The proposed benchmark surpasses GPT-5.2 and Gemini 3 Pro on OlaBench . it delivers an average +23.67% issue resolution and -6.6% human transfer rate versus baseline .

Copied to clipboard

Yongliang Miao, Yangyang Liang, Mengnan Du

Challenge:	Existing reward models rely on a static pooling strategy to condense sequences into scalar scores, which is ill-suited for fine-grained discrimination.
Approach:	They propose a framework that jointly adapts representation and aggregation to address these limitations by integrating a static inductive bias with a representational mismatch.
Outcome:	Experiments on RM-Bench and JudgeBench show that AdaJudge outperforms strong off-the-shelf reward models and traditional pooling baselines.

IEvoAgent: Evolving Conversational Agent based on User Implicit Feedback (2026.acl-long)

Copied to clipboard

Yichen Cai, Jiayang Li, Junyuan Qiu, Jingya Guo, Weitao You, Changyuan Yang, Lingyun Sun, Pei Chen

SGPVT: Self-Generated Proximal Visual Tokens for Mitigating Proximal Collateral Damage in MLLM Unlearning (2026.acl-long)

Challenge:	Existing approaches to optimize conversational agents often rely on explicit preference pairs and expert evaluations.
Approach:	They propose a conversational agent framework that leverages the structured dependency between agent responses and user reactions to extract implicit feedback.
Outcome:	The proposed framework improves on MT-Bench-101, WildBench, and FB-Bech, and shows that mining implicit feedback supports better multi-turn alignment under evolving user preferences.

Copied to clipboard

Jiaqi Li, Zhijing Zhang, Jiahui Geng, Sheng Bi, Chuanyi Zhang, Fan Liu, Guilin Qi

VIGIL: Defending LLM Agents Against Tool-Stream Injection via Verify-Before-Commit (2026.acl-long)

Challenge:	Existing approaches focus on general utility metrics, overlooking the preservation of semantically related concepts.
Approach:	They propose a method that introduces self-generated proximal visual tokens to prevent forgetting vulnerability.
Outcome:	The proposed framework outperforms existing methods in preserving semantically related concepts while achieving effective target unlearning.

Copied to clipboard

Junda Lin, Zhaomeng Zhou, Zhi Zheng, Shuochen Liu, Tong Xu, Yong Chen, Enhong Chen

“I See What You Did There”: Can Large Vision-Language Models Understand Multimodal Puns? (2026.acl-long)

Challenge:	Existing defenses for indirect prompt injection are limited by static protection mechanisms . existing models prioritize injected rules due to strict alignment, whereas static protections sever the feedback loop required for adaptive reasoning.
Approach:	They propose a framework that shifts the paradigm from restrictive isolation to a verify-before-commit protocol.
Outcome:	The proposed framework outperforms state-of-the-art dynamic defenses by reducing the attack success rate by over 22% while more thandoubling utility under attack compared to static baselines.

Copied to clipboard

Naen Xu, Jiayi Sheng, Changjiang Li, Chunyi Zhou, Yuyuan Li, Tianyu Du, Jun Wang, Zhihui Fu, Jinbao Li, Shouling Ji

TRACE: Traversal Retrieval-Augmented Chain of Evidence for Document Understanding (2026.acl-long)

Challenge:	Puns are a common form of rhetorical wordplay that exploits polysemy and phonetic similarity to create humor.
Approach:	They propose a multimodal pun generation pipeline and a model to evaluate their understanding of puns.
Outcome:	The proposed benchmark improves the understanding of multimodal puns by 16.5% in the F1 test.

Copied to clipboard

Liqi He, Zuchao Li, Hao Huang, Ping Wang

FinKario: Event-Enhanced Automated Construction of Financial Knowledge Graph (2026.acl-long)

Challenge:	Long-context Document Visual Question Answering (DocVQA) methods struggle with visual semantics or handling finite context windows.
Approach:	They propose a new approach to longcontext document visual question answering that transforms retrieval into adaptive evidence chain construction using a Bi-Layered Graph.
Outcome:	The proposed approach achieves an average accuracy improvement of 14.07% on M5BookVQA and exhibits robust generalization with a 13.38% gain across four established benchmarks.

Copied to clipboard

Xiang Li, Penglei Sun, Wanyun Zhou, Zikai Wei, Yongqi Zhang, Xiaowen Chu

EvoSci: A Bio-Inspired Multi-Agent Framework for the Evolution of Scientific Discovery (2026.acl-long)

Challenge:	Equity research reports are crucial resources for investors, but lack professional analysis and the rapid evolution of market events outpaces their update cycles.
Approach:	They propose an event-Enhanced automated construction of financial knowledge graph (FinKario) that automatically integrates real-time company fundamentals and market events through prompt-driven extraction guided by professional institutional templates.
Outcome:	The proposed model outperforms financial LLMs by 18.81% and institutional strategies by 17.85% on average in backtesting.

Copied to clipboard

Xiaoyu Xiong, Yuqi Ren, Deyi Xiong

DREAM: Deep Research Evaluation with Agentic Metrics (2026.acl-long)

Challenge:	Existing methods for large language models (LLMs) face challenges in the design of research workflows and multi-role collaboration mechanisms.
Approach:	They propose a multi-agent scientific collaboration framework which integrates bio-inspired evolution with knowledge graph modeling.
Outcome:	EvoSci outperforms baselines in peer-review and ranking evaluations on real-world research topics.

Copied to clipboard

Elad Ben Avraham, ChangHao Li, Ron Dorfman, Roy Ganz, Oren Nuriel, Amir Dudai, Aviad Aberdam, Noah Flynn, Elman Mansimov, Aditya Kalyanpur, Ron Litman

ThinkPersona: Thinking with Persona Graphs for Faithful Individualized Role-Playing (2026.acl-long)

Challenge:	Recent benchmarks propose distinct methodologies, yet they suffer from the Mirage of Synthesis . static evaluators lack the tool-use capabilities required to assess temporal validity and factual correctness .
Approach:	They propose a framework that instantiates the principle of capability parity by making evaluation agentic.
Outcome:	The proposed framework is more sensitive to factual decay than existing benchmarks . large language models increasingly support autonomous, tool-using agents .

Copied to clipboard

Yichen Cai, Pei Chen, Jiayang Li, Jingya Guo, Zejian Li, Changyuan Yang, Lingyun Sun

Sparrow: Text-Anchored Window Attention with Visual-Semantic Glimpsing for Speculative Decoding in Video LLMs (2026.acl-long)

Challenge:	Large Language Models are increasingly utilized as role-playing agents to simulate personas in interactive settings.
Approach:	They propose a role-playing agent trained to explicitly ground responses in individual identity.
Outcome:	The proposed agent can generate persona-consistent responses in long-context dialogues while maintaining general instruction-following capabilities.

Copied to clipboard

Libo Zhang, Zhaoning Zhang, null Hongwanyang, Peng Qiao, Dongsheng Li

SPEAK: Spiking Neurons as an Entropy-Aware Tokenizer for Large Language Models (2026.acl-long)

Challenge:	Existing speculative decoding models face performance collapse due to key-value cache explosion and context window mismatches.
Approach:	They propose a framework that offloads visual computation to the target model by using hidden state reuse.
Outcome:	The proposed framework achieves an average speedup of 2.82x even with 25k visual tokens .

Copied to clipboard

Ming Chen, Wenyao Li, Chao Liang, Shi Gu, Peng Lin, De Ma, Huajin Tang, Qian Zheng, Gang Pan

Unlocking Implicit Experience: Synthesizing Tool-Use Trajectories from Text (2026.acl-long)

Challenge:	Existing tokenizers fail to explicitly leverage historical tokenization results . large language models (LLMs) have demonstrated remarkable effectiveness across NLP tasks .
Approach:	They propose a tokenizer that integrates spiking neurons to explicitly leverage historical tokenization results.
Outcome:	The proposed tokenizer leverages historical tokenization results, but does not selectively leverage history based on contextual relevance.

Copied to clipboard

Zhihao Xu, Rumei Li, Jiahuan Li, Rongxiang Weng, Jingang Wang, Xunliang Cai, Xiting Wang

A Layer-wise Analysis of Supervised Fine-Tuning (2026.acl-long)

Challenge:	Large language models (LLMs) can be used to effectively utilize tools in multi-turn interactions, but acquiring diverse and realistic multi-step tool-use data remains a challenge.
Approach:	They propose a text-based data synthesis pipeline that generates multi-turn tool-use trajectories from text corpora using relevance filtering, workflow tool extraction, trajectory grounding, and complexity refinement.
Outcome:	The proposed model achieves 14.9% improvement on the BFCL V3 Multi-turn benchmark while significantly reducing inference latency and costs.

Copied to clipboard

Qinghua Zhao, Xueling Gong, Xinyu Chen, Zhongfeng Kang, Xinlu Li

Think in Latent Thoughts: A New Paradigm for Gloss-Free Sign Language Translation (2026.acl-long)

Challenge:	Existing methods for fine-tuning ignore depth-dependent heterogeneity of instruction-following . a critical gap remains in understanding where these changes occur across the model's depth and which layers are essential for instruction- following.
Approach:	They propose a method which selectively updates critical intermediate layers . they show that effective alignment is architecturally localized rather than distributed .
Outcome:	The proposed method outperforms standard LoRA up to 10.2% on GSM8K with reduced parameter overhead.

Copied to clipboard

Yiyang Jiang, Li Zhang, Xiao-Yong Wei, Li Qing

CTRAP: Embedding Collapse Trap to Safeguard Large Language Models from Harmful Fine-Tuning (2026.acl-long)

Challenge:	Existing approaches to sign language translation (SLT) assume video segments are directly mappable to spoken-language words.
Approach:	They propose a reasoning-driven SLT framework that uses an ordered sequence of latent thoughts as an explicit middle layer between video and generated text.
Outcome:	The proposed model improves coherence and faithfulness over existing gloss-free methods.

Copied to clipboard

Biao Yi, Tiansheng Huang, Baolei Zhang, Tong Li, Lihai Nie, Zheli Liu, Li Shen

Correct When Paired, Wrong When Split: Decoupling and Editing Modality-Specific Neurons in MLLMs (2026.acl-long)

Challenge:	Fine-tuning-as-a-service exposes models to harmful fine-tuneing attacks . however, inherent general adaptability of LLMs allows them to bypass selective unlearning by rapidly relearning or repurposing their general capabilities for harmful tasks.
Approach:	They propose a paradigm shift that inducing model collapse instead of selective removal by relearning or repurposing general capabilities for harmful tasks.
Outcome:	The proposed model collapse mechanism neutralizes the very general capabilities that attackers exploit, tackling the core issue unaddressed by selective unlearning.

Copied to clipboard

Tingchao Fu, Wenkai Wang, Fanxiao Li, Huadong Zhang, Jinhong Zhang, Dayang Li, Yunyun Dong, Renyang Liu, Wei Zhou

ACIArena: Toward Unified Evaluation for Agent Cascading Injection (2026.acl-long)

Challenge:	Existing knowledge editing paradigms suffer from editing decoupling failures . entity knowledge is sequestered into disentangled modality-specific pathways .
Approach:	They propose a method that explicitly disentangles and localizes modality-specific neuron groups for targeted knowledge.
Outcome:	The proposed method outperforms baselines in reliability and consistency while preserving model locality.

Copied to clipboard

Hengyu An, Minxi Li, Jinghuai Zhang, Naen Xu, Chunyi Zhou, Changjiang Li, Xiaogang Xu, Tianyu Du, Shouling Ji

PLAWBENCH: A Rubric-Based Benchmark for Evaluating LLMs in Real-World Legal Practice (2026.acl-long)

Challenge:	Existing studies consider only limited attack strategies and simplified MAS settings, limiting their generalizability and comprehensive evaluation.
Approach:	They propose a framework to evaluate the robustness of Multi-Agent Systems (MAS) they propose unified evaluation suites spanning attack surfaces and attack objectives .
Outcome:	ACIArena provides a benchmark of 1,356 test cases for evaluating MAS robustness . it covers six widely used MAS implementations and provides measurable results .

Copied to clipboard

Yuzhen Shi, Huanghai Liu, Yiran HU, Song Gaojie, Xu Xinran, Yubo Ma, Tianyi Tang, Li Zhang, Qingjing Chen, Feng Di, Wenbo Lv, Weiheng Wu, Kexin Yang, Sen Yang, Wei Wang, Rongyao Shi, Qiu Yuanyang, Yuemeng Qi, Zhang Jingwen, Sui Xiaoyu, Yifan Chen, Zhang Yi, An Yang, Bowen Yu, Dayiheng Liu, Junyang Lin, Weixing Shen, Bing Zhao, Charles L. A. Clarke, HU Wei

VoxMind: An End-to-End Agentic Spoken Dialogue System (2026.acl-long)

Challenge:	Existing benchmarks for large language models (LLMs) are coarse, single-dimensional metrics and do not explicitly assess fine-grained legal reasoning.
Approach:	They propose a Practical Law Benchmark to evaluate large language models in real-world legal practice scenarios.
Outcome:	The proposed model is based on 850 questions and 13 scenarios with expert-designed evaluation rubrics.

Copied to clipboard

Tianle Liang, Yifu Chen, Shengpeng Ji, Yijun Chen, Zhiyang Jia, Jingyu Lu, Fan Zhuo, Xueyi Pu, Yangzhuo Li, Zhou Zhao

MT3: A Synergistic Multi-Task RL Framework for Specializing MLLMs in Text Image Machine Translation (2026.acl-long)

Challenge:	Existing research on end-to-end spoken dialogue models has focused on core perception and generation, with limited exploration of tool-augmented extensions.
Approach:	They propose a framework to equip end-to-end spoken dialogue models with comprehensive agentic abilities by leveraging a 470-hour AgentChat dataset.
Outcome:	The proposed framework outperforms Gemini-2.5-Pro on spoken agent tasks while maintaining general conversational quality.

Copied to clipboard

Zhaopeng Feng, Yupu Liang, Shaosheng Cao, Jiayuan Su, Jiahan Ren, Zhijie Zhou, Wenxuan Huang, Jian Wu, Zuozhu Liu

Do LLMs Capture Embodied Cognition and Cultural Variation? Cross-Linguistic Evidence from Demonstratives (2026.acl-long)

Challenge:	Text Image Machine Translation (TIMT) is a critical subfield of machine translation . it requires accurate optical character recognition, robust visual-text reasoning, and high-quality translation a challenge .
Approach:	They propose a multi-task optimization framework to specialize MLLMs into expert TIMT models.
Outcome:	The proposed model outperforms baselines on the latest in-domain MIT-10M benchmark.

Copied to clipboard

Yu Wang, Emmanuele Chersoni, Chu-Ren Huang

Challenge:	a new study examines whether large language models acquire embodied cognition and cultural conventions from training data . demonstratives are a natural lens for evaluating linguistic phenomena that reflect cultural variation . aaron e. duan and j. nà: "the complexity of the language model is a major challenge for LLMs"
Approach:	They introduce demonstratives as a probe for grounded knowledge by analyzing 6,400 responses from 320 native speakers.
Outcome:	The proposed model fails to understand proximal–distal contrast and shows no cultural differences . the proposed model is a new probe for evaluating embodied cognition and cultural conventions .

Feeling Rules in Language Models: Mapping Norms of Emotional Appropriateness Across Roles, Institutions, and Intensity (2026.acl-long)

Copied to clipboard

Guangrui Fan, DanDan Liu, Aznul Qalid MD Sabri, Rui Zhang, Pan Lihu

SGT: Securing Open-Source LLMs Against Malicious Fine-tuning via Safety Guidance Trigger (2026.acl-long)

Challenge:	Existing benchmarks measure whether Large language models recognize emotions . authors: LLMs can be used to validate, but they can still judge anger inappropriately .
Approach:	They propose a benchmark to measure whether Large language models validate anger . they use explicit norm judgments and implicit acceptability tests to measure norms .
Outcome:	The study finds that large differences in sanctioning thresholds and institutional norm signatures are not reducible to overall strictness.

Copied to clipboard

Sunguk Shin, Fangzhao Wu, Byung-Jun Lee, Meeyoung Cha, Sungwon Park

Every Token Counts: Generalizing 16M Ultra-Long Context in Large Language Models (2026.acl-long)

Challenge:	open-weight large language models increase vulnerability to malicious fine-tuning . despite these advantages, open-source LLMs increase the potential for misuse .
Approach:	They propose a safety guide for open-weight large language models that guides fine-tuning toward the safety manifold to preserve alignment.
Outcome:	The proposed safety guidance trigger significantly improves robustness against malicious fine-tuning.

Copied to clipboard

Xiang Hu, Zhanchao Zhou, Ruiqi Liang, Zehuan Li, Wei Wu, Jianguo Li

Self-Reflective Generation at Test Time (2026.acl-long)

Challenge:	a recent study explores efficient ultra-long context modeling.
Approach:	They propose to use Hierarchical Sparse Attention to achieve efficient ultra-long context modeling.
Outcome:	The proposed model performs comparable to full-attention baselines on in-domain and out-of-domain tasks.

Copied to clipboard

Jian Mu, Qixin Zhang, Zhiyong Wang, Menglin Yang, Shuang Qiu, Chengwei Qin, Zhongxiang Dai, Yao Shu

GKnow: Measuring the Entanglement of Gender Bias and Factual Gender (2026.acl-long)

Challenge:	Existing self-reflection mechanisms are reactive and inefficient for large language models . a fundamental tension persists between the ability to execute complex multi-step reasoning and the ability of the model to generate coherent traces.
Approach:	They propose a test-time framework that reflects before generating at uncertain points . SRGen utilizes dynamic entropy thresholding to identify high-uncertainty tokens .
Outcome:	The proposed framework can significantly strengthen large language models' reasoning process.

Copied to clipboard

Leonor Veloso, Hinrich Schuetze

Reinforcing Agentic Search Via Reward Density Optimization (2026.acl-long)

Challenge:	Recent studies have focused on mitigating gender bias, but mechanistic interpretations of gender fail to distinguish between factually gendered outputs and gender biased outputs.
Approach:	They propose a benchmark to assess gender knowledge and gender bias in language models . they use neuron ablation to disentangle stereotypical and factual gender .
Outcome:	The proposed benchmark assesses gender knowledge and gender bias in language models across different types of gender-related predictions.

Copied to clipboard

Kun Luo, Hongjin Qian, Zheng Liu, Ziyi Xia, Shitao Xiao, Zhao Cao, Siqi Bao, Jun Zhao, Kang Liu

Scaling External Knowledge Input Beyond Context Windows of LLMs via Multi-Agent Collaboration (2026.acl-long)

Challenge:	Reinforcement Learning with Verifiable Rewards (RLVR) is a promising approach for enhancing agentic search, but its performance is often hindered by reward sparsity .
Approach:	They propose a new research problem to improve the reward obtained per unit of exploration cost by using a system that decomposes long-horizon tasks into intermediate objectives and assigns process-level rewards to provide denser learning signals.
Outcome:	The proposed framework outperforms strong baselines on several agentic search benchmarks and achieves comparable performance to that of advanced proprietary models.

Copied to clipboard

Zijun Liu, Zhennan Wan, Peng Li, Ming Yan, Fei Huang, Yang Liu

M2PO: Multi-Perspective Multi-Pair Preference Optimization for Machine Translation (2026.acl-long)

Challenge:	Existing context window extension methods obstruct scaling external knowledge input.
Approach:	They develop a multi-agent framework to overcome two core bottlenecks in existing agent orchestration designs.
Outcome:	The proposed framework overcomes two core bottlenecks and improves inference-time knowledge integration without longer-context training.

Copied to clipboard

Hao Wang, Linlong Xu, Heng Liu, Yangyang Liu, Xiaohu Zhao, Bo Zeng, Liangying Shao, Yichen Dong, Xinwei Wu, Jiang Zhou, Tianyu Dong, Xiangxiang Zeng, Longyue Wang, Weihua Luo

An Exploration of Mamba for Speech Self-Supervised Models (2026.acl-long)

Challenge:	prevailing methods for machine translation are often hindered by misleading reward signals.
Approach:	They propose a framework that aligns large language models to human preferences . they propose 'M2PO' to correct the bias towards partial errors .
Outcome:	The proposed framework outperforms open-source models and achieves parity with proprietary models.

Copied to clipboard

Tzu-Quan Lin, Heng-Cheng Kuo, Tzu-Chieh Wei, Hsi-Chun Cheng, Chun Wei Chen, Hsien-Fu Hsiao, Yu Tsao, Hung-yi Lee

Ready Jurist One: Benchmarking Language Agents for Legal Intelligence in Dynamic Environments (2026.acl-long)

Challenge:	Mamba-based SSL models are promising for long-sequence modeling, speech unit extraction, and speech self-supervised learning.
Approach:	They propose to use Mamba-based HuBERT models as an alternative to Transformer-based SSL architectures.
Outcome:	The proposed models outperform Transformer-based models in language modeling tasks while showing superior performance on streaming ASR.

Copied to clipboard

Zheng Jia, Shengbin Yue, Wei Chen, Siyuan Wang, Yidong Liu, Zejun Li, Yun Song, Zhongyu Wei

Challenge:	Existing benchmarks for legal intelligence are limited to static evaluation paradigms or simplified scenarios.
Approach:	They introduce J1-ENVS, the first interactive and dynamic legal environment tailored for LLM-based agents.
Outcome:	The proposed framework assesses task performance and procedural compliance across legal proficiency levels.

When Morphology Hides in Plain Sight: Breaking the Isolation in Vietnamese and Beyond (2026.acl-long)

Copied to clipboard

Anh Trac Duc Dinh, Khang Hoang Nhat Vo, Tai Tien Ta, Vinh Cong Doan, Tho Quan

FineLAP: Taming Heterogeneous Supervision for Fine-grained Language-Audio Pretraining (2026.acl-long)

Challenge:	Adaptive Boundary-Token Fusion and a Morpheme-Aware Attention Bias are used to encode monosyllabic morphemes.
Approach:	They propose a morpheme-aware Transformer that augments a pretrained Vietnamese encoder with two lightweight inductive biases.
Outcome:	The proposed morpheme-aware Transformer outperforms strong baselines on Vietnamese POS, NER, and sentence-level classification benchmarks.

Copied to clipboard

Xiquan Li, Xuenan Xu, Ziyang Ma, Wenxi Chen, Haolin He, Qiuqiang Kong, Xie Chen

OCR-Memory: Optical Context Retrieval for Long-Horizon Agent Memory (2026.acl-long)

Challenge:	Existing audio-language models excel at clip-level understanding but struggle with frame-level tasks.
Approach:	They propose a novel training paradigm that advances both clip- and frame-level alignment in CLAP with heterogeneous data.
Outcome:	The proposed training paradigm improves both clip- and frame-level alignment in CLAP with heterogeneous data.

Copied to clipboard

Jinze Li, Yang Zhang, Xin Yang, Jiayi QU, Jinfeng Xu, Shuo Yang, Junhua Ding, Edith Cheuk-Han Ngai

From Past To Path: Masked History Learning for Next-Item Prediction in Generative Recommendation (2026.acl-long)

Challenge:	Existing LLMs are limited by text-context budgets, resulting in token-expensive storage of raw trajectories . Optical Context Retrieval Memory (OCR-Memory) renders historical tra-jectorios into images annotated with unique visual identifiers.
Approach:	They propose a framework that leverages the visual modality as a high-density representation of agent experience.
Outcome:	Optical Context Retrieval Memory (OCRM) renders historical trajectories into images annotated with unique visual identifiers.

Copied to clipboard

Kaiwen Wei, Kejun he, Xiaomian Kang, Jie Zhang, null Ymyang, Li Jin, Zhenyang Li, Jiang Zhong, Richard He Bai, Junnan Zhu

MuSe: Multi-Stage Graph Reasoning via Vision-Language Models (2026.acl-long)

Challenge:	Generative recommendation models inherently bias towards local contexts, failing to capture deeper historical dependencies necessary for understanding complex user intents.
Approach:	They propose a training framework that shifts the objective from simple next-step prediction to deep comprehension of history by entropy-guided masking policy and a curriculum learning scheduler to enhance the framework.
Outcome:	The proposed framework outperforms state-of-the-art generative models on three public datasets and shows that it is more accurate than current models.

Copied to clipboard

Guanyu Wang, Xu Chu, Zhijie Tan, Xinrong Chen, Tong Mo, Weiping Li

Where and What: Reasoning Dynamic and Implicit Preferences in Situated Conversational Recommendation (2026.acl-long)

Challenge:	Graph Neural Networks (GNNs) and graph transformers are inadequate for tasks with limited generalization.
Approach:	They propose a multi-stage graph reasoning framework based on vision-language models that incrementally samples and visualizes task-relevant subgraphs.
Outcome:	The proposed framework outperforms existing benchmarks in Graph-related tasks.

Copied to clipboard

Dongding Lin, Jian Wang, Yongqi Li, Wenjie Li

When Agents Look the Same: Quantifying Distillation-Induced Similarity in Tool-Use Behaviors (2026.acl-long)

Challenge:	Situated conversational recommendation (SCR) uses visual scenes grounded in specific environments and natural language dialogue to deliver contextually appropriate recommendations.
Approach:	They propose a framework that integrates scene transition estimation and Bayesian inverse inference to provide contextually appropriate recommendations.
Outcome:	The proposed framework achieves superiority over baselines on two representative benchmarks on dynamic scene transitions and implicit user intents.

Copied to clipboard

Chenghao Yang, Yuning Zhang, Zhoufutu Wen, Tao Gong, Jiaheng Liu, Qi Chu, Nenghai Yu

Mind the (DH) Gap! A Contrast in Risky Choices Between Reasoning and Conversational LLMs (2026.acl-long)

Challenge:	Existing metrics fail to distinguish mandatory behaviors required for task success from non-mandatory patterns that reflect a model’s autonomous preferences.
Approach:	They propose to use response pattern similarity and action graph similarity to isolate non-mandatory behaviors from mandatory behaviors.
Outcome:	Evaluating 18 models from 8 providers on -Bench and 2-Bench against Claude Sonnet 4.5, the authors find that within-family model pairs score 5.9 pp higher in response pattern similarity and action graph similarity .

Copied to clipboard

Luise Ge, Yongyan Zhang, Yevgeniy Vorobeychik

Beyond Single-View Detection: A Dual-Space Reasoning Framework for Interpretable Harmful Meme Understanding (2026.acl-long)

Challenge:	Large language models are increasingly used in decision support systems and workflows . traditional computational paradigms for decision-making under uncertainty choose an option that maximizes expected utility or payoff .
Approach:	They compare large language models as decision support systems and agentic workflows . they find that LLMs cluster into reasoning models and conversational models .
Outcome:	The proposed models differ in their ability to perform tasks and their ability in a human-like way.

Copied to clipboard

Wenqing Hou, Hongkui Tu, Ye Wang, Yue Zhang, Yuying Liu, Dong Zhu, Liqun Gao, Bin Zhou

From Experts to Bases: Orthogonal Subspace Mixture for Continual Multimodal Instruction Tuning (2026.acl-long)

Challenge:	Existing methods for identifying harmful memes rely on modal alignment or black-box classifiers . BPDMoE-Hate provides visual explanations for viewpoint selection and hierarchical structuring .
Approach:	They propose a framework that conceptualizes harmful meme detection as a process of "viewpoint decoupling and hierarchical fusion" they propose BPDMoE-Hate, which generates adversarial binary perspectives via VLMs and incorporates an adaptive viewpoint gating to facilitate viewpoint selection.
Outcome:	The proposed framework surpasses existing methods in performance and provides visual explanations for viewpoint selection and hierarchical structuring.

Copied to clipboard

Pei Chen, Xilai Wang, null Shiqixu, Zejian Li, Lingyun Sun

Distilling Large Embeddings via Hyperspherical Householder Quantization (2026.acl-long)

Challenge:	Existing parameter-efficient approaches to multimodal Continual Instruction Tuning suffer from knowledge interference and inefficient capacity expansion, limiting scalability.
Approach:	They propose a framework for multimodal Continual instruction tuning that decomposes adaptation weights into a globally shared pool of orthonormal bases to capture task-invariant knowledge.
Outcome:	Experiments show that MoBLoRA outperforms state-of-the-art methods while maintaining superior parameter efficiency.

Copied to clipboard

Yihang Wang, Bin Wu, Yueyang Su, Tianfu Zhang, Yiqi Du, Lei Yu, Jiafeng Guo, Xueqi Cheng

Masked by Consensus: Disentangling Privileged Knowledge in LLM Correctness (2026.acl-long)

Challenge:	Existing methods for quantizing large embeddings rely on Euclidean quantization, which is poorly aligned with the angular geometry induced by contrastive embeddment training.
Approach:	They propose a geometry-aware distillation method that compresses large embeddings into short discrete representations via iterative Householder transformations on the unit hypersphere.
Outcome:	The proposed method reduces decoding cost and maintains strong semantic retrieval accuracy.

Copied to clipboard

Tomer Ashuach, Shai Gretz, Yoav Katz, Yonatan Belinkov, Liat Ein-Dor

Unified Thinker: A General Reasoning Core for Image Generation (2026.acl-long)

Challenge:	Recent research suggests large language models encode meta-information about their own outputs.
Approach:	They investigate whether large language models possess similar privileged knowledge about answer correctness . they train correctness classifiers on question representations from a model’s hidden states and external models .
Outcome:	The proposed model outperforms peer-model models in factual knowledge tasks, but shows no advantage in math reasoning.

Copied to clipboard

Sashuai Zhou, Qiang Zhou, Jijin Hu, Hanqing Yang, Yue Cao, Junpeng Ma, Yinchao Ma, Jun Song, Tiezheng Ge, Cheng Yu, Bo Zheng, Zhou Zhao

Visual and Memory–Augmented Soccer Commentary Generation (2026.acl-long)

Challenge:	generative models struggle with logic-intensive instruction following, exposing a persistent reasoning–execution gap.
Approach:	They propose a task-agnostic reasoning architecture for general image generation . they propose pixel-level feedback to ground the Thinker's policy in pixel feedback .
Outcome:	The proposed system significantly improves image reasoning and generation quality.

Copied to clipboard

Haoran Sun, Natthawut Kertkeidkachorn, Kiyoaki Shirai

InstructDiff: Domain-Adaptive Data Selection via Contrastive Entropy for Efficient LLM Fine-Tuning (2026.acl-long)

Challenge:	Existing datasets produce incomplete commentary that lacks semantic richness and does not convey full visual information present in standard video clips.
Approach:	They propose a method that transforms incomplete annotations into MatchText, a semantically complete and structurally standardized dataset.
Outcome:	The proposed model outperforms baselines on constructed soccer commentary datasets.

Copied to clipboard

Junyou Su, He Zhu, Xiao Luo, Liyu Zhang, Hong-Yu Zhou, Yun Chen, Peng Li, Yang Liu, Guanhua Chen

Jakiro: Boosting Speculative Decoding via Decoupled MoE (2026.acl-long)

Challenge:	Existing data selection methods suffer from severe domain specificity . existing methods for general instruction-following fail on reasoning tasks .
Approach:	They propose a framework that operationalizes contrastive entropy as a domain-adaptive selection criterion through warmup calibration, bi-directional NLL filtering, and entropic-based ranking.
Outcome:	Experiments show that InstructDiff outperforms baseline training on reasoning tasks while using only 10% of the data.

Copied to clipboard

Haiduo Huang, Fuwei Yang, Zhenhua Liu, Pengju Ren

Beyond Word Boundaries: A Hebrew Coreference Benchmark and an Evaluation Protocol for Morphologically Complex Text (2026.acl-long)

Challenge:	Existing methods to accelerate large language model inference have a fundamental limitation: candidates at the same tree layer share identical feature representations, constraining diversity and diminishing overall effectiveness.
Approach:	They propose a decoupled mixture of experts (MoE) into a draft model to generate diverse tokens from distinct feature spaces.
Outcome:	The proposed approach achieves significant speedups over strong baselines, with notable improvements in non-greedy scenarios where token diversity is crucial.

Copied to clipboard

Refael Shaked Greenfeld, Reut Tsarfaty

Reasoning with Ontology Graph: Toward Type-Constrained Knowledge Graph Question Answering (2026.acl-long)

Challenge:	CR methods originally designed for English struggle with Morphologically Rich Languages (MRLs) a single token in Hebrew may consist of multiple anaphors, and word/morpheme boundary discrepancies make mention detection and coreference resolution difficult in MRLs.
Approach:	They propose a CR dataset that identifies mentions at word, sub-word and multi-word levels and an evaluation protocol that directly addresses word/morpheme boundary discrepancies.
Outcome:	The proposed evaluation protocol directly addresses word/morpheme boundary discrepancies in Modern Hebrew, an MRL rich with complex words and pronominal clitics.

Copied to clipboard

Yongxue Shan, Jie Peng, Zixuan Dong, Fei Hu, Xiaodong Wang

Knowledge Representation

Challenge:	Existing knowledge graph question answering methods rely on LLM-induced type systems with inconsistent granularity or perform multi-hop reasoning without explicit target-type constraints.
Approach:	They propose a type-constrained knowledge graph question answering framework that reasons over a relation-centric ontology graph.
Outcome:	The proposed framework achieves state-of-the-art and produces ontology-grounded reasoning chains with substantial Hit@1 gains.

Visual Self-Fulfilling Alignment: Shaping Safety-Oriented Personas via Threat-Related Images (2026.acl-long)

Copied to clipboard

Qishun Yang, Shu Yang, Lijie Hu, Di Wang

GRASPrune: Global Gating for Budgeted Structured Pruning of Large Language Models (2026.acl-long)

Challenge:	Existing methods require explicit safety labels or contrastive data, yet visual inputs enable harmful outputs.
Approach:	They propose a visual self-fulfilling alignment mechanism that fine-tunes vision-language models on neutral VQA tasks without any safety labels.
Outcome:	The proposed approach reduces attack success rate, improves response quality, and mitigates over-refusal while preserving general capabilities.

Copied to clipboard

Ziyang Wang, Jiangfeng Xiao, Chuan Xiao, Ruoxiang LI, Rui Mao, Jianbin Qin

All That Glisters Is Not Gold: A Benchmark for Reference-Free Counterfactual Financial Misinformation Detection (2026.acl-long)

Challenge:	Large language models are expensive to serve because dense FFN blocks, multi-head attention, and KV caches dominate memory.
Approach:	They propose a global budgeted structured pruning framework that prunes FFN channels and attention KV head groups under a single global parameter budget.
Outcome:	The proposed model removes 50% of parameters and achieves 12.18 perplexity on WikiText-2 while maintaining competitive average zero-shot accuracy on five downstream benchmarks.

Copied to clipboard

Yuechen Jiang, Zhiwei Liu, Yupeng Cao, Yueru He, Ziyang Xu, Chen Xu, Zhiyang Deng, Prayag Tiwari, Xi Chen, Alejandro Lopez-Lira, Jimin Huang, Junichi Tsujii, Sophia Ananiadou

Musical Score Understanding Benchmark: Evaluating Large Language Models’ Comprehension of Complete Musical Scores (2026.acl-long)

Challenge:	RFC-Bench evaluates large language models on financial misinformation under realistic news . current models struggle to maintain coherent belief states without external grounding, study finds .
Approach:	They propose a benchmark for evaluating large language models on financial misinformation under realistic news.
Outcome:	The proposed model performs better when context is available, while reference-free settings expose significant weaknesses.

Copied to clipboard

Congren Dai, Yue Yang, Krinos Li, Huichi Zhou, Shijie Liang, Zhang Bo, Enyang Liu, Ge Jin, Hongran An, Haosen Zhang, Peiyuan Jing, KinHei Lee, Zhenxuan Zhang, Xiaobing Li, Maosong Sun

EchoVLM: Dynamic Mixture-of-Experts Vision-Language Model for Universal Ultrasound Intelligence (2026.acl-long)

Challenge:	Existing benchmarks for musical score understanding are narrow in scope, focusing on isolated fragments, short excerpts, or multiple-choice formulations, rather than supporting holistic reasoning over entire scores.
Approach:	They propose a benchmark for score-level musical understanding across textual and visual modalities.
Outcome:	The musical score understanding benchmark contains 1,800 question-answer pairs from works by Bach, Beethoven, Chopin, Debussy, and others.

Copied to clipboard

Chaoyin She, Ruifang Lu, Lida Chen, Wei Wang, Qinghua Huang

Critical-CoT: A Robust Defense Framework against Reasoning-Level Backdoor Attacks in Large Language Models (2026.acl-long)

Challenge:	Ultrasound is the preferred early cancer screening modality due to non-ionizing radiation, cost-effectiveness, and real-time imaging.
Approach:	They propose to use ultrasound-tailored vision-language models with a mixture-of-experts architecture to train ultrasound-specific knowledge across seven anatomical systems.
Outcome:	The proposed model outperforms Qwen2-VL by 7.58 BLEU-1 and 3.45 ROUGE-1 points in report generation.

Copied to clipboard

Vu Tuan Truong, Long Bao Le

How Do LLMs and VLMs Understand Viewpoint Rotation Without Vision? An Interpretability Study (2026.acl-long)

Challenge:	Large Language Models (LLMs) are vulnerable to backdoors that use long-form reasoning to generate a specific word, choice, or class.
Approach:	They propose a mechanism that allows LLMs to develop critical thinking behaviors and detect backdoors by a two-stage fine-tuning.
Outcome:	The proposed mechanism exhibits strong cross-domain and cross-task generalization.

Copied to clipboard

Zhen Yang, Ping Jian, Zhongbin Guo, Zuming Zhang, Chengzhi Li, Yonghong Deng, Xinyue Zhang, Wenpeng Lu

Sycophants in the Courtroom: Are LLMs Fragile to Juridical Authority and Evolving Legal Standards? (2026.acl-long)

Challenge:	Existing studies on spatial intelligence from the perspective of visual-spatial intelligence have not explored whether visual intelligence alone is sufficient to endow models with spatial intelligence.
Approach:	They propose to use a linguistic perspective to investigate spatial intelligence from a theoretical perspective.
Outcome:	The proposed model performs poorly on the proposed dataset while human can easily achieve 100% accuracy.

Copied to clipboard

Lorenzo Molfetta, Alessio Cocchieri, Luca Ragazzi, Ilaria Bartolini, Marco Patella, Gianluca Moro

Challenge:	Recent advances have seen large language models (LLMs) achieve remarkable performance across high-stakes specialized domains.
Approach:	They propose a diagnostic framework that evaluates legal reasoning against medical baselines along four axes (knowledge recall, grounding, confidence, and robustness) they uncover a sharp domain asymmetry when applied to a benchmark that encodes temporal validity and normative relationships.
Outcome:	The proposed framework evaluates legal reasoning against medical baselines along four axes (knowledge recall, grounding, confidence, and robustness) it shows that legal LLMs struggle to assess when retrieved citations are useful or misleading, exhibiting overconfidence in perturbed contexts and sensitivity to superficial formatting cues.

Valid Survey Simulations with Limited Human Data: The Roles of Prompting, Fine-Tuning, and Rectification (2026.acl-long)

Copied to clipboard

Stefan Krsteski, Giuseppe Russo, Serina Chang, Robert West, Kristina Gligorić

ODUTQA-MDC: A Task for Open-Domain Underspecified Tabular QA with Multi-turn Dialogue-based Clarification (2026.acl-long)

Challenge:	Large language models (LLMs) are a cost-effective and time-consuming way to capture public opinion and behavior, but their outputs are often biased and yield invalid estimates.
Approach:	They propose to use large language models to generate survey responses and rectification methods that debias population estimates to find out how human responses are best allocated between them.
Outcome:	The proposed methods reduce bias below 5% and increase sample size by up to 14% under a fixed budget.

Copied to clipboard

Zhensheng Wang, ZhanTeng Lin, Wenmian Yang, Kun Zhou, Yiquan Zhang, Weijia Jia

CuBridge: An LLM-Based Framework for Understanding and Reconstructing High-Performance Attention Kernels (2026.acl-long)

Challenge:	Existing approaches to tabular QA are limited to closed-domain scenarios . existing approaches do not solve the core challenge of generating correct answers without user clarification .
Approach:	They propose a benchmark to tackle underspecified or uncertain queries in tabular question answering . they propose ODUTQA-MDC task and a multi-agent framework to detect ambiguities .
Outcome:	The proposed framework excels at detecting ambiguities, clarifying them through dialogue, and refining answers.

Copied to clipboard

Xing Ma, Yangjie Zhou, Wu Sun, Zihan Liu, Jingwen Leng, Yun Lin, Shixuan Sun, Minyi Guo, Jin Song Dong

Learning from Contrasts: Synthesizing Reasoning Paths from Diverse Search Trajectories (2026.acl-long)

Challenge:	Existing approaches to support diverse attention variants trade performance for flexibility . expert-written kernels achieve high efficiency but are difficult to adapt .
Approach:	They propose a framework that adapts expert-written attention kernels to GPUs . they use a structured lift–transfer–lower workflow to make execution explicit .
Outcome:	The proposed framework outperforms existing frameworks and compilers on diverse variants and GPU platforms.

Copied to clipboard

Peiyang Liu, Zhirui Chen, Xi Wang, Di Liang, Youru Li, Zhi Cai, Wei Ye

Defenses Against Prompt Attacks Learn Surface Heuristics (2026.acl-long)

Challenge:	MCTS methods retain only the single highest-reward trajectory, discarding comparative signals present in the many explored paths.
Approach:	They propose a framework that transforms supervision extraction into a synthesis procedure.
Outcome:	The proposed framework matches or exceeds baselines on 60K CRPS-synthesized examples on out-of-domain benchmarks.

Copied to clipboard

Li Li, Chenxiao Yu, Zhiyu Ni, Hao Li, Charith Peris, Chaowei Xiao, Yue Zhao

Memory-Augmented LLM-based Multi-Agent System for Automated Feature Generation on Tabular Data (2026.acl-long)

Challenge:	Large language models (LLMs) are increasingly deployed in security-sensitive applications . recent defenses rely on supervised fine-tuning with benign and malicious labels . position bias arises when benign content placed later in a prompt is rejected at much higher rates .
Approach:	They analyze three recurring shortcut behaviors induced by supervised fine-tuning . position bias arises when benign content placed later in a prompt is rejected . token trigger bias occurs when strings common in attack data raise rejection probability .
Outcome:	The proposed model overrides intended logic when adversarial instructions appear . the proposed model has low rejection rates but narrow correlations in defense data .

Copied to clipboard

FengXian Dong, Zhi Zheng, Xiao Han, Wei Chen, Jingqing Ruan, Tong Xu, Yong Chen, Enhong Chen

Learning from the Irrecoverable: Error-Localized Policy Optimization for Tool-Integrated LLM Reasoning (2026.acl-long)

Challenge:	Existing methods for automated feature generation rely on predefined operator libraries and do not incorporate feature semantics, limiting their ability to produce high-quality features.
Approach:	They propose a Memory-Augmented LLM-based Multi-Agent System (MALMAS) that decomposes the generation process into agents with distinct responsibilities.
Outcome:	The proposed method extracts informative features from raw tabular data without manual intervention and is crucial for accurate, generalizable machine learning.

Copied to clipboard

Qiao Liang, Yuke Zhu, Chao Ge, Lei Yang, Ying Shen, Bo Zheng, Sheng Guo

Learning from Evolving Training Dynamics: An Entropy-Maximizing Data Curation Strategy for LLM Supervised Post-Training (2026.acl-long)

Challenge:	Tool-integrated reasoning (TIR) enables LLM agents to solve tasks through planning, tool use, and iterative revision, but outcome-only reinforcement learning suffers from sparse, delayed rewards and weak step-level credit assignment.
Approach:	They propose a tool-integrated reasoning approach that localizes the first irrecoverable step and leverages it for fine-grained credit assignment.
Outcome:	The proposed algorithm outperforms strong Agentic RL benchmarks in math, science QA, and code execution with additional gains in Pass@K and Major@K scaling, rollout ranking quality, and tool-call efficiency.

Copied to clipboard

Mengxiang Zhang, Lingyuan Liu

Challenge:	EVO-Curate is a dynamic data curation framework that synchronizes sample complexity with the maturing capacity of the Large Language Model (LLM).
Approach:	They propose a dynamic data curation framework that synchronizes sample complexity with maturing capacity of the Large Language Model (LLM) they use an Adaptive Dynamics Measurer to synthesize instantaneous difficulty and historical variability into a multidimensional utility score.
Outcome:	The proposed framework outperforms standard training baselines and traditional CL methods on instruction following, mathematical reasoning, and code generation architectures while maintaining manageable computational overhead.

Reinforcement Learning on Pre-Training Data (2026.acl-long)

Copied to clipboard

Siheng Li, Kejiao Li, Zenan Xu, Guanhua Huang, Kun Li, Haoyuan Wu, null Wujiajia, Zihao Zheng, Chenchen Zhang, Kun Shi, Xue Gong, Qi Yi, Ruibin Xiong, Tingqiang Xu, Yuhao Jiang, Jianfeng Yan, Yuyuan Zeng, Guanghui Xu, Jinbao Xue, Zhijiang xu, Zheng Fang, Shuai LI, Qibin Liu, Xiaoxue Li, Zhuoyu Li, Yangyu Tao, Fei Gao, Cheng Jiang, Bochao Wang, Kai Liu, Jianchen Zhu, Wai Lam, Bo Zhou, Di Wang

Limited Linguistic Diversity in Embodied AI Datasets (2026.acl-long)

Challenge:	Recent progress in large language models is driven by scaling of training compute through pre-training with nexttoken prediction (NTP) or post-training (RL) Pre-training using NTP enables models to acquire extensive knowledge and skills from general data, but it suffers from data inefficiency and catastrophic forgetting in continual learning settings.
Approach:	They propose to scale training compute through pre-training with next-token prediction (NTP) or post-training by scaling reinforcement learning (RL) to improve learning from general data.
Outcome:	Experiments on multiple benchmarks and models show that the proposed approach improves continual pre-training and provides a strong foundation for post-training on Qwen3-8B-Base.

Copied to clipboard

Selma Liliane Wanna, Agnes Luhtaru, Jonathan Salfity, Ryan Barron, Juston Moore, Cynthia Matuszek, Mitch Pryor

Challenge:	Language is a key modality in Vision-Language-Action (VLA) models, yet its linguistic characteristics are poorly documented.
Approach:	They propose to audit VLA corpora to characterize what kinds of instructions they contain . they quantify instruction language along complementary dimensions including lexical variety, duplication and overlap, semantic similarity, and syntactic complexity.
Outcome:	The proposed dataset audit of several widely used VLA datasets shows that many datasets rely on highly repetitive, template-like commands with limited structural variation yielding a narrow distribution of instruction forms.

Towards Mitigating Modality Bias in Vision-Language Models for Temporal Action Localization (2026.acl-long)

Copied to clipboard

Jiaqi Li, Guangming Wang, Shuntian Zheng, Minzhe Ni, Xiaoman Lu, Guanghui Ye, Yu Guan

CreditDecoding: Accelerating Parallel Decoding in Diffusion Large Language Models with Trace Credit (2026.acl-long)

Challenge:	Existing vision-language models overemphasize linguistic priors, leading to modality bias.
Approach:	They propose a vision-language aggregation framework that mitigates modality bias in TAL by preserving vision as the dominant signal while adaptively exploiting language only when beneficial.
Outcome:	Experiments on THUMOS14 show that the proposed model outperforms state-of-the-art models by up to 3.2% mAP.

Copied to clipboard

Kangyu Wang, Zhiyun Jiang, Haibo Feng, Weijia Zhao, Lin Liu, Jianguo Li, Zhenzhong Lan, Weiyao Lin

Question Difficulty Estimation for Large Language Models via Answer Plausibility Scoring (2026.acl-long)

Challenge:	Diffusion large language models generate text through iterative denoising with bidirectional attention, enabling richer contextual dependencies.
Approach:	They propose a training-free parallel decoding method that fuses Trace Credit with current logits to boost the confidence of correct but underconfident tokens.
Outcome:	The proposed method achieves 5.48 times speedup with +0.48 accuracy on LLaDA-8B and is orthogonal to mainstream inference optimizations.

Copied to clipboard

Jamshid Mozafari, Bhawna Piryani, Adam Jatowt

Vector Calligrapher: Generating Scalable Vector Graphics via Structured Linguistic Supervision (2026.acl-long)

Challenge:	Existing approaches to estimate question difficulty rely on readability formulas, retrieval-based signals, or popularity statistics.
Approach:	They propose a method that estimates question difficulty by computing the entropy of plausibility scores over candidate answers.
Outcome:	The proposed method outperforms baselines across four QA datasets and shows strong robustness across hyperparameter variations and question types.

Copied to clipboard

Bo Zhou, Xikang Chen, Yan Gong, Yin Zhang

CoQuIR: A Comprehensive Benchmark for Code Quality-Aware Information Retrieval (2026.acl-long)

Challenge:	Existing approaches to generate SVG-based fonts struggle with semantic ambiguity and inefficiency . edward mcginley: generic text tokenizers fragment coordinate-dense SVG XML into excessively long sequences .
Approach:	They propose a system that treats SVG generation as a conditional language modeling task . they propose linguistic supervision framework that decomposes typographic style into interpretable linguistic dimensions .
Outcome:	The proposed system improves CLIP score by +23% while reducing geometric error by 48% and boosts generation efficiency by 18% Command-per-Token (C/T) ratio.

Copied to clipboard

Jiahui Geng, Fengyu Cai, Shaobo Cui, Qing Li, Liangwei Chen, Chenyang Lyu, Haonan Li, Derui Zhu, Alexander Pretschner, Heinz Koeppl, Fakhri Karray

Sherry: Hardware-Efficient 1.25-Bit Ternary Quantization via Fine-grained Sparsification (2026.acl-long)

Challenge:	Existing benchmarks focus on functional relevance while neglecting code quality.
Approach:	They propose a multilingual benchmark to evaluate quality-aware code retrieval . they include fine-grained quality annotations over 42,725 queries and 134,907 code snippets .
Outcome:	The proposed benchmarks show that state-of-the-art models fail to separate buggy or insecure code from robust counterparts.

Copied to clipboard

Hong Huang, Decheng Wu, Qiangqiang Hu, Guanghua Yu, Jinhai Yang, Jianchen Zhu, Xue Liu, Dapeng Wu

DMN: A Compositional Framework for Jailbreaking Multimodal LLMs with Multi-Image Inputs (2026.acl-long)

Challenge:	ternary quantization is a powerful solution for resource-constrained edge devices . current implementations suffer from a fundamental misalignment with commodity hardware .
Approach:	They propose a hardware-efficient ternary quantization framework that packs weights into five bits to restore power-of-two alignment.
Outcome:	The proposed framework reduces weights to -1, 0, +1 while preserving power-of-two alignment.

Copied to clipboard

Wenzhuo Xu, Zhipeng Wei, Zonghao Ying, Deyue Zhang, Dongdong Yang, Xiangzheng Zhang, Quanchen Zou

PIAST: Rapid Prompting with In-context Augmentation for Scarce Training data (2026.acl-long)

Challenge:	Existing jailbreak methods only use a single image, restricting the attack space . Existing frameworks only use single image to distribute harmful requests across multiple images .
Approach:	They propose a compositional jailbreak framework that leverages Distributed instruction, Multimodal evidence and a Number chain task to fully enhance the jailbreak performance.
Outcome:	The proposed framework achieves attack success rates of over 90% on GPT-4o, Gemini-2.5-pro and Claude Sonnet 4 .

Copied to clipboard

Paweł Batorski, Paul Swoboda

ProgressLM: Towards Progress Reasoning in Vision-Language Models (2026.acl-long)

Challenge:	Existing prompt generation methods are impractical in time and data constrained settings.
Approach:	They propose a fast automatic prompt construction algorithm that augments human instructions by generating a small set of few shot examples.
Outcome:	The proposed method outperforms existing prompting methods on classification, simplification, and MedQA.

Copied to clipboard

Jianshu Zhang, Chengxuan Qian, Haosen Sun, Haoran Lu, Dingcheng Wang, Letian Xue, Han Liu

Forest Before Trees: Latent Superposition for Efficient Visual Reasoning (2026.acl-long)

Challenge:	Existing models for task progress estimation lack long-horizon and dynamic reasoning . estimating how much of a task has been completed requires long-term reasoning based on partial information.
Approach:	They propose a benchmark for evaluating progress reasoning from a single observation . they instantiate a two-stage paradigm that combines episodic retrieval with mental simulation .
Outcome:	The proposed benchmark improves on 14 VLMs on a small scale and shows common failure patterns.

Copied to clipboard

Yubo Wang, Juntian Zhang, Yichen Wu, Yankai Lin, Nils Lukas, Yuhan Liu

Beyond Pedagogical Principles: Multi-Horizon Preference Optimization for Efficient Socratic Tutoring (2026.acl-long)

Challenge:	Recent latent reasoning methods suffer from a bandwidth bottleneck . explicit textual rationales suffer from premature semantic collapse .
Approach:	They propose a new paradigm that reformulates visual deduction via Dynamic Windowed Alignment Learning.
Outcome:	The proposed paradigm achieves state-of-the-art performance among latent reasoning methods surpassing the strong baseline Monet by 5.03% on average.

Copied to clipboard

Xin Shi, Chao Zhang, Yifan Zhu, Xueqiao Zhang, Yawei Luo

Protecting Language Models Against Unauthorized Distillation through Trace Rewriting (2026.acl-long)

Challenge:	Existing methods for developing LLMs are constrained by static data or sparse reward signals in online settings.
Approach:	They propose a framework that iteratively refines tutor agents using a multi-horizon reward function within a dynamic teacher-student simulation environment.
Outcome:	The proposed framework improves model performance and balances principles and effectiveness compared to baselines.

Copied to clipboard

Xinhang Ma, William Yeoh, Ning Zhang, Yevgeniy Vorobeychik

The Confidence Dichotomy: Analyzing and Mitigating Miscalibration in Tool-Use Agents (2026.acl-long)

Challenge:	Knowledge distillation is a widely adopted technique for transferring capabilities from LLMs to smaller, more efficient student models.
Approach:	They propose methods for modifying teacher-generated reasoning traces to achieve two objectives that deter unauthorized distillation: (1) anti-distillation, or degrading the training usefulness of query responses; and (2) API watermarking, which embeds verifiable signatures in student models.
Outcome:	The proposed method achieves strong anti-distillation effect while maintaining or even improving teacher performance.

Copied to clipboard

Weihao Xuan, Qingcheng Zeng, Heli Qi, Yunze Xiao, Junjue Wang, Naoto Yokoya

XOXO: Stealthy Cross-Origin Context Poisoning Attacks against AI Coding Assistants (2026.acl-long)

Challenge:	a fundamental pillar of trustworthiness is calibration, which refers to an agent’s ability to express confidence that reliably reflects its actual performance.
Approach:	They propose a reinforcement learning framework that jointly optimizes task accuracy and calibration, supported by a holistic benchmark of reward designs.
Outcome:	The proposed framework improves calibration across tool types and shows that trained agents achieve superior calibration and exhibit robust generalization from local training environments to noisy web settings and to distinct domains such as mathematical reasoning.

Copied to clipboard

Adam Štorek, Mukur Gupta, Noopur Bhatt, Aditya Gupta, Janie Kim, Prashast Srivastava, Suman Jana

SafeMCP: Proactive Power Regulation for LLM Agent Defense via Environment-Grounded Look-Ahead Reasoning (2026.acl-long)

Challenge:	Cross-Origin Context Poisoning (XOXO) exploits this automatic context inclusion by subtly manipulating code without changing its semantics.
Approach:	They propose a novel attack that exploits automatic context inclusion by subtly manipulating code without changing its semantics.
Outcome:	The proposed attack achieves 73.20% success rates against eight state-of-the-art models including GPT 4.1 and Claude 3.5 Sonnet v2 and vulnerability injection rates up to 66.67%.

Copied to clipboard

Lichao Wang, ZhaoXing Ren, Tianzhuo Yang, Jiaming Ji, Chi Harold Liu, Yaodong Yang, Juntao Dai

C2: Scalable Rubric-Augmented Reward Modeling from Binary Preferences (2026.acl-long)

Challenge:	Large Language Model (LLM) agents are expanding their action spaces to operate in complex environments.
Approach:	They propose a server-side defense plugin that constrains tool acquisition via predictive reasoning regarding future safety risks.
Outcome:	Experiments on PowerSeeking Bench, ToolEmu, and AgentHarm show that SafeMCP achieves a safe equilibrium, effectively mitigating risks while preserving agent utility.

Copied to clipboard

Akira Kawabata, Saku Sugawara

Challenge:	Existing methods for rubric-augmented verification require costly rubric annotations, limiting scalability.
Approach:	They propose a framework that allows a reward model to collaborate critically with a rubric generator trained solely from binary preferences.
Outcome:	The proposed framework outperforms reasoning reward models trained on binary preferences with gains of up to 6.5 points on RM-Bench and 6.0 points length-controlled win rate on AlpacaEval 2.0.

LLaDA 1.5: Variance-Reduced Preference Optimization for Large Language Diffusion Models (2026.acl-long)

Copied to clipboard

Fengqi Zhu, Rongzhen Wang, Shen Nie, Xiaolu Zhang, Chunwei Wu, Jun Zhou, Yankai Lin, Ji-Rong Wen, Chongxuan Li

HarmRLVR: Weaponizing Verifiable Rewards for Harmful LLM Alignment (2026.acl-long)

Challenge:	Masked diffusion language models have achieved significant progress in language modeling . however, the systematic analysis and empirical validation of their alignment on general tasks remains underexplored.
Approach:	They propose a framework that analyzes the bias and variance of preference optimization loss and gradient based on Direct Preference Optimization.
Outcome:	The proposed model outperforms its SFT-only predecessor on general benchmarks . it consistently outperformed other strong language models and ARMs on general tasks .

Copied to clipboard

Yuexiao Liu, Lijun Li, Xingjun Wang, Jing Shao

Just Pass Twice: Efficient Token Classification with LLMs for Zero-Shot NER (2026.acl-long)

Challenge:	Recent advances in Reinforcement Learning with Verifiable Rewards (RLVR) have gained significant attention due to their objective and verifiably verifier reward signals.
Approach:	They propose to exploit RLVR for alignment reversibility by using GRPO to reverse alignment with merely 64 harmful prompts without responses.
Outcome:	The proposed method outperforms fine-tuning and RLHF in reasoning and code generation tasks while maintaining general capabilities.

Copied to clipboard

Ahmed Ewais, Ahmed Hashish, Amr Ali

Lifting Optimized Binaries to Canonical Compiler IR via Structure-Aware Retrieval and Iterative Verification (2026.acl-long)

Challenge:	Existing methods to name entity recognition use autoregressive decoding, hallucinated entities, and formatting errors.
Approach:	They propose a method that allows causal LLMs to perform discriminative token classification with full bidirectional context.
Outcome:	The proposed method surpasses the previous best method on zero-shot NER benchmarks by +7.9 F1 on average across CrossNER and MIT benchmarks.

Copied to clipboard

Xiaoao Zhu, Jie Ren, Zhiqiang Li, Jie Zheng, Zhanyong Tang, Zheng Wang

Provably Safe Offline-to-Online RL: Decoupling Learning from Data-Driven Safety Enforcement (2026.acl-long)

Challenge:	Existing methods for decompiling binary code are brittle due to compiler optimizations that distort control-flow and data-flow structure.
Approach:	They propose a system that lifts optimized binaries to canonical compiler intermediate representation (IR) BRIDGE uses control-flow-aware retrieval-augmented generation with feedback-driven verification .
Outcome:	The proposed system outperforms seven baselines on humanEval-Decompile and MBPP, lifting x86-64 and ARM64 binaries to LLVM IR.

Copied to clipboard

Kaitong Cai, Jusheng Zhang, Keze Wang

SHAPE: Unifying Safety, Helpfulness and Pedagogy for Educational LLMs (2026.acl-long)

Challenge:	Hybrid offline–online reinforcement learning (O2O RL) promises both sample efficiency and robust exploration, but suffers from instability due to distribution shift between offline and online data.
Approach:	They propose a framework that decouples policy optimization from safety enforcement . they propose dynamic curricula that gradually extend temporal horizons and anneal offline–online data mixing .
Outcome:	The proposed framework preserves the exploratory value of online interactions without collapsing to conservative policies.

Copied to clipboard

Sihang Zhao, Kangrui Yu, Youliang Yuan, Pinjia He, Hongyi Wen

Evolving Sparsity: Leveraging Token Importance Dynamics for Efficient LLM Decoding with Sparse Attention (2026.acl-long)

Challenge:	Existing educational LLMs are vulnerable to pedagogical jailbreaks where students use answer-inducing prompts to elicit solutions rather than scaffolded instructions.
Approach:	They propose a graph-augmented tutoring pipeline that infers prerequisite concepts from queries and identifies mastery gaps.
Outcome:	The proposed method improves safety under two pedagogical jailbreak scenarios while maintaining near-ceiling helpfulness under the same evaluation protocol.

Copied to clipboard

Ruizi Han, Miao Zhang, Ziyue Qiao, Liqiang Nie

MOA: Multi-Objective Alignment for Role-Playing Agents (2026.acl-long)

Challenge:	Efficient long-context inference remains a major challenge for large language models (LLMs), as the cost of attention computation during auto-regressive decoding grows linearly with the context length.
Approach:	They propose to model token importance as a dynamic process that evolves over decoding steps and propagates through model layers.
Outcome:	The proposed method outperforms baseline sparse attention methods and achieves speedups of up to 5.36 for attention latency and 2.33 for end-to-end decoding.

Copied to clipboard

Chonghua Liao, Ke Wang, Yuchuan Wu, Ruoran Li, Fei Huang, Yongbin Li

Challenge:	Prior work on role-playing agents relies on supervised fine-tuning or reinforcement learning with scalarized rewards, but these approaches do not address the coordination of multiple reward dimensions during optimization.
Approach:	They propose a reinforcement-learning framework that enables multi-dimensional, fine-grained rubric optimization for general RPAs.
Outcome:	Experiments on PersonaGym and RoleMRC show that MOA improves multi-dimensional role-playing performance over supervised and standard RL baselines.

Losing our Tail, Again: (Un)Natural Selection & Multilingual LLMs (2026.acl-long)

Copied to clipboard

Eva Vanmassenhove

From Verbatim to Gist: Distilling Pyramidal Multimodal Memory via Semantic Information Bottleneck for Long-Horizon Video Agents (2026.acl-long)

Challenge:	a few million years ago, we lost our tails, and with them, the narratives and identities they carry.
Approach:	They argue that multilingual large language models are threatening our linguistic diversity . they argue that models collapse towards what is likely, driven by statistical biases .
Outcome:	a new paper shows that models can collapse and lead to loss of linguistic diversity . the authors argue that models are a field that values and protects multilingual diversity - a key challenge for the field .

Copied to clipboard

Niu Lian, Yuting Wang, Hanshu Yao, Jinpeng Wang, Bin Chen, Yaowei Wang, Min Zhang, Shu-Tao Xia

Chain-of-Memory: Lightweight Memory Construction with Dynamic Evolution for LLM Agents (2026.acl-long)

Challenge:	Existing multimodal large language models struggle with long-horizon video understanding due to limited context windows and static memory mechanisms that fail to mirror human cognitive efficiency.
Approach:	They propose a pyramidal multimodal memory architecture grounded in Fuzzy-Trace Theory that structures memory hierarchically into a Sensory Buffer, Episodic Stream, and Symbolic Schema.
Outcome:	The proposed architecture achieves state-of-the-art on both offline and streaming tasks, demonstrating robust generalization and validating the effectiveness of cognition-inspired memory organization.

Copied to clipboard

Xiucheng Xu, Bingbing Xu, Tian Xueyun, Zihe Huang, Rongxin Chen, Li Yunfan, Huawei Shen

WebCoderBench: Benchmarking Web Application Generation with Comprehensive and Interpretable Evaluation Metrics (2026.acl-long)

Challenge:	Existing paradigms for large language model (LLM) agents use memory construction and retrieval-augmented generation.
Approach:	They propose a framework that advocates for a paradigm shift toward lightweight construction paired with sophisticated utilization.
Outcome:	Experiments show that CoM outperforms baselines with accuracy gains of 7.5%–10.4% while reducing computational overhead to approximately 2.7% of token consumption and 6.0% of latency compared to complex memory architectures.

Copied to clipboard

Chenxu Liu, Yingjie Fu, Wei Yang, Ying Zhang, Tao Xie

Efficient Test-Time Scaling of Multi-Step Reasoning by Probing Internal States of Large Language Models (2026.acl-long)

Challenge:	Web applications (web apps) are a key arena for large language models to demonstrate their code generation capabilities and commercial potential.
Approach:	a new benchmark for large language models (LLMs) is designed to provide real-world user requirements and generalizable evaluation metrics.
Outcome:	a new benchmark for large language models (LLMs) provides a real-world, generalizable, and interpretable evaluation score . the benchmark measures user requirements, expression styles and human-preference-aligned weights . a web application can be used to demonstrate its commercial potential, authors say .

Copied to clipboard

Jingwei Ni, Ekaterina Fadeeva, Tianyi Wu, Mubashara Akhtar, Jiaheng Zhang, Elliott Ash, Markus Leippold, Timothy Baldwin, See-Kiong Ng, Artem Shelmanov, Mrinmaya Sachan

Adaptive Prompt Structure Factorization: A Framework for Self-Discovering and Optimizing Compositional Prompt Programs (2026.acl-long)

Challenge:	Existing verification approaches, such as Process Reward Models, are computationally expensive and limited to specific domains.
Approach:	They propose a transformer-based probe that uses internal states of frozen LLMs to estimate credibility of reasoning steps during generation.
Outcome:	The proposed probes match or exceed PRMs that are up to 810 larger.

Copied to clipboard

Haoyue Liu, Zhichao Wang, Yongxin Guo, Haoran Shou, Xiaoying Tang

Cognitive Alpha Mining via LLM-Driven Code-Based Evolution (2026.acl-long)

Challenge:	Large language models (LLMs) exhibit strong capabilities in reasoning, coding, and complex generation, yet their performance is highly sensitive to prompt design.
Approach:	They propose an API-only framework that decomposes a single prompt into semantic factors and updates selected factors while freezing the rest.
Outcome:	The proposed framework outperforms strong baselines, improves accuracy by up to +4.29 percentage points on average, and reduces optimization cost by 45–87% tokens on MultiArith while reaching peak validation in 1 step.

Copied to clipboard

Fengyuan Liu, Yi Huang, Sichun Luo, Yuqi Wang, Yazheng Yang, Xinye Li, Zefa Hu, Junlan Feng, Qi Liu

Why Do Emotions Change? Appraisal-Guided Reasoning for Emotion–Cause Triplet Extraction in Conversations (2026.acl-long)

Challenge:	Existing approaches to finding effective predictive signals from financial data are limited by their complexity and low signal-to-noise ratio.
Approach:	They propose a framework that combines code-level alpha representation with LLM-driven reasoning and evolutionary search.
Outcome:	The proposed framework combines code-level alpha representation with LLM-driven reasoning and evolutionary search.

Copied to clipboard

Qiao Liang, Ying Shen, Yao Liu, Tiantian Chen, Lin Zhang

KV-Embedding: Training-free Text Embedding via Internal KV Re-routing in Decoder-only LLMs (2026.acl-long)

Challenge:	Existing methods for multi-turn, multi-speaker multimodal affect understanding are difficult to maintain conversation-level consistency under within-speaks' emotion shifts.
Approach:	They propose a framework that combines appraisal-guided structured generation with graph-structured reinforcement learning to extract triplets from multi-turn multimodal conversations.
Outcome:	The proposed framework outperforms baselines on public MECTEC benchmarks and improves structure-aware metrics on emotion shift coherence and core events.

Copied to clipboard

Yixuan Tang, Yi Yang

Fingerprinting LLMs via Prompt Injection (2026.acl-long)

Challenge:	Recent work shows that decoder-only LLMs can serve as strong embedding backbones when fine-tuned with contrastive objectives.
Approach:	They propose a framework that activates the latent representation power of frozen LLMs by rerouting the final token's KV states as a prepended prefix.
Outcome:	The proposed framework outperforms existing training-free baselines by 10% on MTEB and maintains robust performance on sequences up to 4,096 tokens.

Copied to clipboard

Yuepeng Hu, Zhengyuan Jiang, Mengyuan Li, Osama Ahmed, Zhicong Huang, Cheng Hong, Neil Zhenqiang Gong

Exploring Layer Activation Dynamic of CoT via Knowledge Probe (2026.acl-long)

Challenge:	Existing provenance detection methods for large language models are infeasible for already published models and compare outputs using hand-crafted or random prompts.
Approach:	They propose a detection framework that constructs fingerprints by exploiting LLMs’ inherent vulnerability to prompt injection.
Outcome:	The proposed framework achieves high true positive rates while keeping false positive rates near zero.

Copied to clipboard

Chuanxin Zhang, Jiajun Liu, Yao He, Wenjun Ke, Peng Wang, Yankun Le, Sirui Liu, Zhaoyu Yang

From Trust to Compromise: Outcome-Verified LLM Phishing Simulation and Real-Time Defense (2026.acl-long)

Challenge:	Chain-of-thought reasoning has emerged as a crucial paradigm for multi-step reasoning tasks.
Approach:	They propose a multi-stage probing framework that enforces structured reasoning with three explicit stages: keyword extraction, theorem generation, and computation execution.
Outcome:	The proposed framework enforces structured reasoning with three explicit stages: keyword extraction, theorem generation, and computation execution.

Copied to clipboard

Tulika Tewari, Nalin Asanka Gamagedara Arachchilage, Jagat Sesh Challa, Dhruv Kumar

EvoNarrator: Modeling Scientific Evolution for Feasible Hypothesis Generation (2026.acl-long)

Challenge:	Large Language Models (LLMs) excel as conversational agents, but existing simulators focus on PII requests within the chat.
Approach:	They propose a large language model that generates human-like language and maintains conversational context to automate social engineering attacks.
Outcome:	The proposed model improves dialogue-level detection over a real-time baseline.

Copied to clipboard

Xiaoying Le, Pengfei Qian, Yuanzhao Zhai, Xu Zhang, Qian Liu, Feng Dawei, Bo Ding

Challenge:	Scientific discovery evolution does not occur ex nihilo but is characterized by structural deepening and reconfiguration of existing functionalities.
Approach:	They propose a framework for hypothesis generation based on evolutionary narratives . they extract structured P-M-L-F quadruples from citation networks and introduce a mechanism to assess their semantic compatibility.
Outcome:	The proposed framework reduces logical disconnects by evaluating its semantic compatibility.

DecoCal: Decoding with Calibration in Diffusion Large Language Models (2026.acl-long)

Copied to clipboard

Fan Xu, Huixuan Zhang, Xiaojun Wan

LiveFact: A Dynamic, Time-Aware Benchmark for LLM-Driven Fake News Detection (2026.acl-long)

Challenge:	Diffusion Large Language Models (DLLMs) generate text via iterative token denoising . but decoding is challenging, with many tokens appearing predictable early .
Approach:	They propose a Decoding framework that performs Calibration of token-level confidence across diffusion steps and leverages the calibrated results to guide decoding decisions.
Outcome:	Experiments on multiple DLLMs and benchmarks show that DecoCal improves generation accuracy compared to existing strategies.

Copied to clipboard

Cheng Xu, Changhong Jin, Yingjie Niu, Nan Yan, Yuke Mei, Shuhao Guan, Liming Chen, Tahar Kechadi

Beyond "I Don’t Know": Evaluating LLM Self-Awareness in Discriminating Data and Model Uncertainty (2026.acl-long)

Challenge:	Current evaluation frameworks are static and vulnerable to benchmark data contamination . current models are ineffective at assessing reasoning under temporal uncertainty .
Approach:	They propose a live-based benchmark that simulates the real-world "fog of war" they propose evaluating models on their ability to reason with evolving, incomplete information .
Outcome:	The proposed model outperforms proprietary state-of-the-art models in classification and evidence mode . it also provides a component to monitor BDC explicitly .

Copied to clipboard

Jingyi Ren, Ante Wang, Yunghwei Lai, Xiaolong Wang, Linlu Gong, Weitao Li, Weizhi Ma, Yang Liu

Challenge:	Prior studies treat refusal as a generic "I don't know" lack of distinction limits downstream action decisions like requesting clarification or invoking external tools.
Approach:	They propose a benchmark to evaluate explicit uncertainty attribution in large language models.
Outcome:	The proposed method improves uncertainty attribution while preserving answer accuracy.

Unlearners Can Lie: Evaluating and Improving Honesty in LLM Unlearning (2026.acl-long)

Copied to clipboard

Renjie Gu, Jiazhen Du, Yihua Zhang, Sijia Liu

MatchTIR: Fine-Grained Supervision for Tool-Integrated Reasoning via Bipartite Matching (2026.acl-long)

Challenge:	Existing methods for unlearning in large language models often hallucinate, generate abnormal token sequences, or behave inconsistently, raising safety and trust concerns.
Approach:	They propose a formal definition of unlearning honesty that preserves both utility and honesty on retained knowledge and ensures effective forgetting while encouraging the model to acknowledge its limitations.
Outcome:	The proposed method achieves highest rejection rate and refusal stability on Q A tasks from the forget set, nearly double the second-best method.

Copied to clipboard

Changle Qu, Sunhao Dai, Hengyi Cai, Jun Xu, Shuaiqiang Wang, Dawei Yin

Knowledge is Not Enough: Injecting RL Skills for Continual Adaptation (2026.acl-long)

Challenge:	Existing reinforcement learning methods rely on outcome- or trajectory-level rewards, assigning uniform advantages to all steps within a trajectory.
Approach:	They propose a framework that introduces fine-grained supervision via bipartite matching-based turn-level reward assignment and dual-level advantage estimation.
Outcome:	The proposed framework surpasses the majority of 8B competitors on three benchmarks.

Copied to clipboard

Pingzhi Tang, Yiding Wang, Muhan Zhang

BhashaSutra: A Task-Centric Unified Survey of Indian NLP Datasets, Corpora, and Resources (2026.acl-long)

Challenge:	Large Language Models face the "knowledge cutoff" problem because their parametric memory remains frozen after pretraining, preventing them from natively internalizing new information or tools on the fly.
Approach:	They propose a framework that supports modular skill transfer for efficient and effective knowledge adaptation by extracting a domain-agnostic **Skill Vector from a source domain.
Outcome:	Experiments on knowledge-incorporation QA (SQuAD, LooGLE) and agentic tool-use benchmarks show that the proposed framework outperforms state-of-the-art self-editing SFT by 9.9 points.

Copied to clipboard

Raghvendra Kumar, Devankar Raj, Sriparna Saha

Challenge:	Existing reviews focus on a few high-resource languages or embed Indian languages within broad multilingual settings, limiting coverage of low-resourced and culturally diverse varieties.
Approach:	They present a unified survey of Indian NLP resources, covering 200+ datasets, 50+ benchmarks, and 100+ models, tools, and systems across text, speech, multimodal, and culturally grounded tasks.
Outcome:	The proposed survey covers 200+ datasets, 50+ benchmarks, and 100+ models, tools, and systems across text, speech, multimodal, and culturally grounded tasks.

A Dual-Task Paradigm to Investigate Sentence Comprehension Strategies in Language Models (2026.acl-long)

Copied to clipboard

Rei Emura, Saku Sugawara

Diffusion-CAM: Faithful Visual Explanations for dMLLMs (2026.acl-long)

Challenge:	Existing methods do not directly target the balance between memory and sentence processing, which is central to human working memory.
Approach:	They propose a dual-task paradigm that combines arithmetic computation with sentence comprehension . they show a greater accuracy gap between plausible sentences and implausible sentences .
Outcome:	The proposed paradigm shows that plausibility-based comprehension mirrors humans’ rational inference.

Copied to clipboard

Haomin Zuo, Yidi Li, Luoxiao Yang, Xiaofeng Zhang

Beyond Noise: Characterizing Creative Potential in Unverifiable LLM Hallucinations (2026.acl-long)

Challenge:	Existing Class Activation Mapping methods are ill-suited for interpreting non-autoregressive behaviors of diffusion-based architectures.
Approach:	They propose to use a method to generate parallel activation maps by probing intermediate representations in the transformer backbone to capture latent features and their class-specific gradients.
Outcome:	Experiments show that Diffusion-CAM significantly outperforms SoTA methods in localization accuracy and visual fidelity.

Copied to clipboard

Yu Yan, Chunhong Zhang, Haiyu Zhao, Ziyang Zeng, Zihao Liu, Yongkang Wu, Jianzhou Diao, YiJie Chen, Shujie Wang, Zheng Hu

LLM-SLM Collaborative Framework of Idiomatic Expression Generation (2026.acl-long)

Challenge:	Large Language Models generate outputs that extend beyond established knowledge . prior work does not characterize the unverifiable space as a whole .
Approach:	They propose a novelty-verifiability characterization that distinguishes Creative Synthesis from Groundless Fabrication by a conceptual creation task.
Outcome:	The proposed model distinguishes Creative Synthesis (Region A) from Groundless Fabrication (Regium B) it shows that Region A is non-negligible and robust, persisting across generation strategies, models, domains, and embedding choices.

Copied to clipboard

Hui Gao, Changhao Song, Peng Zhang, Jing Zhang, Chang Yang, Liuxian Ge

Focus-dLLM: Accelerating Long-Context Diffusion LLM Inference via Confidence-Guided Context Focusing (2026.acl-long)

Challenge:	Existing methods for idiomatic expression generation lack parallel data and manual annotations.
Approach:	They propose an iterative LLM-SLM collaborative framework that replaces human supervision for idiomatic expression data generation.
Outcome:	The proposed framework outperforms DeepSeek-R1 in Chinese Idiom Polishing with a 25.2% improvement in accuracy.

Copied to clipboard

Lingkun Long, Yushi Huang, Shihao Bai, Ruihao Gong, Jun Zhang, Ao Zhou, Jianlei Yang

Table Question Answering in the Era of Large Language Models: A Comprehensive Survey of Tasks, Methods, and Evaluation (2026.acl-long)

Challenge:	Existing methods for estimating attention importance for tokens are ineffective . dLLMs require bidirectional attention, which limits inference efficiency .
Approach:	They propose a training-free attention sparsification framework for efficient long-context inference . they propose 'sink-aware pruning strategy' to accurately estimate and remove redundant computation .
Outcome:	The proposed approach offers 29 lossless speedup under 32K context length.

Copied to clipboard

Wei Zhou, Bolei Ma, Annemarie Friedrich, Mohsen Mesgar

RouteMoA: Dynamic Routing without Pre-Inference Boosts Efficient Mixture-of-Agents (2026.acl-long)

Challenge:	Table Question Answering (TQA) aims to answer natural language questions using tabular data.
Approach:	They propose a systematic overview of TQA research using large language models and summarize available benchmarks based on task features.
Outcome:	The proposed framework provides a comprehensive overview of the current state of the art in the field of Table Question Answering.

Copied to clipboard

Jize Wang, Han Wu, Zhiyuan You, Yiming Song, Yijun Wang, Zifei Shan, Yining Li, Songyang Zhang, Xinyi Le, Cailian Chen, Xinping Guan, Dacheng Tao

Patches of Nonlinearity: Instruction Vectors in Large Language Models (2026.acl-long)

Challenge:	Existing methods for mixing-of-agents (MoA) lack model selection criteria and struggle with large model pools.
Approach:	They propose a mixture-of-agents framework with dynamic routing that uses a lightweight scorer to perform initial screening and refines the model scores through self- and cross-assessment.
Outcome:	The proposed framework outperforms existing methods for large model pools and tasks . it reduces cost by 89.8% and latency by 63.6% in the large-scale model pool.

Copied to clipboard

Irina Bigoulaeva, Jonas Rohweder, Subhabrata Dutta, Iryna Gurevych

Reasoning Gets Harder for LLMs Inside A Dialogue (2026.acl-long)

Challenge:	Despite the success of instruction-tuned language models, little is known about how they process instructions internally.
Approach:	They propose a method to localize instruction processing in language models that is free from patching assumptions.
Outcome:	The proposed method disentangles the implicit linear assumptions of patching-based techniques.

Copied to clipboard

Ivan Kartáč, Mateusz Lango, Ondrej Dusek

Capability Decomposition for Unified Information Extraction via Hierarchical Mixture-of-Experts (2026.acl-long)

Challenge:	Large Language Models (LLMs) achieve strong performance on many reasoning benchmarks, yet these evaluations typically focus on isolated tasks that differ from real-world usage in task-oriented dialogue (TOD).
Approach:	They propose to use a dynamic benchmark to examine how framing reasoning tasks within task-oriented dialogue (TOD) affect LLM performance.
Outcome:	The proposed model performs well on isolated tasks and in task-oriented dialogues, but performance is inconsistent between them.

Copied to clipboard

Jing Zhou, Peng Wang, Wenjun Ke, Jiajun Liu, Yao He

PDR: A Plug-and-Play Positional Decay Framework for LLM Pre-training Data Detection (2026.acl-long)

Challenge:	Existing methods for IE tasks suffer from inconsistent schema representation and implicitly intermediate reasoning . UC-UIE adopts a low-rank adapted hierarchical Mixture-of-Experts adapter for UIE tasks .
Approach:	They propose a framework that decomposes IE reasoning into three universal capabilities . UC-UIE adopts a low-rank Adaptation adapter to fine-tune LLMs for IE tasks .
Outcome:	The proposed framework outperforms full-parameter tuning methods with 1.24% trainable parameters and outperformed existing methods in generalization and interpretability.

Copied to clipboard

Jinhan Liu, Yibo Yang, Ruiying Lu, Piotr Piękos, Yimeng Chen, Peng Wang, Dandan Guo

PRISM: Probabilistic Reward Model with Inherent Structural Modeling (2026.acl-long)

Challenge:	Existing likelihood-based methods for detecting pretraining data are limited in black-box, zero-shot settings.
Approach:	They propose a training-free and plug-and-play framework that reweights token-level scores to amplify distinct signals from early positions while suppressing noise from later ones.
Outcome:	The proposed framework amplifys signals from early positions while suppressing noise from later positions.

Copied to clipboard

Yuhang Zhou, Yixin Cao, Yuchen Ni, Shihan Dou, Xutian Chen, Ge Zhang, Xiang Liu, Guangnan Ye

Cognitive Modeling

Challenge:	Existing evaluators compress diverse human judgments into a single scalar, leading to brittle alignment and reward hacking.
Approach:	They propose a Gaussian-based reinterpretation of reward evaluation as a conditional distribution and a mixture of Gaussians to capture conflicting preference dimensions.
Outcome:	The proposed model outperforms scalar baselines in accuracy and generalization.

Multilingual Idioms in Sentences and Conversations Across High-, Medium-, and Low-Resource Languages (2026.acl-long)

Copied to clipboard

Saeed Almheiri, Bilal Elbouardi, Salsabila Zahirah Pranida, Irina Nikishina, Ashwath Rao B, Parameswari Krishnamurthy, Muhammad Cendekia Airlangga, Rifo Ahmad Genadi, Nguyen Phan Gia Bao, Amir Hossein Yari, Hawau Olamide Toyin, Nurdaulet Mukhituly, Mena Attia, Besher Hassan, Ahmad Fathan Hidayatullah, Tatsuki Kuribayashi, Haonan Li, Suma Bhat, Fajri Koto

Author-in-the-Loop Response Generation and Evaluation: Integrating Author Expertise and Intent in Responses to Peer Review (2026.acl-long)

Challenge:	idioms are a major challenge for multilingual NLP because their meanings shift between figurative and literal usage, often requiring context for accurate interpretation.
Approach:	They propose a multilingual idiom dataset that provides idiomatic expressions in both sentence-level and conversational contexts.
Outcome:	The proposed model performs well with low-resource idioms, but lacks contextual inference.

Copied to clipboard

Qian Ruan, Iryna Gurevych

WeightLoRA: Keep Only Necessary Adapters (2026.acl-long)

Challenge:	Existing ARG work lacks author inputs and controls and no evaluation measures response reflection of author signals and effectiveness in addressing reviewer concerns.
Approach:	They propose a novel author-in-the-loop framework that integrates domain expertise and author-only information into author response generation (ARG) they also propose re3Align, a large-scale dataset of aligned review–response–revision triplets, where revisions proxy author signals and REspGen, an author- in-the loop ARG framework supporting flexible author input, multi-attribute control, and evaluation-guided refinement.
Outcome:	Experiments with SOTA LLMs show that author input and evaluation-guided refinement improves author response quality and controllability–quality trade-offs.

Copied to clipboard

Andrey Veprikov, Vladimir Solodkin, Zyl Alexander, Andrey Savchenko, Aleksandr Beznosikov

Reward Modeling for Scientific Writing Evaluation (2026.acl-long)

Challenge:	Low-rank adaptation (LoRA) adds trainable adapters to selected layers, but requires significant memory to train large models and intuition on which layers to add adapters.
Approach:	They propose a method which adds trainable adapters to selected layers . they compare weightLoRA with different adaptive approaches to reduce trainable parameters while maintaining consistent or even superior metric values.
Outcome:	The proposed method reduces the number of trainable parameters while maintaining the capability to obtain consistent or even superior metric values.

Copied to clipboard

Furkan Şahinuç, Subhabrata Dutta, Iryna Gurevych

Deriving Character Logic from Storyline as Codified Decision Trees (2026.acl-long)

Challenge:	Existing models for scientific writing evaluation are primarily optimized for general-purpose benchmarks with fixed scoring rubrics and evaluation criteria.
Approach:	They propose to train scientific writing evaluation models that leverage domain knowledge . they use a two-stage evaluation framework that optimizes evaluation preferences and refines reasoning capabilities .
Outcome:	The proposed model generalizes effectively across tasks and to previously unseen settings.

Copied to clipboard

Letian Peng, Kun Zhou, Longfei Yun, Yupeng Hou, Jingbo Shang

Challenge:	Existing behavioral profiles are unstructured, weakly validated, and unusable . existing models are weakly valid, leading to brittle agent behavior . Using codified decision trees, we show that CDT outperforms previous methods .
Approach:	They propose a data-driven framework that induces an executable decision structure from narrative data.
Outcome:	The proposed framework outperforms human-written profiles and prior profiles on multiple benchmarks.

GASim: A Graph-Accelerated Hybrid Framework for Social Simulation (2026.acl-long)

Copied to clipboard

Xuan Zhou, Yanhui Sun, Hantao Yao, Allen He, Yongdong Zhang, Wu Liu

Mock Worlds, Real Skills: Building Small Agentic Language Models with Synthetic Tasks, Simulated Environments, and Rubric-Based Rewards (2026.acl-long)

Challenge:	Large-scale social simulators require high latency due to expensive memory retrieval and sequential ABM execution.
Approach:	They propose a graph-accelerated hybrid multi-agent framework for large-scale social simulations that uses large language models and numerical agent-based models to scale up simulations.
Outcome:	The proposed framework delivers 9.94 speedup over the traditional framework and consumes less than 20% of tokens.

Copied to clipboard

Yuanjie Lyu, Chengyu Wang, Lei Shen, Jun Huang, Tong Xu

Investigating Cross-Modal Skill Injection: Scenarios, Methods, and Hyperparameters (2026.acl-long)

Challenge:	Existing agentic training data are narrow in task variety and easily solved . real-world APIs lack diversity and are unstable for large-scale reinforcement learning rollout processes.
Approach:	They propose a framework that synthesizes diverse tool-use training data and simulates complete environments.
Outcome:	The proposed framework synthesizes diverse tool-use training data and simulates complete environments.

Copied to clipboard

Zhiyu Xu, Lean Wang, Yuanxin Liu, Lei Li, Hao Zhou, Fandong Meng, Jie Zhou, Xu Sun

The Bidirectional Process Reward Model (2026.acl-long)

Challenge:	Existing research lacks systematic analysis of the applicability and methodology of cross-modal skill injection.
Approach:	They investigate the applicability and methodology of cross-modal skill injection by integrating a domain-expert LLM into a VLM.
Outcome:	The proposed method enables transfer of domain-specific expertise from Large Language Models (LLMs) to VLMs without incurring additional training data requirements or significant computational overhead.

Copied to clipboard

Lingyin Zhang, Jun Gao, Xiaoxue Ren, Ziqiang Cao

AgentMark: Utility-Preserving Behavioral Watermarking for Agents (2026.acl-long)

Challenge:	Process reward models (PRMs) assign fine-grained scores to intermediate reasoning steps within a solution trajectory.
Approach:	They propose a bidirectional evaluation paradigm that integrates a parallel evaluation stream alongside the L2R evaluation scheme and a gating mechanism to fuse the reward scores.
Outcome:	The proposed model surpasses unidirectional baselines in multiple benchmarks, LLM objectives and sampling policies.

Copied to clipboard

Kaibo Huang, Jin Tan, Yukun Wei, Wanling Li, Zipei Zhang, Hui Tian, Zhongliang Yang, Linna Zhou

Challenge:	Recent advances in large language models (LLMs) have improved text generation and reasoning.
Approach:	They propose a behavioral watermarking framework that embeds multi-bit identifiers into planning decisions while preserving utility.
Outcome:	The proposed framework embeds multi-bit provenance into planning decisions while preserving utility.

Generative Gamer: Learning Equilibrium Strategy by LLM-driven Dynamic Deduction (2026.acl-long)

Copied to clipboard

Yadong Zhang, Xinshu Shen, Yupei Ren, Shangqing Zhao, Man Lan

Probing for Reading Times (2026.acl-long)

Challenge:	Large Language Models (LLMs) falter in domains requiring deep strategic reasoning.
Approach:	They propose a framework that trains LLMs to reason like an expert player . they propose action pruning based on policy confidence, state pruning via value estimation and branch pruning inspired by alpha-beta principles to train the model effectively.
Outcome:	Experiments on Tic-Tac-Toe and Leduc Poker show that GenGamer significantly improves the strategic capabilities of large language models.

Copied to clipboard

Eleftheria Tsipidi, Samuel Kiegeland, Francesco Ignazio Re, Tianyang Xu, Mario Giulianelli, Karolina Stanczak, Ryan Cotterell

No More Stale Feedback: Co-Evolving Critics for Open-World Agent Learning (2026.acl-long)

Challenge:	a large body of work on probing has demonstrated that language model representations encode a wealth of linguistic information, but it remains unclear whether they also capture cognitive signals about human processing.
Approach:	They use regularized linear regression to compare language model representations against scalar predictors.
Outcome:	The representations from early layers outperform surprisal in predicting early-pass measures such as first fixation and gaze duration.

Copied to clipboard

Zhicong Li, Lingjie Jiang, Yulan Hu, Xingchen Zeng, Yixia Li, Xiangwen Zhang, Guanhua Chen, Zheng Pan, Xin Li, Yong Liu

Challenge:	Current methods for training Large Language Model agents rely on static or offline critic models, which fail to adapt as the policy evolves.
Approach:	They propose a framework that integrates a critique and a policy to optimize the policy and critic through a synchronized co-evolutionary loop.
Outcome:	The proposed framework yields more stable training and higher long-horizon task success across open-world environments.

Continuous Interpretive Steering for Scalar Diversity (2026.acl-long)

Copied to clipboard

Ye-eun Cho

A Scalable Multi-LLM Collaboration System with Retrieval-based Selection and Exploration-Exploitation-Driven Enhancement (2026.acl-long)

Challenge:	Existing studies on pragmatic inference in large language models rely on prompt-based manipulations to elicit a pragmatic interpretation.
Approach:	They propose a method that probes graded pragmatic interpretation by treating activation-level steering strength as a continuous experimental variable.
Outcome:	The proposed method increases pragmatic interpretations globally but collapses item-level variation whereas graded activation steering yields differentiated interpretive shifts aligned with scalar diversity grades.

Copied to clipboard

Shengji Tang, Jianjian Cao, Weihao Lin, Jiale Hong, Bo Zhang, Shuyue Hu, Lei Bai, Tao Chen, Wanli Ouyang, Peng Ye

From Form to Logic: Masked Reconstruction and Reasoning Distillation for Short Video Fake News Detection (2026.acl-long)

Challenge:	Existing multi-LLM collaboration systems often encounter scalability challenges when integrating new LLMs and tasks.
Approach:	They propose a Scalable Multi-LLM Collaboration System to coordinate multiple open-source LLMs.
Outcome:	The proposed system outperforms prevailing closed-source LLMs on eight mainstream benchmarks on multiple tasks.

Copied to clipboard

Qingyan Wang, Lianwei Wu, Botao Wang, null Wangkang, Yaxiong Wang

UR2 : Unify RAG and Reasoning through Reinforcement Learning (2026.acl-long)

Challenge:	Existing detectors that detect short video fake news suffer from global-alignment bias and lack generative reasoning are too late.
Approach:	They propose a Perception-Cognition Dual-driven Detector that jointly observes the form and probes the logic for short video fake news detection.
Outcome:	The proposed detector outperforms baseline detectors on real-world datasets while improving interpretability and robustness in data scarcity scenarios.

Copied to clipboard

Weitao Li, Boran Xiang, Xiaolong Wang, Jingyi Ren, Ante Wang, Zhinan Gou, Weizhi Ma, Yang Liu

LLM-as-Scheduler: Agentic Workflow Dynamic Scheduling (2026.acl-long)

Challenge:	Existing attempts to unify large language models are limited to open-domain QA with fixed retrieval settings.
Approach:	They propose a general reinforcement learning framework that dynamically coordinates retrieval and reasoning.
Outcome:	The proposed framework outperforms existing paradigms on open-domain QA, MMLU-Pro, medical, and mathematical reasoning tasks.

Copied to clipboard

Dawei Xiang, Kexin Chu, Wenyan Xu, Wenhui Zhang, Wei Zhang

Mathematical Proof as a Litmus Test: Revealing Failure Modes of Advanced Large Reasoning Models (2026.acl-long)

Challenge:	Experiments show that LAS cuts token usage by 43% and reduces end-to-end latency by more than 36%, while causing at most a 1.4 percentage-point drop in accuracy compared with a strong fixed workflow.
Approach:	They propose a system that dynamically chooses the right workflow for each query.
Outcome:	Experiments show that LAS cuts token usage by 43% and reduces end-to-end latency by more than 36% while causing at most a 1.4 percentage-point drop in accuracy compared with a strong fixed workflow.

Copied to clipboard

Dadi Guo, Jiayu Liu, Zhiyuan Fan, Zhitao He, Haoran Li, Yuxin Li, Yumeng Wang, Yi R. Fung

Memory-R1: Enhancing Large Language Model Agents to Manage and Utilize Memories via Reinforcement Learning (2026.acl-long)

Challenge:	Large reasoning models have demonstrated remarkable mathematical problem-solving abilities, but their true reasoning shortcomings are often hidden.
Approach:	They propose to leverage the rigor and methodological complexity of mathematical proofs as a diagnostic tool to expose hidden failures.
Outcome:	The proposed model evaluation exploits the rigor and complexity of proof problems to uncover 10 fine-grained errors.

Copied to clipboard

Sikuan Yan, Xiufeng Yang, Zuchao Huang, Ercong Nie, Zifeng Ding, Zonggen Li, Xiaowen Ma, Jinhe Bi, Kristian Kersting, Jeff Z. Pan, Hinrich Schuetze, Volker Tresp, Yunpu Ma

CostBench: Evaluating Multi-Turn Cost-Optimal Planning and Adaptation in Dynamic Environments for LLM Tool-Use Agents (2026.acl-long)

Challenge:	Large Language Models (LLMs) are stateless and limited by a finite context window, preventing them from maintaining knowledge across long conversations or evolving tasks.
Approach:	They propose a reinforcement learning framework that empowers LLMs to actively manage external memory through two specialized agents.
Outcome:	The proposed framework outperforms baselines and benchmarks across diverse question types, three benchmarks, and multiple model scales.

Copied to clipboard

Jiayu Liu, Cheng Qian, Zhaochen Su, Qing Zong, Shijue Huang, Bingxiang He, Yi R. Fung

Decoding Scientific Experimental Images: The SPUR Benchmark for Perception, Understanding, and Reasoning (2026.acl-long)

Challenge:	Existing evaluations of Large Language Models (LLMs) focus on task completion, but neglect a crucial capability: the ability to devise and adjust cost-optimal plans in response to changing environments.
Approach:	They propose a scalable, cost-centric benchmark to evaluate agents’ economic reasoning and replanning abilities.
Outcome:	Evaluating leading open-sourced and proprietary models on CostBench reveals a substantial gap in cost-aware planning .

Copied to clipboard

Junpeng Ding, Zichen Tang, Haihong E, Mengyuan Ji, Yang Liu, Haolin Tian, Haiyang Sun, Pengqi Sun, Yang Xu, Yichen Liu, Haocheng Gao, Zijie Xi, Ruomeng Jiang, Peizhi Zhao, Rongjin Li, Yuanze Li, Jiacheng Liu, Zhongjun Yang, Jintong Chen, Siying Lin

SegTune: Structured and Fine-Grained Control for Song Generation (2026.acl-long)

Challenge:	Xu and Peng, 2025) . . SPUR is a comprehensive benchmark for scientific experimental image perception, understanding, and reasoning, comprising 4,264 question-answering (QA) pairs derived from 1,084 expert-curated images.
Approach:	They propose to use 4,264 question-answering (QA) pairs derived from 1,084 expert-curated images to evaluate the visual perception of multimodal large language models (MLLMs) . they also propose to utilize cross-panel relation understanding to evaluate MLLM’s ability to decipher intricate cross-panel relations.
Outcome:	The proposed model is based on 4,264 question-answering pairs derived from 1,084 expert-curated images.

Copied to clipboard

Yuejiao Wang, Zihao Ji, Pengfei Cai, Xu Li, Haorui Zheng, Zewen Song, Zhongliang Liu, Chen Zhang, Pengfei Wan

Making Large Language Models Efficient Dense Retrievers (2026.acl-long)

Challenge:	Recent advances in neural song generation have enabled high-quality synthesis from lyrics and global textual prompts.
Approach:	They propose a framework that allows users to specify local musical descriptions aligned to song segments.
Outcome:	The proposed framework outperforms baselines in musicality and controllability.

Copied to clipboard

Yibin Lei, Shwai He, Ang Li, Andrew Yates

Lightweight LLM Agent Memory with Small Language Models (2026.acl-long)

Challenge:	Recent studies have shown that fine-tuning large language models for dense retrieval yields strong performance, but their substantial parameter counts make them computationally inefficient.
Approach:	They propose a framework for developing efficient retrievers that performs coarse-to-fine compression through a coarse-grained coarse-tuning strategy.
Outcome:	The proposed framework reduces model size and inference cost while preserving performance of full-size models.

Copied to clipboard

Jiaquan Zhang, Chaoning Zhang, Shuxu Chen, Zhenzhen Huang, Pengcheng Zheng, Zhicheng Wang, Ping Guo, Fan Mo, Sung-Ho Bae, Jie Zou, Jiwei Wei, Yang Yang

HeteroSpec: Leveraging Contextual Heterogeneity for Efficient Speculative Decoding (2026.acl-long)

Challenge:	Existing external memory systems for LLMs have low online overhead but are unstable in accumulating latency over long interactions.
Approach:	They propose a lightweight memory system for better agent memory driven by Small Language Models . lightmem modularizes memory retrieval, writing, and long-term consolidation . they show consistent gains across model scales and high efficiency .
Outcome:	The proposed system improves agent memory but has low latency and low online overhead . it separates online processing from offline consolidation to enable efficient memory invocation . the proposed system achieves an average F1 improvement of 2.5 over A-MEM on LoCoMo .

Copied to clipboard

Siran Liu, Yang Ye, Qianchao Zhu, Zane Cao, Yongchao He

A Model of the Language Process (2026.acl-long)

Challenge:	Autoregressive decoding limits the inference throughput of Large Language Models due to its sequential dependency.
Approach:	They propose a framework that allocates verification effort in proportion to candidate uncertainty.
Outcome:	Speculative decoding achieves an average speedup over state-of-the-art methods . a small subset of high-confidence predictions accounts for most successful verifications .

Copied to clipboard

Brandon Duderstadt, Hayden Helm

Rethinking Table Pruning in TableQA: From Sequential Revisions to Gold Trajectory-Supervised Parallel Search (2026.acl-long)

Challenge:	Language is a process that changes over time as new vocabulary emerges, word meanings shift, and narratives progress.
Approach:	They introduce a BERT style transformer encoder that models language by jointly learning to predict document contents and classify document publication dates.
Outcome:	The proposed model can predict document contents and classify document publication dates and accurately detects changes in word meanings.

Copied to clipboard

Yu Guo, Shenghao Ye, Shuangwu Chen, Zijian Wen, Tao Zhang, Bai Qirui, Dong Jin, Yunpeng Hou, Huasen He, null Jianyang, Xiaobin Tan

Interleaved Tool-Call Reasoning for Protein Function Understanding (2026.acl-long)

Challenge:	Existing pruning methods rely on sequential revisions and unreliable critique signals . Existing methods fail to detect the loss of answer-critical data .
Approach:	They propose a table pruning framework which transforms table pruning to gold trajectory-supervised parallel search.
Outcome:	The proposed framework outperforms the strongest baseline pruning framework by 3.2% on various tabular reasoning tasks.

Copied to clipboard

Chuanliu Fan, Zicheng Ma, Huanran Meng, Aijia Zhang, Wenjie Du, Jun Zhang, Ziqiang Cao, Guohong Fu

Among Us: Language of Conspiracy Theorists on Mainstream Reddit (2026.acl-long)

Challenge:	Recent advances in large language models have highlighted the effectiveness of chain-of-thought reasoning in symbolic domains such as mathematics and programming.
Approach:	They propose a tool-augmented protein reasoning agent that unifies problem decomposition, tool invocation, and grounded answer generation.
Outcome:	The proposed protein function understanding agent outperforms text-only reasoning models with an average performance improvement of 103%.

Copied to clipboard

Francesco Corso, Giuseppe Russo, Francesco Pierri, Gianmarco De Francisci Morales

Challenge:	Conspiracy theories are influential, alternative narratives that explain events through the actions of secretive, malevolent groups.
Approach:	They analyze a large-scale longitudinal dataset of over 500 million comments on reddit . they show that users exhibit distinctive linguistic patterns that enable machine learning models to distinguish them from the general population within individual communities.
Outcome:	The proposed model outperforms global classifiers by 17 percentage points.

HybridKV: Hybrid KV Cache Compression for Efficient Multimodal Large Language Model Inference (2026.acl-long)

Copied to clipboard

Bowen Zeng, Feiyang Ren, Jun Zhang, Xiaoling Gu, Ke Chen, Lidan Shou, Huan Li

Breaking Down and Building Up: Mixture of Skill-Based Vision-and-Language Navigation Agents (2026.acl-long)

Challenge:	Multimodal Large Language Models (MLLMs) are hindered by the rapid growth of key–value (KV) caches.
Approach:	They propose a hybrid KV cache compression framework that reduces KV memory by 7.9 and speeds up decoding by 1.52.
Outcome:	Experiments on 11 multimodal benchmarks show that HYBRIDKV cuts KV cache memory by 7.9 and speeds up decoding by 1.52.

Copied to clipboard

Tianyi Ma, Yue Zhang, Zehao Wang, Parisa Kordjamshidi

Bridging Distance and Spectral Positional Encodings via Anchor-Based Diffusion Geometry Approximation (2026.acl-long)

Challenge:	Vision-and-Language Navigation (VLN) is a subfield of embodied AI that integrates natural language understanding, visual perception, and sequential decision-making to allow autonomous agents to navigate and interact within visual environments.
Approach:	They propose a modular framework that introduces structured, skill-based reasoning into Transformer-based VLN agents.
Outcome:	The proposed framework decomposes navigation into atomic skills handled by a specialized agent.

Copied to clipboard

Yan Zimo, Zheng Xie, Runfan Duan, Chang Liu, Wumei Du

A Goal Without a Plan Is Just a Wish: Efficient and Effective Global Planner Training for Long-Horizon Agent Tasks (2026.acl-long)

Challenge:	Molecular graph learning benefits from positional signals that capture local neighborhoods and global topology.
Approach:	They propose to use anchor-based distance encodings to approximate diffusion geometry.
Outcome:	The proposed model outperforms models without positional encodings on DrugBank with a shared GNP-based DDI backbone.

Copied to clipboard

Shuzheng Si, Haozhe Zhao, Kangyang Luo, Gang Chen, Fanchao Qi, Minjia Zhang, Baobao Chang, Maosong Sun

Stereotype Bias in a Bilingual Setting: A Culturally Grounded Evaluation in Kazakhstan (2026.acl-long)

Challenge:	Recent advances in large language models (LLMs) have leapt from static chatbots to versatile agents that tackle complex tasks such as science experiments.
Approach:	They propose a plan-and-execute framework and propose 'EAGLET' to enhance the executor agent's planning abilities without human effort.
Outcome:	The proposed method outperforms existing methods on three long-horizon tasks and reduces training costs by 8 compared to baselines.

Copied to clipboard

Nurkhan Laiyk, Daniil Orel, Ayana Mussabayeva, Maiya Goloburda, Kamila Kuishibekova, Liya Goloburda, Diana Turmakhan, Preslav Nakov, Yuxia Wang, Fajri Koto

Arguments that Alter Minds: LLM Rationales Sway Human (and LLM) Notions of Plausibility (2026.acl-long)

Challenge:	Stereotype bias in language models is largely understudied in English . language models perform strongly on downstream NLP tasks, but they are pre-trained on large text corpora .
Approach:	They use a dataset to assess stereotype bias in language models in Kazakhstan . they find that stereotype bias is most pronounced in code-mixed inputs .
Outcome:	The proposed dataset shows that stereotype bias is most pronounced in code-mixed inputs.

Copied to clipboard

Shramay Palta, Peter A. Rankel, Sarah Wiegreffe, Rachel Rudinger

GLARE: Agentic Reasoning for Legal Judgment Prediction (2026.acl-long)

Challenge:	Experiments with LLMs reveal similar patterns of influence on human plausibility judgments of commonsense benchmark answers.
Approach:	They find that human plausibility judgments of commonsense benchmark answers are affected by implausibility arguments for or against an answer.
Outcome:	The results show that human judges find LLM rationales convincing and that human annotators agree on the most plausible answer when the plausibility gap is wide.

Copied to clipboard

Xinyu Yang, Chenlong Deng, Zhicheng Dou

Vista-LLM: Decoupled Query-Guided Visual Token Pruning for Efficient Long-Video Large Language Models (2026.acl-long)

Challenge:	Large language models struggle with fine-grained distinctions between similar charges.
Approach:	They propose an agentic legal reasoning framework that actively retrieves external knowledge during decision-making.
Outcome:	The proposed model outperforms baseline models on complex cases involving confusing or rare charges on real-world datasets.

Copied to clipboard

Zhenyu Li, Zuchao Li, Ping Wang, Lefei Zhang, Haojun Ai

RoboFailRing: Retrieval-Augmented and Language Grounding Failure Detection for VLM-enabled Robotic Manipulation (2026.acl-long)

Challenge:	Long-video understanding is bottlenecked by the high cost of processing massive visual tokens.
Approach:	They propose a decoupled framework for query-guided visual token pruning . their method reduces visual tokens by 90% and accelerates inference by 98% .
Outcome:	The proposed framework reduces visual tokens by 90% and accelerates inference while retaining over 98% of baseline performance on average.

Copied to clipboard

Chenduo Ying, Linkang Du, Yuanchao Shu, Peng Cheng

Scaling Law for Multimodal Large Language Model Supervised Fine-Tuning (2026.acl-long)

Challenge:	RoboFailRing enables timely failure detection during task execution and enhances reasoning accuracy of VLMs.
Approach:	They propose a robot-based failure detection system that enables timely failure detection . they evaluate a large-scale simulated dataset and provide a grounded failure report .
Outcome:	The proposed method achieves rapid failure detection and returns similarity-based decision on large-scale simulated failures.

Copied to clipboard

YiFan Zhang, Tao Yu, Feng Li, Chaoyou Fu, Yibo Hu, Kun Wang, Qingsong Wen, Zhang Zhang, Liang Wang, Rong Jin

CrochetBench: Can Vision-Language Models Move from Describing to Doing in Crochet Domain? (2026.acl-long)

Challenge:	supervised fine-tuning (SFT) is crucial for multimodal large language models, yet a comprehensive scaling law is lacking . et al.: scaling laws focus on model size, pre-training tokens, and MLLM SFT data volumes .
Approach:	They propose two scaling laws to guide optimal model-data configuration . they propose one applicable when training data volumes are well defined by researchers .
Outcome:	The proposed scaling laws provide valuable recommendations for optimal resource allocation . they show that the proposed laws are more accurate than existing models .

Copied to clipboard

Peiyu Li, Xiaobao Huang, Ting Hua, Nitesh V Chawla

DORA: A Dual-Objective Reinforcement Learning Framework for Effective and Efficient Multimodal Agentic Search (2026.acl-long)

Challenge:	Existing multimodal large language models cannot generate executable procedures . authors propose a new benchmark to assess procedural competence in multimodal models .
Approach:	They propose a new benchmark to assess procedural competence in multimodal large language models . they use a CrochetPARADE DSL representation to enable structural validation and functional evaluation .
Outcome:	The proposed model enables structural validation and functional evaluation via execution.

Copied to clipboard

Guangming Qin, Yuhao Deng, Yukun Zhao, Zhenyang Li, Junfeng Wang, Dawei Yin, Ye Yuan, Guoren Wang, Yizhou Yan, Chengliang Chai, Lei Cao

IUQ: Interrogative Uncertainty Quantification for Long-Form Large Language Model Generation (2026.acl-long)

Challenge:	Existing methods to train large language models overlook quality of intermediate search results . existing methods often invoke search calls during reasoning, making inference inefficient .
Approach:	They propose a dual-objective reinforcement learning framework to improve search strategies of MLLMs . DORA outperforms state-of-the-art methods, achieving up to 8.4% higher accuracy .
Outcome:	The proposed model outperforms state-of-the-art methods while reducing search calls by 9.7%.

Copied to clipboard

Haozhi Fan, Jinhao Duan, Kaidi Xu

CODESTRUCT: Code Agents over Structured Action Spaces (2026.acl-long)

Challenge:	Recent approaches to quantify uncertainty in LLMs produce short or constrained answer sets, but many real-world applications require long-form and free-form text generation.
Approach:	They propose a framework that leverages inter-sample consistency and intra-sampled faithfulness to quantify the uncertainty in long-form LLM outputs.
Outcome:	The proposed framework provides reliable measures of claim-level uncertainty and the model’s faithfulness over two widely used long-form generation datasets.

Copied to clipboard

Myeongsoo Kim, Chao-Chun Hsu, Dingmin Wang, Shweta Garg, Varun Kumar, Murali Krishna Ramanathan

Answering the Wrong Question: Reasoning Trace Inversion for Abstention in LLMs (2026.acl-long)

Challenge:	LLM-based code agents treat repositories as unstructured text, fail to produce valid patches . authors propose a structure-aware interface that exposes a codebase as a programmable action space .
Approach:	They propose to reframe the codebase as a structured action space where agents operate on named AST entities rather than text spans.
Outcome:	Evaluated on six LLMs, the proposed framework improves Pass@1 accuracy by 1.2-5.0% and reduces token consumption by 12-38%.

Copied to clipboard

Abinitha Gourabathina, Inkit Padhi, Manish Nagireddy, Subhajit Chaudhury, Prasanna Sattigeri

Beyond Single-shot Writing: Deep Research Agents are Unreliable at Multi-turn Report Revision (2026.acl-long)

Challenge:	Large Language Models have been shown to have worse abstention abilities than reasoning models . a new class of abstraction methods is developed to improve absttention performance .
Approach:	They propose a class of abstention methods that generate reasoning trace and reconstruct most likely query from it.
Outcome:	The proposed method beats baselines in 33 out of 36 settings.

Copied to clipboard

Bingsen Chen, Boyan Li, Ping Nie, Yuyu Zhang, Xi Ye, Chen Zhao

Challenge:	Existing benchmarks for Deep Research Agents (DRAs) treat report generation as a single-shot writing task.
Approach:	They propose an evaluation suite that establishes multi-turn report revision as a new axis.
Outcome:	The evaluation suite establishes multi-turn report revision as a new axis.

FinCall-Surprise: A Large Scale Multi-modal Benchmark for Earning Surprise Prediction (2026.acl-long)

Copied to clipboard

Dong Shu, Yanguang Liu, Huopu Zhang, Mengnan Du

ALDEN: Reinforcement Learning for Active Navigation and Evidence Gathering in Long Documents (2026.acl-long)

Challenge:	Existing models for earnings surprise prediction rely on expensive, proprietary data.
Approach:	They propose to use textual transcripts and audio recordings to build a dataset for earnings surprise prediction.
Outcome:	The proposed dataset includes 2,688 unique conference calls from 2019 to 2021.

Copied to clipboard

Tianyu Yang, Terry Ruas, Yijun Tian, Jan Philip Wahle, Daniel Kurzawe, Bela Gipp

Reinforced Informativeness Optimization for Long-Form Retrieval-Augmented Generation (2026.acl-long)

Challenge:	Visually rich documents (VRDs) combine text, tables, and figures within complex, semantically structured layouts.
Approach:	They propose a multi-turn reinforcement learning framework that fine-tunes VLMs as interactive agents capable of actively navigating long, visually rich documents.
Outcome:	The proposed framework achieves state-of-the-art on five long-document benchmarks.

Copied to clipboard

Yuhao Wang, Ruiyang Ren, Yucheng Wang, Xin Zhao, Jing Liu, Hua Wu, Haifeng Wang

Android Coach: Improve Online Agentic Training Efficiency with Single State Multiple Actions (2026.acl-long)

Challenge:	Existing reinforcement learning systems lack verifiable reward mechanisms for long-form question answering . current systems lack reliable long-term answers due to lack of factual content .
Approach:	They propose a framework for reinforced verifiable informativeness optimization . it defines informativeness as measurable and externally verifier objective for RL .
Outcome:	Experiments show that RioRAG achieves higher factual recall and faithfulness . the proposed framework is based on a framework that uses nugget-centric verification with cross-source checks .

Copied to clipboard

Guo Gan, Yuxuan Ding, Cong Chen, Yuwei Ren, Yin Huang, Hong Zhou

Inside Out: Evolving User-Centric Core Memory Trees for Long-Term Personalized Dialogue Systems (2026.acl-long)

Challenge:	Existing reinforcement learning methods are expensive due to high latency and sample inefficiency . Currently, RL is limited to one-to-one state-action pairs .
Approach:	They propose a framework that shifts the training paradigm to Single State Multiple Actions and introduce a group-wise advantage estimator based on the averaged critic outputs.
Outcome:	The proposed framework achieves 7.5% and 8.3% success rate improvements on AndroidLab and AndroidWorld over UI-TARS-1.5-7B and attains 1.4x higher training efficiency than existing methods.

Copied to clipboard

Jihao Zhao, Ding Chen, Zhaoxin Fan, Kerun Xu, Mengting Hu, Bo Tang, Feiyu Xiong, Zhiyu li

FinChart-Bench: Benchmarking Financial Chart Comprehension in Vision-Language Models (2026.acl-long)

Challenge:	Existing personalized dialogue systems struggle to reconcile unbounded interactions with finite context constraints.
Approach:	They propose a framework that utilizes a globally maintained PersonaTree as the carrier of long-term user profiling.
Outcome:	The proposed framework outperforms existing systems in suppressing contextual noise and persona inconsistency.

Copied to clipboard

Dong Shu, Haoyang Yuan, Yuchen Wang, Yanguang Liu, Huopu Zhang, Mengnan Du

From Charts to Code: A Hierarchical Benchmark for Multimodal Models (2026.acl-long)

Challenge:	FinChart-Bench is the first benchmark specifically focused on real-world financial charts.
Approach:	They propose a benchmark specifically focused on real-world financial charts.
Outcome:	The proposed benchmark evaluates 26 state-of-the-art LVLMs on FinChart-Bench.

Copied to clipboard

Jiahao Tang, Henry Hengyuan Zhao, Lijian Wu, Zijian Zhang, Yifei Tao, Dongxing Mao, Yang Wan, Jingru Tan, Min Zeng, Min Li, Alex Jinpeng Wang

Learning While Staying Curious: Entropy-Preserving Supervised Fine-Tuning via Adaptive Self-Distillation for Large Reasoning Models (2026.acl-long)

Challenge:	Chart2Code is a new benchmark for evaluating the natural language to chart code generation capabilities of large multimodal models.
Approach:	They introduce Chart2Code, a new benchmark for evaluating the natural language to chart code generation capabilities of large multimodal models.
Outcome:	The proposed benchmark is the first to scale task complexity while capturing diverse scenarios.

Copied to clipboard

Hao Wang, Hao Gu, Hongming Piao, Kaixiong Gong, Yuxiao Ye, Xiangyu Yue, Sirui Han, Yike Guo, Dapeng Wu

Toward Consistent World Models with Multi-Token Prediction and Latent Semantic Enhancement (2026.acl-long)

Challenge:	Recent advances establish "SFT-then-RL" as the defacto paradigm for enhancing large reasoning mod- els on automatically verifiable tasks.
Approach:	They propose an entropy-preserving SFT method to enhance exploration capabilities through intrinsic curiosity.
Outcome:	The proposed method outperforms the vanilla method on reasoning tasks by 2.5 points . it also outperformed the vanilla SFT by 2.9 points on out-of-distribution tasks .

Copied to clipboard

Qimin Zhong, Hao Liao, Haiming Qin, Mingyang Zhou, Rui Mao, Wei Chen, Naipeng Chao

SpiderFlow: Efficient Topology-Aware Scheduling for LLM Training Across Decentralized GPU Clusters (2026.acl-long)

Challenge:	Existing methods to learn internal world models rely on one-step supervision . however, standard MTP suffers from structural hallucinations .
Approach:	They propose a method which anchors predictions to ground-truth hidden state trajectories.
Outcome:	The proposed method bridges the gap between discrete tokens and continuous state representations, reducing structural hallucinations, and improving robustness to perturbations.

Copied to clipboard

Zihan Chang, Shuibing He, Bo Zhou, Sheng Xiao, Siling Yang, Rui Wang, Zhe Pan

Unleashing Spatial Reasoning in Multimodal Large Language Models via Textual Representation Guided Reasoning (2026.acl-long)

Challenge:	Existing approaches to training large language models lack topologyaware task scheduling mechanisms and model parallelization strategies.
Approach:	They propose a topology-aware scheduling system specifically designed for decentralized GPU clusters . they propose heuristic methods at the inter-cluster level with ILP-based optimization within clusters.
Outcome:	The proposed system reduces job completion time by 1.2-1.3 and improves throughput by 1.12-1.25 . it also reduces scheduling overhead by 20-90 on average compared to state-of-the-art scheduling systems.

Copied to clipboard

Jiacheng Hua, Yishu Yin, Yuhang Wu, Tai Wang, Yifei Huang, Miao Liu

UniCorn: Towards Self-Improving Unified Multimodal Models through Self-Generated Supervision (2026.acl-long)

Challenge:	Existing Multimodal Large Language Models struggle with 3D spatial reasoning as they fail to construct structured abstractions of the 3D environment depicted in video inputs.
Approach:	They propose a prompting method that induces MLLMs to generate 3D representations as reasoning traces for more accurate spatial question answering.
Outcome:	Extensive experiments on VSI-Bench and OST-Bech show that TRACE improves over prior prompting strategies across a diverse range of MLLM backbones.

Copied to clipboard

Zhen Fang, Ruiyan Han, XinYu Sun, Yuchen Ma, Ziheng Wang, Yu Zeng, Zehui Chen, Lin Chen, Wenxuan Huang, Wei-Jie Xu, Yi Cao, Feng Zhao

Why Multimodal In-Context Learning Lags Behind? Unveiling the Inner Mechanisms and Bottlenecks (2026.acl-long)

Challenge:	Unified Multimodal Models have achieved remarkable success in cross-modal comprehension, but a gap persists in their ability to translate internal knowledge into faithful and controllable synthesis.
Approach:	They propose a self-improvement framework that partitions a single UMM into three collaborative roles: Proposer, Solver, and Judge.
Outcome:	The proposed framework improves on TIIF, DPG, CompBench and UniCycle benchmarks.

Copied to clipboard

Yu Wang, Sharon Li

Current Agents Fail to Leverage World Model as Tool for Foresight (2026.acl-long)

Challenge:	In-context learning (ICL) enables models to adapt to new tasks via inference-time demonstrations.
Approach:	They propose a simple inference-stage enhancement method that reinforces task mapping transfer.
Outcome:	The proposed method strengthens task mapping transfer in multimodal models . it performs comparable to text-only ICL in zero-shot settings but degrades significantly under few-shot demonstrations.

Copied to clipboard

Cheng Qian, Emre Can Acikgoz, Bingxuan Li, Xiusi Chen, Yuji Zhang, Bingxiang He, Qinyu Luo, Gokhan Tur, Dilek Hakkani-Tür, Yunzhu Li, Heng Ji

Language Model as Planner and Formalizer under Constraints (2026.acl-long)

Challenge:	Generative world models could be used to enhance agents' cognition . agents are expected to operate in settings where tasks unfold over long horizons and involve intricate chains of interdependent decisions.
Approach:	They propose to use vision-language models as external simulators to enhance cognition . they find that agents rarely invoke simulation and misuse predicted rollouts .
Outcome:	The proposed model could be used to predict future states rather than short-horizon reasoning . the model could also be used for real-world planning and robotics .

Copied to clipboard

Cassie Huang, Stuti Mohan, Ziyi Yang, Stefanie Tellex, Li Zhang

HeLa-Mem: Hebbian Learning and Associative Memory for LLM Agents (2026.acl-long)

Challenge:	Large language models (LLMs) have been widely used in planning but lack interpretability and control.
Approach:	They propose to augment widely used planning benchmarks with manually annotated, fine-grained, and rich natural language constraints spanning four formally defined categories.
Outcome:	The proposed model outperforms existing models in 4 state-of-the-art reasoning LLMs, 4 formal languages, and 4 datasets.

Copied to clipboard

Jinchang Zhu, Jindong Li, Cheng Zhang, Jiahong Liu, Menglin Yang

Locket: Robust Feature-Locking Technique for Language Models (2026.acl-long)

Challenge:	Existing memory systems represent conversation history as unstructured embedding vectors, retrieving information through semantic similarity.
Approach:	They propose a bio-inspired memory architecture that models memory as a dynamic graph with Hebbian learning dynamics.
Outcome:	The proposed architecture leverages both semantic similarity and learned associations . it can be used to build a bio-inspired memory graph with Hebbian learning dynamics .

Copied to clipboard

Lipeng He, Vasisht Duddu, N. Asokan

Can Factual Opinions Be Edited (Manipulated) in Large Language Models? (2026.acl-long)

Challenge:	Existing FLoTEs (e.g., password-locked models) fail to meet these criteria.
Approach:	They propose a feature-locking technique that is effective in refusing locked features . they propose scalable FLoTEs that can selectively disable specific features of a model .
Outcome:	The proposed solution is effective (100% refusal rate), utility-preserving ( 7% utility degradation), robust (5% attack success rate), and scalable to multiple features and clients.

Copied to clipboard

Yuanpu Cao, Ziyi Yin, Fenglong Ma, Jinghui Chen

Language of Thought Shapes Output Diversity in Large Language Models (2026.acl-long)

Challenge:	Existing methods for factual opinion editing focus on atomic facts, ignoring the risks associated with factual opinions.
Approach:	They propose a method that achieves opinion–evidence alignment without relying on explicit instructions to edit factual opinions.
Outcome:	The proposed method achieves opinion–evidence alignment without relying on explicit instructions.

Copied to clipboard

Shaoyang Xu, Wenxuan Zhang

Aligning Backchannel and Dialogue Context Representations via Contrastive LLM Fine-Tuning (2026.acl-long)

Challenge:	Using multiple thinking languages, we show that controlling the language used during model thinking provides a novel and structural source of output diversity.
Approach:	They propose to control the language used during model thinking to provide a novel source of output diversity.
Outcome:	The proposed methods show that controlling the language used during model thinking provides a novel and structural source of output diversity.

Copied to clipboard

Livia Qian, Gabriel Skantze

Closing the Spatial Execution Gap in Digital Whiteboards via Verifiable Reinforcement Learning (2026.acl-long)

Challenge:	Prior work on predicting backchannel timing has focused on lexical form and prosody, but the relationship between lexico-prosodic form and meaning remains underexplored.
Approach:	They propose a framework for fine-tuning large language models on dialogue transcripts to derive rich contextual representations; and a joint embedding space for dialogue contexts and backchannel realizations.
Outcome:	The proposed framework improves context-backchannel retrieval and human perception is more sensitive to extended conversational context and embeddings align more closely with human judgments than raw WavLM features.

Copied to clipboard

Chang Liu, Benjamin Wagley, Zibo Wang, Mehmet E. Belviranli, Bo Wu

ExecVerify: White-Box RL with Verifiable Stepwise Rewards for Code Execution Reasoning (2026.acl-long)

Challenge:	Large language models suffer from a fundamental Spatial Execution Gap, failing to translate visual semantics into precise, schema-valid coordinate operations in interactive environments.
Approach:	They propose a pipeline that leverages Group Relative Policy Optimization to enforce a strict Identify-Reason-Verify protocol and train on execution-verifiable rewards.
Outcome:	The proposed pipeline outperforms a state-of-the-art frontier model by 16.75% in operation accuracy.

Copied to clipboard

Lingxiao Tang, He Ye, Zhaoyang Chu, Muyang Ye, Zhongxin Liu, Xiaoxue Ren, Lingfeng Bao

Prune as You Generate: Online Rollout Pruning for Faster and Better RLVR (2026.acl-long)

Challenge:	Existing methods for code execution reasoning are limited by the difficulty of the training data.
Approach:	They propose a model that uses reinforcement learning to reward correct answers from execution traces.
Outcome:	The proposed model improves pass@1 by up to 5.9% on code generation tasks over strong baselines.

Copied to clipboard

Haobo Xu, Sirui Chen, Ruizhong Qiu, Yuchen Yan, Chen Luo, Monica Xiao Cheng, Jingrui He, Hanghang Tong

Challenge:	Reinforcement Learning with Verifiable Rewards (RLVR) has improved reasoning capabilities of Large Language Models (LLMs).
Approach:	They propose an online pruning method that prunes rollouts while steering correct ones to enhance learning signals.
Outcome:	The proposed method improves average accuracy by +2.30 to +2.99 across GRPO and DAPO on Qwen-3 and LLaMA-3.2 models.

Mitigating Legal Hallucinations via Symbolic Constraints and Analogical Precedents (2026.acl-long)

Copied to clipboard

Zixuan Huang, Yanxiang Ma, Luhan Wang, Yunke Wang, Duo Shi, Chang Xu

How Tokenization Limits Phonological Knowledge Representation in Language Models and How to Improve Them (2026.acl-long)

Challenge:	Existing methods for finetuning and retrieval-augmented generation suffer from hallucination risk and semantic drift.
Approach:	They propose a framework for a dual-retriever based on the legal syllogism and the nature of different legal data.
Outcome:	The proposed framework mitigates hallucinations while improving explainability of legal reasoning.

Copied to clipboard

Disen Liao, Freda Shi

Illusions of the Gold Standard: A Large-scale Analysis of Human Evaluation Protocols for Long-form Text Generation (2026.acl-long)

Challenge:	Tokenization is the first step in every language model (LM), yet it never takes the sounds of words into account.
Approach:	They propose a lightweight IPA-based fine-tuning method that infuses phonological awareness into LMs.
Outcome:	The proposed method improves phonological awareness across three phonology-related tasks while preserving math and general reasoning ability.

Copied to clipboard

Katelyn X. Mei, Yi-Li Hsu, Minjoon Choi, Zongwan Cao, Chenjun Xu, Bingbing Wen, Su Lin Blodgett, Lucy Lu Wang

Omni-RewardBench: Toward a Comprehensive Evaluation of Generative Reward Models Across Modalities (2026.acl-long)

Challenge:	a large-scale analysis of human evaluation protocols for long-form generation tasks is lacking in current practice . current protocols lack proper standardization and operationalization, which can limit validity of evaluation .
Approach:	They conduct a large-scale analysis of human evaluation protocols for long-form generation tasks in *CL conference papers from 2023–2025.
Outcome:	The proposed evaluation protocols lack standardization and operationalization, the authors show . they also find that the evaluation protocols are inadequate for specific domains and tasks .

Copied to clipboard

Chi-Min Chan, Yujin Zhou, Pengcheng Wen, Boqin Yin, Jiaming Ji, Juntao Dai, Wei Xue, Sirui Han, Yike Guo

Challenge:	Existing evaluation benchmarks for ORMs are largely text-centric or limited to bimodal tasks . a new study examines the effectiveness of Omni-RewardBench for ORms across modalities .
Approach:	They propose a hybrid automatic-annotation and human-verification pipeline to construct high-quality evaluation data.
Outcome:	The proposed model is the first benchmark for comprehensive evaluation of ORMs across modalities.

From Answers to Arguments: Toward Trustworthy Clinical Diagnostic Reasoning with Toulmin-Guided Curriculum Goal-Conditioned Learning (2026.acl-long)

Copied to clipboard

Chen Zhan, Xiaoyu Tan, Gengchen Ma, Yu-Jie Xiong, Xiaoyan Jiang, Xihe Qiu

Verifiable LLM-Generated Text Detection via Projected Semantic-Structural Distributions (2026.acl-long)

Challenge:	Large Language Models (LLMs) are obstructed by their opaque and often unreliable reasoning.
Approach:	They propose a framework for trustworthy clinical argumentation by adapting the Toulmin model to the diagnostic process.
Outcome:	The proposed method achieves diagnostic accuracy comparable to resource-intensive RL methods while offering a more stable and efficient training pipeline.

Copied to clipboard

Ruochong Xiong, Qien Li, Wangwang Lian, Yulong Wan, Hanlin Xue, Zhouxing Tan, Han Yang, Fengyu Lu, Junfei Liu

Is Human-Like Text Liked by Humans? Multilingual Human Detection and Preference Against AI (2026.acl-long)

Challenge:	Existing methods for detecting LLM-Generated text suffer from distribution misalignment and limited interpretability.
Approach:	They propose a statistical framework utilizing supervised subspace learning to extract compact features and construct conditional semantic distributions based on syntactic structures.
Outcome:	The proposed framework is superior in cross-domain, cross-model, and adversarial scenarios.

Copied to clipboard

Yuxia Wang, Rui Xing, Jonibek Mansurov, Giovanni Puccetti, Zhuohan Xie, Minh Ngoc Ta, Jiahui Geng, Jinyan Su, Mervat Abassy, Saadeldine Eletter, Kareem Elozeiri, Nurkhan Laiyk, Maiya Goloburda, Tarek Mahmoud, Raj Vardhan Tomar, Alexander Aziz, Ryuto Koike, Masahiro Kaneko, Artem Shelmanov, Ekaterina Artemova, Vladislav Mikhailov, Akim Tsvigun, Alham Fikri Aji, Nizar Habash, Iryna Gurevych, Preslav Nakov

Powerful Training-Free Membership Inference Against Fine-Tuned Autoregressive Language Models (2026.acl-long)

Challenge:	Prior studies have shown that distinguishing text generated by Large Language Models from human-written text is challenging for humans and often no better than random guessing.
Approach:	They conduct extensive case study to determine the upper bound of human detection accuracy.
Outcome:	The findings challenge previous conclusions on human detection accuracy across languages and domains.

Copied to clipboard

David Ilić, David Stanojević, Kostadin Cvejoski

Communication-Efficient Desire Alignment for Proactive Embodied Human–Agent Interaction (2026.acl-long)

Challenge:	Existing methods for auditing fine-tuned language models have limited detection rates . membership inference attacks aim to determine if a specific record was in a model's training set .
Approach:	They propose a membership inference attack that exploits memorization at error positions . EZ-MIA achieves 3.8 higher detection than previous state-of-the-art .
Outcome:	The proposed attack achieves 3.8 higher detection than previous state-of-the-art models . EZ-MIA achieves 8 higher detectability than prior work, requiring no model training .

Copied to clipboard

Yuanfei Wang, Xinju Huang, Fangwei Zhong, Yaodong Yang, Yizhou Wang, Yuanpei Chen, Hao Dong

When Seeing Overrides Knowing: Disentangling Knowledge Conflicts in Vision-Language Models (2026.acl-long)

Challenge:	Effective real-world human–agent interactions are long-term and repeated.
Approach:	They propose a simulation that uses a proxy user with value-driven preferences and natural language behavior to evaluate how agents adapt to users across interactions and satisfy their desires.
Outcome:	HA-Desire, a home assistance simulation, shows that agents can adapt to user needs and provide proactive assistance within limited communication.

Copied to clipboard

Francesco Ortu, Zhijing Jin, Diego Doimo, Alberto Cazzaniga

ICDAGENT: Empowering Agentic Large Language Models for Explainable Medical Coding (2026.acl-long)

Challenge:	Vision-language models combine visual and textual information to perform complex tasks. conflicts between internal knowledge and external visual input can lead to hallucinations and unreliable predictions.
Approach:	They propose to use a dataset to identify attention heads that deliberately contradict internal commonsense knowledge to resolve cross-modal conflicts.
Outcome:	The proposed model can be manipulated to find out which visual inputs are conflicting . the model can then be orientated towards internal parametric knowledge or visual information .

Copied to clipboard

Ziyi Yin, Yuanpu Cao, Ting Wang, Jinghui Chen, Fenglong Ma

Hyperion: Private Token Sampling with Homomorphic Encryption (2026.acl-long)

Challenge:	Existing models lack convincing, human-understandable explanations, making them difficult for physicians to trust and use in practice.
Approach:	They propose a framework that aims to automatically assign ICD codes to clinical notes while providing explicit justifications for each assignment.
Outcome:	The proposed framework achieves effective ICD coding with accurate explanations using two collaborative LLM agents: a coding agent and a critical agent.

Copied to clipboard

Lawrence Lim, Jiaming Liu, Vikas Kalagi, Divyakant Agrawal, Amr El Abbadi

Map of Encoders – Mapping Sentence Encoders using Quantum Relative Entropy (2026.acl-long)

Challenge:	Large language models (LLMs) enable an extraordinary range of applications, yet their adoption is limited by a fundamental concern: users must send sensitive data to remote model providers.
Approach:	They propose an efficient homomorphic encryption algorithm for inverse transform sampling that allows private token sampling under HE.
Outcome:	The proposed method samples tokens in 0.14 seconds on GPU, achieving a 100 latency improvement over prior work.

Copied to clipboard

Gaifan Zhang, Danushka Bollegala

Mango: Multi-Agent Web Navigation via Global-View Optimization (2026.acl-long)

Challenge:	a method to compare and visualise sentence encoders at scale is proposed . we map encoder LLMs using QRE-based feature vectors, which are then projected to 2D .
Approach:	They propose a method to compare and visualise sentence encoders at scale by creating a map of encoder . they construct a QRE-based map of sentences covering 1101 publicly available sentence encoded sentences .
Outcome:	The proposed method compares sentence encoders at scale by creating a map of encoder models . it shows that the map accurately reflects relationships between encoder and unit base encoder .

Copied to clipboard

Weixi Tong, Yifeng Di, Tianyi Zhang

Challenge:	Existing web agents typically begin exploration from the root URL, which is inefficient for complex websites with deep hierarchical structures.
Approach:	They propose a multi-agent web navigation method that leverages the website structure to dynamically determine optimal starting points.
Outcome:	The proposed method achieves 63.6% success rate on WebVoyager, outperforming the best baseline by 7.3%, and 52.5% success rate with open-source and closed-source models.

DPWriter: Reinforcement Learning with Diverse Planning Branching for Creative Writing (2026.acl-long)

Copied to clipboard

Qian Cao, Yahui Liu, Wei Bi, Yi Zhao, Ruihua Song, Xiting Wang, Ruiming Tang, Guorui Zhou, Han Li

Challenge:	Existing methods for enhancing large language models (LLMs) lack explicit mechanisms for guiding diverse exploration and instead prioritize efficiency and performance over diversity.
Approach:	They propose a reinforcement learning-based framework that decomposes the generation process into explicitly planned intermediate steps and introduces divergence at the planning phase based on diversity variation.
Outcome:	The proposed method significantly outperforms existing baselines on creative writing benchmarks on a semi-structured long chain-of-thought (CoT) it introduces divergence at the planning phase based on diversity variation, alongside a group-aware diversity reward to encourage distinct trajectories.

Iterative Dual-Model Alignment for Story Evaluation (2026.acl-long)

Copied to clipboard

Bruce Qin, Dan Goldwasser

Discovery and Reinforcement of Tool-Integrated Reasoning Chains via Rollout Trees (2026.acl-long)

Challenge:	Existing evaluators of large language models are static and lack the ability to refine their reasoning through interaction.
Approach:	They propose an Alpha–Beta Learning framework that trains two complementary 8B models: an Alpha classifier that assesses pairwise story engagement, and a Beta generator that produces structured, rubric-guided comparative explanations.
Outcome:	The proposed framework outperforms strong single-model baselines on human-annotated story-pair datasets in both accuracy and explanation quality across multiple iterative rounds.

Copied to clipboard

Kun Li, Zenan Xu, Junan Li, Zengrui Jin, Jinghao Deng, Zexuan Qiu, Bo Zhou

When Identity Skews Debate: Anonymization for Bias-Reduced Multi-Agent Reasoning (2026.acl-long)

Challenge:	Existing approaches to augment Large Language Models (LLMs) with computational capabilities have focused on short Chain-of-thought (CoT) integrating tool-use into long CoT remains underexplored due to the scarcity of training data and the challenge of integrating it without compromising the model’s intrinsic long-chain reasoning.
Approach:	They propose a framework that enables spontaneous tool-use during long CoT reasoning without additional human annotation.
Outcome:	Experiments on AIME and GPQA-Diamond show that DART significantly outperforms existing methods, successfully harmonizing tool execution with long CoT reasoning.

Copied to clipboard

Hyeong Kyu Choi, Jerry Zhu, Sharon Li

Talking to a Know-It-All GPT or a Second-Guesser Claude? How Repair reveals distinct Multi-Turn Behavior in LLMs (2026.acl-long)

Challenge:	Multi-agent debate (MAD) aims to improve large language model reasoning by letting multiple agents exchange answers and then aggregate their opinions.
Approach:	They propose a principled framework that joins sycophancy and self-bias to mitigate and quantify identity bias in multi-agent debate by removing identity markers from prompts.
Outcome:	The proposed framework joins identity-driven sycophancy and self-bias to mitigate and quantify identity bias in multi-agent debate.

Copied to clipboard

Clara Lachenmaier, Hannah Bultmann, Sina Zarrieß

HistLens: Mapping Idea Change across Concepts and Corpora (2026.acl-long)

Challenge:	a lot of research aims to mitigate these problems by introducing specific computational solutions.
Approach:	They examine how large language models engage in the interactive process of repair in multi-turn dialogues around solvable and unsolvable math questions.
Outcome:	The models respond to user-initiated repair differently from one another . the models exhibit their own characteristic form of unreliability in the context of repair .

Copied to clipboard

Yi Jing, Weiyun Qiu, Yihang Peng, Zhifang Sui

ReFEree: Reference-Free and Fine-Grained Method for Evaluating Factual Consistency in Real-World Code Summarization (2026.acl-long)

Challenge:	Existing approaches to diachronic semantics and discourse analysis focus on a single concept or corpus, argues a new paper.
Approach:	They propose a framework for multi-concept, multi-corpus conceptual-history analysis that decomposes concept representations into interpretable features and tracks activation dynamics over time and across sources.
Outcome:	The proposed framework decomposes concept representations into interpretable features and tracks their activation dynamics over time and across sources.

Copied to clipboard

Suyoung Bae, CheolWon Na, Jaehoon Lee, Yumin Lee, YunSeok Choi, Jee-Hyong Lee

MeanAudio: Fast and Faithful Text-to-Audio Generation with Mean Flows (2026.acl-long)

Challenge:	Existing methods for evaluating factual consistency are primarily designed for short summaries of isolated code snippets.
Approach:	They propose a reference-free and fine-grained method for evaluating factual consistency in real-world code summaries.
Outcome:	The proposed method achieves highest correlation with human judgment among 13 baselines, improving 15-18% over the previous state-of-the-art.

Copied to clipboard

Xiquan Li, Junxi Liu, Yuzhe Liang, Zhikang Niu, Wenxi Chen, Xie Chen

ModeX: Evaluator-Free Best-of-N Selection for Open-Ended Generation (2026.acl-long)

Challenge:	Recent advances in Text-to-Audio Generation (TTA) systems suffer from slow inference speed, authors report . authors demonstrate that MeanAudia achieves state-of-the-art performance in single-step audio generation .
Approach:	They propose a text-to-audio generator capable of rendering realistic sound with only one function evaluation.
Outcome:	The proposed system achieves state-of-the-art performance in single-step audio generation.

Copied to clipboard

Hyeong Kyu Choi, Sharon Li

What Makes LLMs Effective Sequential Recommenders? A Study on Preference Intensity and Temporal Context (2026.acl-long)

Challenge:	Existing approaches to selecting a single high-quality output from multiple generations are limited in their applicability and efficiency.
Approach:	They propose a method that generalizes majority voting to open-ended text generation . modeX leverages structural information across multiple generation paths to select a "modal" output .
Outcome:	The proposed framework outperforms standard single- and multi-path baselines in open-ended tasks.

Copied to clipboard

Zhongyu Ouyang, Qianlong Wen, Chunhui Zhang, Yanfang Ye, Soroush Vosoughi

Challenge:	Existing preference-alignment approaches rely on binary pairwise comparisons, overlooking preference intensity and temporal context.
Approach:	They propose a unified preference optimization framework that maps both explicit and implicit feedback into a common preference signal and constructs adaptive reward margins that jointly account for preference intensity and interaction recency.
Outcome:	The proposed framework outperforms state-of-the-art recommendations while maintaining behavioral patterns aligned with human decision-making.

TeamFusion: Supporting Open-ended Teamwork with Multi-Agent Systems (2026.acl-long)

Copied to clipboard

Jiale Liu, Victor Bursztyn, Lin Ai, Haoliang Wang, Sunav Choudhary, Saayan Mitra, Qingyun Wu

Challenge:	Many group decisions are open-ended, and aggregation approaches suppress minority perspectives . team members must surface hidden assumptions, discuss disagreements, negotiate acceptable trade-offs .
Approach:	They propose a multi-agent system that instantiates a proxy agent for each team member . they also conduct a structured discussion to elicit agreements and disagreements .
Outcome:	The proposed system outperforms direct aggregation on two teamwork tasks . it can judge how well individual views are represented in team decisions and consensually good deliverables .

FormulaSPIN: Self-Play Fine-Tuning for Natural Language to Spreadsheet Formula Generation (2026.acl-long)

Copied to clipboard

Cy Xie

Stable Signer: Hierarchical Sign Language Generative Model (2026.acl-long)

Challenge:	Existing approaches to writing formulas rely on static supervised data, which quickly saturates on limited annotations.
Approach:	They propose a self-play framework that breaks the ceiling of supervised fine-tuning by enabling iterative self-improvement without any additional data.
Outcome:	The proposed framework outperforms existing approaches to fine-tuning on static data while enabling iterative self-improvement without additional data.

Copied to clipboard

Sen Fang, Yalin Feng, Hongbin Zhong, Yanxin Zhang, Dimitris N. Metaxas

DisCal: Distribution-Aware Calibration for Mathematical Reasoning Under Character-Level Noisy Inputs (2026.acl-long)

Challenge:	Sign Language Production (SLP) is the process of converting complex input text into a real video.
Approach:	They propose a new sign language generative model that streamlines redundant structure and optimizes the task objective.
Outcome:	The proposed model streamlines redundant structure and optimizes objective . it generates high-quality and multi-style sign language videos with hand gestures .

Copied to clipboard

Bo Zhang, Jiawei Zhang, Cong Gao, Bingxu Han, Minghao Hu, Jun Zhang, Yunbo Cao, Zhunchen Luo, Wen Yao, Guotong Geng, Zhong Wang

SARA: Selective and Adaptive Retrieval-augmented Generation with Context Compression (2026.acl-long)

Challenge:	Existing methods for calibration of large reasoning models (LRMs) focus on clean inputs, leaving noise unexplored.
Approach:	They propose a confidence calibration framework for character-level noisy inputs that extracts uncertainty signals from both the empirical answer distribution and the model’s predictive distribution and integrates them via a learned calibrator.
Outcome:	Experiments on multiple mathematical reasoning benchmarks show that DisCal outperforms existing calibration methods under noisy inputs, reducing expected calibration error (ECE) by up to 39.21% and improving Area Under the Receiver Operating Characteristic Curve (AUROC) by 31.44%.

Copied to clipboard

Yiqiao Jin, Kartik Sharma, Vineeth Rakesh, Yingtong Dou, Menghai Pan, Mahashweta Das, Srijan Kumar

FinChain: A Symbolic Benchmark for Verifiable Chain-of-Thought Financial Reasoning (2026.acl-long)

Challenge:	Retrieval-augmented generation (RAG) extends large language models with external knowledge, but it must balance limited effective context, redundant retrieved evidence, and the loss of fine-grained facts.
Approach:	They propose a hybrid RAG framework that uses natural-language snippets and semantic compression vectors to preserve passages in text form and compress remaining evidence into interpretable vectors for iterative evidence reranking.
Outcome:	The proposed framework improves answer relevance, answer correctness and semantic similarity across 9 datasets and 5 open-source LLMs.

Copied to clipboard

Zhuohan Xie, Daniil Orel, Rushil Thareja, Dhruv Sahnan, Hachem Madmoun, Fan Zhang, Debopriyo Banerjee, Georgi Nenkov Georgiev, Xueqing Peng, Lingfei Qian, Jimin Huang, Jinyan Su, Aaryamonvikram Singh, Rui Xing, Rania Elbadry, Chen Xu, Haonan Li, Fajri Koto, Ivan Koychev, Tanmoy Chakraborty, Yuxia Wang, Salem Lahlou, Veselin Stoyanov, Sophia Ananiadou, Preslav Nakov

AI use in American newspapers is widespread, uneven, and rarely disclosed (2026.acl-long)

Challenge:	Existing benchmarks emphasize final numerical answers while neglecting intermediate reasoning steps.
Approach:	They propose a symbolic benchmark for verifiable Chain-of-Thought evaluation in finance . FINCHAIN spans 58 topics across 12 financial domains and three difficulty levels .
Outcome:	The proposed benchmark aims to bridge symbolic reasoning and factual verification.

Copied to clipboard

Jenna Russell, Marzena Karpinska, Destiny Akinode, James Zhou, Katherine Thai, Bradley Emi, Max Spero, Mohit Iyyer

Challenge:	a large-scale dataset of 186K articles from 1.5K newspapers published in the summer of 2025 is audited.
Approach:	They audit 186K articles from 1.5K newspapers published in summer of 2025 . they use Pangram, a state-of-the-art AI detector, to detect whether articles are partially or fully AI-generated .
Outcome:	The findings highlight the need for greater transparency and updated editorial standards regarding the use of AI in journalism to maintain public trust.

For-Value: Efficient Forward-Only Data Valuation for finetuning LLMs and VLMs (2026.acl-long)

Copied to clipboard

Wenlong Deng, Qi Zeng, Jiaming Zhang, Minghui Chen, Zixin Ding, Christos Thrampoulidis, Boying Gong, Xiaoxiao Li

The Illusion of Specialization: Unveiling the Domain-Invariant "Standing Committee" in Mixture-of-Experts Models (2026.acl-long)

Challenge:	Existing methods for data valuation rely on gradient computations, making them prohibitive for billion-parameter models.
Approach:	They propose a forward-only data valuation framework that enables efficient batch-scalable value estimation while maintaining effectiveness.
Outcome:	The proposed framework matches or outperforms gradient-based baselines in detecting influential data and mislabeled data while achieving significant efficiency improvements.

Copied to clipboard

Yan Wang, Yitao Xu, Nanhan Shen, Jinyan Su, Jimin Huang, Zining Zhu

Challenge:	Mixture of Experts models are widely assumed to achieve domain specialization through sparse routing.
Approach:	They propose a framework that analyzes routing behavior at the level of expert groups rather than individual experts.
Outcome:	The proposed framework analyzes routing behavior at the level of expert groups rather than individual experts.

Find Your Optimal Teacher: Personalized Data Synthesis via Router-Guided Multi-Teacher Distillation (2026.acl-long)

Copied to clipboard

Hengyuan Zhang, Shiping Yang, Xiao Liang, Chenming Shang, Yuxuan Jiang, Chaofan Tao, Jing Xiong, Hayden Kwok-Hay So, Ruobing Xie, Angel X. Chang, Ngai Wong

Single-Pass, Depth-Selective Reading for Multi-Aspect Sentiment Analysis (2026.acl-long)

Challenge:	Existing studies show that stronger models are not always optimal teachers, suggesting a mismatch between the teacher’s output and the student’s learning ability.
Approach:	They propose a method that routes each prompt to its optimal teacher via a query-level router that jointly considers the student models’ learnability and teacher models’ response quality.
Outcome:	The proposed method outperforms baselines on six benchmarks including instruct tuning and math reasoning settings.

Copied to clipboard

Yan Xia, Zhuangzhuang Pan, Amirrudin Kamsin, Chee Seng Chan

PARASITE: Conditional System Prompt Poisoning to Hijack LLMs (2026.acl-long)

Challenge:	Existing models re-encode the sentence for each aspect or rely on static use of deep representations, leading to redundant computation and limited adaptivity.
Approach:	They propose a single-pass inference framework that encodes each sentence once to construct a reusable, depth-ordered substrate.
Outcome:	Experiments show that DABS reduces end-to-end computation by 60% in multi-aspect settings.

Copied to clipboard

Viet Pham, Thai Le

Generating Literature-Driven Scientific Theories at Scale (2026.acl-long)

Challenge:	Large Language Models (LLMs) are increasingly deployed via third-party system prompts downloaded from public marketplaces.
Approach:	They propose a framework that optimizes system prompts to trigger LLMs to output compromised responses only for specific queries.
Outcome:	The proposed framework achieves up to 70% F1 reduction on targeted queries with minimal degradation to general capabilities.

Copied to clipboard

Peter Jansen, Peter Clark, Doug Downey, Daniel S Weld

Controllable Memory Usage: Balancing Anchoring and Innovation in Long-Term Human–Agent Interaction (2026.acl-long)

Challenge:	Contemporary automated scientific discovery systems focus on generating experiments, but higher-level activities such as theory building remain underexplored.
Approach:	They propose to synthesize theories from scientific literature using literature-grounding versus parametric knowledge.
Outcome:	The proposed method matches existing evidence better than parametric LLM memory generation.

Copied to clipboard

Zisu Huang, Muzhao Tian, Xiaohua Wang, Jingwen Xu, Zhengkang Guo, Qi Qian, Kaitao Song, Jiakang Yuan, Changze Lv, Xiaoqing Zheng

Challenge:	Existing systems that use memory as an "all-or-nothing" approach to memory usage are often static and rely on experience-following tendencies.
Approach:	They propose a framework that allows users to dynamically regulate memory reliance by adding context into the model's prompt.
Outcome:	The proposed model outperforms prompting and memory masking strategies in multiple scenarios.

Digitizing Nepal’s Written Heritage: A Comprehensive HTR Pipeline for Old Nepali Manuscripts (2026.acl-long)

Copied to clipboard

Anjali Sarawgi, Esteban Garces Arias, Christof Zotter

XMark: Reliable Multi-Bit Watermarking for LLM-Generated Texts (2026.acl-long)

Challenge:	Using a line-level transcription approach, we explore encoder-decoder architectures and data-centric techniques to improve recognition accuracy for Old Nepali manuscripts.
Approach:	They propose a line-level transcription approach and explore encoder-decoder architectures and data-centric techniques to improve recognition accuracy.
Outcome:	The proposed model achieves a 4.9% error rate and is highly reliable.

Copied to clipboard

Jiahao Xu, Rui Hu, Olivera Kotevska, Zikai Zhang

Believing without Seeing: Quality Scores for Contextualizing Vision-Language Model Explanations (2026.acl-long)

Challenge:	Existing methods for embedding binary messages into LLM-generated text suffer from key limitations, such as a poor trade-off between text quality and decoding accuracy.
Approach:	They propose a method for embedding binary messages into Large Language Model (LLM)-generated text that uses a limited number of tokens to decode and recover the encoded message.
Outcome:	The proposed method significantly outperforms existing methods in multiple downstream tasks and will be made publicly available upon acceptance.

Copied to clipboard

Keyu He, Tejas Srinivasan, Brihi Joshi, Xiang Ren, Jesse Thomason, Swabha Swayamdipta

MONETA: Multimodal Industry Classification through Geographic Information with Multi Agent Systems (2026.acl-long)

Challenge:	Prior work has found that explanations can easily convince users that inaccurate VLM predictions are correct.
Approach:	They propose to evaluate two complementary qualities of VLM-generated explanations via two quality scoring functions to improve their accuracy.
Outcome:	The proposed explanations improve accuracy on the A-OKVQA, VizWiz, and MMMU-Pro tasks by 11.1%, including a 15.4% reduction in falsely believing incorrect predictions.

Copied to clipboard

Arda Yüksel, Gabriel Thiem, Susanne Walter, Patrick Felka, Gabriela Alves Werb, Ivan Habernal

CASPER in the Machine: Insights into Character Variety in LLM-Generated Stories (2026.acl-long)

Challenge:	MONETA is the first multimodal industry classification benchmark with text and geospatial sources.
Approach:	They propose a multimodal industry classification benchmark using text and geospatial sources.
Outcome:	The proposed model increases the accuracy of the existing models by 22.80%.

Copied to clipboard

Anneliese Brei, Abhisheik Sharma, Nicholas Sanaie, Lu Wang, Snigdha Chaturvedi

Language Models Struggle to Use Representations Learned In-Context (2026.acl-long)

Challenge:	Increasing numbers of authors are using AI to assist in the process of writing stories.
Approach:	They analyze 8 category-pairs of character that assess how characters are portrayed in short stories . they find similarities between LLMs and human-written stories based on categories .
Outcome:	The analysis includes questions on popular LLMs and recently published human-written stories.

Copied to clipboard

Michael A. Lepori, Tal Linzen, Ann Yuan, Katja Filippova

SlideAgent: Hierarchical Agentic Framework for Multi-Page Visual Document Understanding (2026.acl-long)

Challenge:	a recent study shows that large language models are capable of inducing rich representations of data that are seen in-context . a novel task, adaptive world modeling, shows that even the most performant LLMs cannot reliably leverage novel semantics defined in-constitut.
Approach:	They propose to use in-context representations to induce rich representations of data . they also propose to probe models using a novel task to enable flexible deployment .
Outcome:	The proposed model can use in-context representations to complete simple downstream tasks.

Copied to clipboard

Yiqiao Jin, Rachneet Kaur, Zhen Zeng, Sumitra Ganesh, Srijan Kumar

YIELD: A Large-Scale Dataset and Evaluation Framework for Information Elicitation Agents (2026.acl-long)

Challenge:	Multimodal large language models (MLLMs) are a promising tool for document understanding, but they are not able to handle complex multi-page visual documents.
Approach:	They propose a flexible agentic framework for understanding multi-modal, multi-page, and multi-layout documents . SlideAgent employs specialized agents and decomposes reasoning into three specialized levels .
Outcome:	a new agentic framework improves accuracy over open-source and proprietary models . it decomposes reasoning into three levels to capture themes and visual cues . the framework is based on a multimodal large language model and a MLLM .

Copied to clipboard

Victor De Lima, Grace Hui Yang

Spatial-Agent: Agentic Geo-spatial Reasoning with Scientific Core Concepts (2026.acl-long)

Challenge:	Existing conversational agents (CAs) are designed to satisfy user needs through user-driven interactions. however, many real-world settings, such as academic interviewing, require agents that can elicit information from users.
Approach:	They propose to support Information Elicitation Agents (IEAs) in which the agent’s goal is to elicit information from users to support the agent's institutional or task-oriented objectives.
Outcome:	The proposed agent-based model improves the performance of a 26M-token dataset of 2,281 human-to-human dialogues on multiple foundation LLMs and human evaluation confirms the results.

Copied to clipboard

Riyang Bao, Cheng Yang, Dazhou Yu, Zhexiang Tang, Gengchen Mai, Liang Zhao

Rethinking Evaluation for LLM Hallucination Detection: A Desiderata, A New RAG-based Benchmark, New Insights (2026.acl-long)

Challenge:	Existing LLM-based agents lack inherent spatial awareness, relying on web search or text matching while hallucinating spatial relationships.
Approach:	They propose a spatial-based agent that can perform real-world geospatial computations . they use natural-language questions to parse into executable workflows based on geoFlow Graphs - directed acyclic graphs with nodes corresponding to spatial concepts and edges representing transformations.
Outcome:	The proposed agent outperforms existing baselines on MapEval-API and MapQA benchmarks while producing interpretable and executable geospatial workflows.

Copied to clipboard

Wenbo Chen, Veena Padmanabhan, Tootiya Giyahchi, Elaine Wong, Leman Akoglu

Min-k Sampling: Decoupling Truncation from Temperature Scaling via Relative Logit Dynamics (2026.acl-long)

Challenge:	Existing benchmarks lack long context and label noise for stress-testing detectors . a new RAG-based HDB that underwent a rigorous human annotation process is developed .
Approach:	They propose a desiderata of properties for hallucination detection benchmarks to exhibit . they build a RAG-based HDB that underwent a rigorous human annotation process .
Outcome:	The proposed benchmark exhibits all desirable properties of existing HDBs . existing benchmarks lack realistic label noise for stress-testing detectors despite human annotation .

Copied to clipboard

Yuanhao Ding, Meimingwei Li, Esteban Garces Arias, Matthias Aßenmacher, Christian Heumann, Chongsheng Zhang

Challenge:	Existing methods for decoding large language models have extreme sensitivity to temperature parameter T.
Approach:	They propose a dynamic truncation strategy that analyzes the local shape of the logit distribution to identify "semantic cliffs" they show that Min-k consistently improves text quality even under extreme temperature settings .
Outcome:	The proposed method achieves strict temperature invariance and low sensitivity to hyperparameter choices.

Bridging the Memorization-Utilization Gap: Near-Lossless Context Compression via Reinforcement Learning (2026.acl-long)

Copied to clipboard

Yujan Ting, Xu Tang, Terrence Chen, Weijing Huang

PRIME: A Process-Outcome Alignment Benchmark for Verifiable Reasoning in Mathematics and Engineering (2026.acl-long)

Challenge:	Recent advances in context compression have failed to effectively utilize compressed representations for downstream tasks.
Approach:	They propose a holistic training paradigm that uses outcome-based RL to enable implicit expansion.
Outcome:	The proposed model outperforms previous models on NIAH, LongBench and multi-hop reasoning.

Copied to clipboard

Xiangfeng Wang, Hangyu Guo, Yanlin Lai, Mitt Huang, Liang Zhao, Chengyuan Yao, Yinmin Zhang, Qi Han, null Xiaoxiaoren, Chun Yuan, Tong Xu, Zheng Ge, Xiangyu Zhang, Daxin Jiang

ZARA: Training-Free Motion Time-Series Reasoning via Evidence-Grounded LLM Agents (2026.acl-long)

Challenge:	Current outcome-centric verification paradigms neglect potential errors in the derivation process.
Approach:	They propose a process-aware RLVR training paradigm utilizing verifiers selected via PRIME.
Outcome:	The proposed approach outperforms the baseline verification paradigm on AIME24, AIME25, and Beyond-AIME models.

Copied to clipboard

Zechen Li, Baiyu Chen, Hao Xue, Flora D. Salim

AutoJudger: An Agent-Driven Framework for Efficient Benchmarking of MLLMs (2026.acl-long)

Challenge:	Existing approaches to human activity recognition are constrained to fixed activity sets . lack of training-free adaptation to new behavior leads to hallucinations and weak grounding .
Approach:	They propose a knowledge- and retrieval-augmented agentic framework for motion time-series reasoning in a training-free inference setting.
Outcome:	The proposed framework generalizes robustly to unseen subjects and across datasets . it can be used to train-free inference in a training-free environment .

Copied to clipboard

Xuanwen Ding, Chengjun Pan, Zejun Li, Jiwen Zhang, Siyuan Wang, Zhongyu Wei

Implicit Representations of Grammaticality in Language Models (2026.acl-long)

Challenge:	Evaluating multimodal large language models (MLLMs) is becoming increasingly expensive as benchmarks grow in scale and cross-modality complexity.
Approach:	They propose an adaptive evaluation framework for efficient benchmarking that treats evaluation as an interview-like process by keeping a hypothesized ability structure of the evaluated model and actively selecting the informative questions.
Outcome:	Experiments on four representative multimodal benchmarks show that **A2-Judger significantly improves sample efficiency while maintaining reliable evaluation results.

Copied to clipboard

Yingshan Susan Wang, Linlu Qiu, Zhaofeng Wu, Roger P. Levy, Yoon Kim

Discovering and Causally Validating Emotion-Sensitive Neurons in Large Audio-Language Models (2026.acl-long)

Challenge:	Pretrained language models generate grammatically well-formed text and discriminate well between grammatical and ungrammatically sentences in tightly controlled minimal pairs.
Approach:	They propose a method to train pretrained LMs for representations of grammaticality by applying perturbations to a naturalistic text corpus.
Outcome:	The proposed model outperforms probability-based models on human-curated grammaticality judgment benchmarks and performs worse than string probabilities on plausibility benchmarks.

Copied to clipboard

Xiutian Zhao, Björn Schuller, Berrak Sisman

Uncovering Intervention Opportunities for Suicide Prevention with Language Model Assistants (2026.acl-long)

Challenge:	Emotion is a central dimension of spoken communication, yet we lack a mechanistic account of how LALMs encode it internally.
Approach:	They propose to use emotion-sensitive neurons in large audio-language models to study their interpretations.
Outcome:	The proposed models show that they can be used to make decisions on emotion . the results show that the ESNs exhibit non-uniform clustering with partial cross-dataset transfer .

Copied to clipboard

Jaspreet Ranjit, Hyundong Justin Cho, Claire J. Smerdon, Yoonsoo Nam, Myles Phung, Jonathan May, John R. Blosnich, Swabha Swayamdipta

The Imperfective Paradox in Large Language Models (2026.acl-long)

Challenge:	Using language models, annotators can help develop novel suicide interventions . 85% of cases where LM predictions disagree with existing annotations are analyzed .
Approach:	They propose a human-in-the-loop algorithm that leverages language models as an assistant to annotators and experts to facilitate data-driven insights from NVDRS data.
Outcome:	The proposed algorithm can be used to support the development of novel suicide interventions . it finds that LM predictions match existing data annotations about 85% of the time .

Copied to clipboard

Bolei Ma, Yusuke Miyao

MalruleLib: Large-Scale Executable Misconception Reasoning with Step Traces for Modeling Student Thinking in Mathematics (2026.acl-long)

Challenge:	Existing models rely on surface-level probabilistic heuristics to grasp compositional semantics of events . authors: current open-weight models operate as predictive narrative engines rather than faithful reasoners .
Approach:	They propose a diagnostic dataset to probe the imperfective paradox . they uncover a pervasive Teleological Bias in open-weight models .
Outcome:	The proposed dataset reveals a pervasive Teleological Bias in open-weight models . the findings suggest that these models operate as predictive narrative engines rather than faithful reasoners .

Copied to clipboard

Xinghe Chen, Naiming Liu, Shashank Sonkar

How to Improve LLMs’ Performance on Specific Languages: A Perspective on LLM-Derived Language Similarity (2026.acl-long)

Challenge:	MalruleLib is a learning-science-grounded framework that translates documented misconceptions into executable procedures and generates step-by-step traces of malrule-consistent student reasoning.
Approach:	They propose a learning-science-grounded framework that translates documented misconceptions into executable procedures and generates step-by-step traces of malrule-consistent student reasoning.
Outcome:	The framework translates misconceptions into executable procedures and generates step-by-step traces of malrule-consistent student reasoning.

Copied to clipboard

Xinhe Shi, Qingcheng Zeng, Weihao Xuan, Linchao Zhu

MATCH: Modulating Attention via In-Context Retrieval for Long-Context Transformers (2026.acl-long)

Challenge:	Large language models (LLMs) exhibit uneven performance across languages.
Approach:	They propose to use a framework to quantify the similarity within each language pair through both the lenses of language-specific performance patterns and cross-lingual transferability.
Outcome:	The proposed approach outperforms traditional linguistic typology and cross-lingual transferability measures on multilingual LLMs.

Copied to clipboard

Linrui Ma, Chun Hei Lo, Xinyu Wang, Peng Lu, Xihao Yuan, Hanting Chen, Kai Han, Xinghao Chen, Chengjun Zhan, Hanlin xu, Yichun Yin, Lifeng Shang, Feng Wen, Boxing Chen, Yufei Cui

Protecting Bystander Privacy via Selective Hearing in Audio LLMs (2026.acl-long)

Challenge:	Existing approaches to improve efficiency often enforce rigid structural constraints such as local attention windows.
Approach:	They propose a framework that augments sparse-attention mechanisms with dynamically integrated in-context information through an efficient retrieval system.
Outcome:	Empirical results show that MATCH significantly improves the performance of sparse-attention models on synthetic and real-world natural-language tasks.

Copied to clipboard

Xiao Zhan, Guangzhi Sun, Jose Such, Phil Woodland

LiveCLKTBench: Towards Reliable Evaluation of Cross-Lingual Knowledge Transfer in Multilingual LLMs (2026.acl-long)

Challenge:	Audio Large language models capture speech from unintended bystanders, raising privacy risks that existing benchmarks and defences did not consider.
Approach:	They propose to evaluate selective hearing by evaluating a model’s ability to attend to an intended main speaker while refusing to process or reveal information about incidental bystander speech.
Outcome:	The proposed model can attend to an intended main speaker while refusing to process or reveal information about incidental bystander speech.

Copied to clipboard

Pei-Fu Guo, Yun-Da Tsai, Chun-Chia Hsu, Kai-Xin Chen, Ya An Tsai, Kai-Wei Chang, Nanyun Peng, Mi-Yen Yeh, Shou-De Lin

To Lie or Not to Lie? Investigating The Biased Spread of Global Lies by LLMs (2026.acl-long)

Challenge:	Evaluating cross-lingual knowledge transfer in large language models is challenging, as correct answers in a target language may arise either from genuine transfer or from prior exposure during pre-training.
Approach:	They propose a pipeline to isolate and measure cross-lingual knowledge transfer by identifying self-contained, time-sensitive knowledge entities from real-world domains and generating factual questions.
Outcome:	The proposed pipeline analyzes multiple LLMs across five languages and shows that cross-lingual transfer is strongly influenced by linguistic distance and often asymmetric across language directions.

Copied to clipboard

Zohaib Khan, Mustafa Dogan, Ifeoma Okoh, Pouya Sadeghi, Siddhartha Shrestha, Sergius Justus Chesami Nyah, Mahmoud O. Mokhiamar, Michael J Ryan, Tarek Naous

Robertha: Eigenspectrum Regularized Attention for Robust Natural Language Understanding (2026.acl-long)

Challenge:	Misinformation is on the rise, and the strong writing capabilities of LLMs lower the barrier for malicious actors to produce and disseminate false information.
Approach:	They introduce a multilingual parallel dataset of 440 misinformation generation prompt templates and 6,867 entities, spanning 8 languages and 195 countries.
Outcome:	The proposed model reduces misinformation generation across languages and countries . it also reduces the risk of misinformation being spread across countries based on the model's performance .

Copied to clipboard

Andreia Podasca, Anup Das

Agentic Rubrics as Contextual Verifiers for SWE Agents (2026.acl-long)

Challenge:	Existing robustness methods sacrifice clean performance or fail to generalize to higher corruption levels.
Approach:	They propose a mechanism that uses semantic patterns to pull corrupted embeddings toward correct representations by Eigenspectrum Regularization.
Outcome:	The proposed mechanism outperforms robustness methods on 13 GLUE and SuperGLUE tasks while maintaining competitive clean performance.

Copied to clipboard

Mohit Raghavendra, Anisha Gunjal, Bing Liu, Yunzhong He

Challenge:	Large Language Models (LLMs) have rapidly advanced on coding tasks, enabling increasingly capable software engineering agents for real-time code editing and bug fixing.
Approach:	They propose to use a rubric checklist to create a context-grounded rubric for SWE agents.
Outcome:	The proposed rubrics achieve a score of 54.2% on Qwen3-Coder-30B-A3B and 40.6% on Qween3-332B .

Evaluating the Impact of Verbal Multiword Expressions on Machine Translation (2026.acl-long)

Copied to clipboard

Linfeng Liu, Saptarshi Ghosh, Tianyu Jiang

MedVerse: Efficient and Reliable Medical Reasoning via DAG-Structured Parallel Execution (2026.acl-long)

Challenge:	Verbal multiword expressions (VMWEs) are difficult for machine translation because their meanings are often not recoverable from their component words.
Approach:	They analyze the impact of verbal idioms, verb-particle constructions, and light verb constructions on machine translation quality from English to multiple languages.
Outcome:	The proposed system improves translation quality by focusing on verb idioms, verb-particle constructions and light verb constructions.

Copied to clipboard

Jianwen Chen, Xinyu Yang, Peng Xia, Arian Azarang, Yueh Z Lee, Gang Li, Hongtu Zhu, Yun Li, Beidi Chen, Huaxiu Yao

MetFuse: Figurative Fusion between Metonymy and Metaphor (2026.acl-long)

Challenge:	Recent advances in large reasoning models have broadened the capabilities of medical artificial intelligence.
Approach:	They propose a reasoning framework for complex medical inference that reformulates medical reasoning as a parallelizable directed acyclic graph process based on Petri Net theory.
Outcome:	The proposed reasoning framework improves strong general-purpose LLMs by up to 8.9%.

Copied to clipboard

Saptarshi Ghosh, Tianyu Jiang

Stylistic Analysis

Challenge:	Metonymy and metaphor are two fundamental linguistic phenomena in figurative language that involve concept mapping.
Approach:	They propose a framework that transforms a literal sentence into three figurative variants . they propose 'metonymic, metaphoric, and hybrid' datasets that can be used to map metonymy and metaphor .
Outcome:	The proposed framework transforms a literal sentence into three figurative variants . hybrid examples yield the largest gains on metonymy tasks, the study shows .

Adaptive Constraint Propagation: Scaling Structured Inference for Large Language Models via Meta-Reinforcement Learning (2026.acl-long)

Copied to clipboard

Ibne Farabi Shihab, Sanjeda Akter, Anuj Sharma

STReasoner: Empowering LLMs for Spatio-Temporal Reasoning in Time Series via Spatial-Aware Reinforcement Learning (2026.acl-long)

Challenge:	Large language models increasingly require structured inference, says aaron sagar . meta-learning learns universal constraint propagation policies without task-specific training . standard schedulers are inexpensive but myopic, because they optimize local effects .
Approach:	MetaJuLS learns universal constraint propagation policies applicable across languages and tasks without task-specific retraining.
Outcome:	MetaJuLS achieves 1.5-2.0 speedups over GPU-optimized baselines while maintaining accuracy within 0.2% of state-of-the-art parsers.

Copied to clipboard

Juntong Ni, Shiyu Wang, Qi He, Ming Jin, Wei Jin

When Vision-Language Models Judge Without Seeing: Exposing Informativeness Bias (2026.acl-long)

Challenge:	Existing models focus on predictive accuracy over reasoning, a gap exists . time series data are ubiquitous in real-world systems and exhibit complex spatio-temporal structures.
Approach:	They propose a time series reasoning model that integrates time series, graph structure, and text for explicit reasoning.
Outcome:	The proposed model achieves average accuracy gains between 17% and 135% at 0.004x the cost of proprietary models and generalizes robustly to real-world data.

Copied to clipboard

Xiaohan Zou, Roshan Sridhar, Mohammadtaher Safarzadeh, Dan Roth

Hallucination Detection in LLMs with Topological Divergence on Attention Graphs (2026.acl-long)

Challenge:	VLM-as-a-Judges pay limited attention to the image content when making decisions, and often blindly favor the more informative answer, even when they can recognize it conflicts with the image.
Approach:	They propose a judging paradigm that first corrects inconsistencies with the image content in candidate answers, and then compares the answers against this corrected version.
Outcome:	The proposed paradigm reduces informativeness bias by up to 17%, resulting in performance gains of up to 9.8%.

Copied to clipboard

Alexandra Bazarova, Andrei Volodichev, Aleksandr Yugay, Andrey Shulga, Alina Ermilova, Konstantin Polev, Julia Belikova, Rauf Parchiev, Dmitry Simakov, Maxim Savchenko, Andrey Savchenko, Serguei Barannikov, Alexey Zaytsev

Exploring Concreteness Through a Figurative Lens (2026.acl-long)

Challenge:	Large language models (LLMs) are prone to producing so-called hallucinations, i.e., content that is factually or contextually incorrect.
Approach:	They propose a TOpology-based HAllucination detector which quantifies the structural properties of graphs induced by attention matrices.
Outcome:	The proposed detector achieves state-of-the-art or competitive results on several benchmarks while requiring minimal annotated data and computational resources.

Copied to clipboard

Saptarshi Ghosh, Tianyu Jiang

Stylistic Analysis

Challenge:	Static concreteness ratings are widely used in NLP, yet a word’s concreteness can shift with context, especially in figurative language such as metaphor, where common concrete nouns can take abstract interpretations.
Approach:	They conduct a layer-wise and geometric analysis of LLM hidden representations across four model families to examine how models distinguish literal vs. figurative usage.
Outcome:	The results show that LLMs separate literal and figurative usage in early layers and that mid-to-late layers compress concreteness into a one-dimensional direction consistent across models.

Mediocrity is the key for LLM as a Judge Anchor Selection (2026.acl-long)

Copied to clipboard

Shachar Don-Yehiya, Asaf Yehudai, Leshem Choshen, Omri Abend

Challenge:	a poor selection of an anchor can dramatically reduce correlation with human rankings . traditional reference-based metrics are often ill-suited for open-ended generation .
Approach:	They evaluate 22 different anchors on a Arena-Hard-v2.0 dataset and quantify the effect size of anchor selection.
Outcome:	The proposed model is better or worse than all other models, but it is rarely indicative of the relative ranking of the models.

When "Correct" Is Not Safe: Can We Trust Functionally Correct Patches Generated by Code Agents? (2026.acl-long)

Copied to clipboard

Yibo Peng, James Song, Lei Li, Xinyu Yang, Mihai Christodorescu, Ravi Mangal, Corina S. Pasareanu, Haizhong Zheng, Beidi Chen

CoG: Controllable Graph Reasoning via Relational Blueprints and Failure-Aware Refinement over Knowledge Graphs (2026.acl-long)

Challenge:	Code agents are increasingly trusted to autonomously fix bugs on platforms such as GitHub, yet their security evaluation focuses on functional correctness.
Approach:	They propose to attack functionally correct yet vulnerable (FCV) patches by combining multi-turn reasoning with tool invocation and environment interaction.
Outcome:	The proposed FCV-Attack achieves an attack success rate of 40.7% on GPT-5 Mini + OpenHands.

Copied to clipboard

Yuanxiang Liu, Songze Li, Xiaoke Guo, Zhaoyan Gong, Qifei Zhang, Huajun Chen, Wen Zhang

Latent Agents: A Post-Training Procedure for Internalized Multi-Agent Debate (2026.acl-long)

Challenge:	Existing approaches to large language models often exhibit cognitive rigidity, causing reasoning stagnation.
Approach:	They propose a training-free framework that mimics the interplay between intuition and deliberation.
Outcome:	The proposed framework outperforms state-of-the-art approaches on three benchmarks.

Copied to clipboard

John Seon Keun Yi, Aaron Mueller, Dokyun Lee

Act-Adaptive Margin: Dynamically Calibrating Reward Models for Subjective Ambiguity (2026.acl-long)

Challenge:	Multi-agent debate is compute-intensive and requires long transcripts before answering questions.
Approach:	They propose a framework that distills multi-agent debate into a single LLM by combining debate structure learning with internalization via dynamic reward scheduling and length clipping.
Outcome:	The proposed model matches or exceeds explicit multi-agent debate performance using 93% fewer tokens across multiple models and benchmarks.

Copied to clipboard

Feiteng Fang, Dingwei Chen, Xiang Huang, Ting-En Lin, Yuchuan Wu, Xiong Liu, Jing Ye, Ziqiang Liu, Haonan Zhang, Liang Zhu, Hamid Alinejad-Rokny, Min Yang, Yongbin Li

Challenge:	Existing approaches to reward modeling in reinforcement learning tasks are limited when dealing with ambiguous preferences.
Approach:	They propose to use AAM to dynamically calibrate preference margins using the Bradley-Terry model's internal parameter knowledge to improve reward modeling in subjective tasks.
Outcome:	The proposed approach improves reward modeling by dynamically calibrating preference margins using the model’s internal parameter knowledge.

To Think or Not to Think: The Hidden Cost of Meta-Training with Excessive CoT Examples (2026.acl-long)

Copied to clipboard

Vignesh Kothapalli, Ata Fatahibaarzi, Hamed Firooz, Maziar Sanjabi

VIGNETTE: Socially Grounded Bias Evaluation for Vision-Language Models (2026.acl-long)

Challenge:	Chain-of-thought (CoT) prompting and in-context learning (ICL) have unlocked significant reasoning capabilities in large language models (LLMs).
Approach:	They propose a meta-training technique to learn reasoning tasks in-context using CoT examples.
Outcome:	The proposed methods improve performance on novel reasoning tasks even when there are no CoT examples available in-context.

Copied to clipboard

Chahat Raj, Bowen Wei, Aylin Caliskan, Antonios Anastasopoulos, Ziwei Zhu

BiMind: A Dual-Head Reasoning Model with Attention-Geometry Adapter for Incorrect Information Detection (2026.acl-long)

Challenge:	Existing studies on VLM bias focus on portrait-style images and gender-occupation associations . existing studies ignore broader and more complex social stereotypes and their implied harm .
Approach:	They propose a large-scale VQA benchmark for evaluating bias in vision-language models . they use a question-answering framework that spans factuality, perception, stereotyping, and decision making .
Outcome:	The proposed framework examines bias in vision-language models using 30M+ images . findings reveal subtle, multifaceted, and surprising stereotypical patterns .

Copied to clipboard

Zhongxing Zhang, Emily K. Vraga, Jisu Huh, Jaideep Srivastava

Autonomous Knowledge Graph Exploration with Adaptive Breadth-Depth Retrieval (2026.acl-long)

Challenge:	Existing detection approaches struggle to balance content verification with knowledge modification under collapsed attention geometries.
Approach:	They propose a dual-head reasoning framework that disentangles content-internal reasoning from knowledge-augmented reasoning.
Outcome:	The proposed model outperforms detection approaches and provides interpretable diagnostics on when and why knowledge matters.

Copied to clipboard

Joaquin Polonuer, Lucas Vittor, Iñaki Arango, Ayush Noori, David A. Clifton, Luciano Del Corro, Marinka Zitnik

Retrieval Heads are Dynamic (2026.acl-long)

Challenge:	ARK: Adaptive Retriever of Knowledge is a tool-using KG retriever that allows a language model to control breadth-depth tradeoffs without requiring a fragile seed selection or pre-set hop depth.
Approach:	They propose a tool-using KG retriever that gives a language model control over breadth-depth tradeoff using global lexical search over node descriptors and one-hop neighborhood exploration that composes into multi-hop traversal.
Outcome:	The proposed model improves on a teacher's dataset by +7.0, +26.6, and +13.5% while retaining 98.5% of the teacher' s Hit@1 rate.

Copied to clipboard

Yuping Lin, Zitao Li, Yue Xing, Pengfei He, Yingqian Cui, Yaliang Li, Bolin Ding, Jingren Zhou, Jiliang Tang

SynthAgent: Adapting Web Agents with Synthetic Supervision (2026.acl-long)

Challenge:	Recent studies have identified "retrieval heads" in Large Language Models responsible for extracting information from input contexts.
Approach:	They propose to examine retrieval heads from a dynamic perspective . they establish that retrieval head activation is highly dynamic and functionally irreplaceable .
Outcome:	The proposed model's hidden state encodes a predictive signal for future retrieval head patterns, indicating an internal planning mechanism.

Copied to clipboard

Zhaoyang Wang, Yiming Liang, Xuchao Zhang, Qianhui Wu, Siwei Han, Anson Bastos, Rujia Wang, Chetan Bansal, Baolin Peng, Jianfeng Gao, Saravan Rajmohan, Huaxiu Yao

More Than Meets the Eye: Measuring the Semiotic Gap in Vision-Language Models via Semantic Anchorage (2026.acl-long)

Challenge:	Existing studies have focused on synthetic supervision but have encountered data quality issues.
Approach:	They propose a fully synthetic supervision framework that aims at improving data quality via dual refinement of both tasks and trajectories.
Outcome:	The proposed framework outperforms existing methods on standardized benchmarks and shows promising results on a standardized test.

Copied to clipboard

Wei He

LLM Agents in Law: Taxonomy, Applications, and Challenges (2026.acl-long)

Challenge:	Vision-Language Models excel at photorealistic generation, but struggle to represent abstract meanings.
Approach:	They propose a benchmark that replaces high-fidelity visual detail with schematic iconicity by generating paired, sense-anchored visualizations for literal and idiomatic readings.
Outcome:	The proposed benchmark replaces high-fidelity visual detail with schematic iconicity by generating paired, sense-anchored visualizations for literal and idiomatic readings.

Copied to clipboard

Shuang Liu, Ruijia Zhang, Ruoyun Ma, Yujia Deng, Lanyi Zhu, Jiayu Li, Zelong Li, Zhibin Shen, Mengnan Du

Challenge:	Large language models (LLMs) have improved the legal domain, but deployment of standalone models faces significant limitations regarding hallucination, outdated information, and verifiability.
Approach:	They present a survey of LLM agents for legal tasks and analyze their architectures . they analyze the transition from standard legal LLMs to legal agents .
Outcome:	The proposed architectures bridge the gap between technical capabilities and domain-specific needs.

BenchMarker: An Education-Inspired Toolkit for Highlighting Flaws in Multiple-Choice Benchmarks (2026.acl-long)

Copied to clipboard

Nishant Balepur, Bhavya Rajasekaran, Hyunjin Jane Oh, Michael Xie, Atrey Desai, Vipul Gupta, Steven James Moore, Eunsol Choi, Rachel Rudinger, Jordan Lee Boyd-Graber

Challenge:	Multiple-choice question answering (MCQ) is standard in NLP, but benchmarks lack rigorous quality control.
Approach:	They propose an education-inspired toolkit that uses LLM judges to flag flaws in MCQs . they validate the tool with annotations and run it to audit 12 benchmarks based on 19-rule education rubric .
Outcome:	The proposed toolkit flags three common MCQ flaws based on a 19-rule education rubric . contaminated MCqs tend to inflate accuracy, while writing errors lower it and change rankings .

Model Internal Sleuthing: Finding Lexical Identity and Inflectional Features in Modern Language Models (2026.acl-long)

Copied to clipboard

Michael Li, Nishant Subramani

HAT: Hallucination Annotation for Translation (2026.acl-long)

Challenge:	Prior work suggests hierarchical organization where different layers specialize in capturing distinct levels of linguistic structure.
Approach:	They probe 25 models from BERT Base to Qwen2.5-7B focusing on linguistic properties: lexical identity and inflectional features.
Outcome:	The proposed model maintains inflectional features across layers while trading off lexical identity for compact, predictive representations.

Copied to clipboard

Rajen Chatterjee, Xintong Li, Paisarn Charoenpornsawat, Allen Lee

DMRetriever: A Family of Models for Improved Text Retrieval in Disaster Management (2026.acl-long)

Challenge:	Hallucinations in machine translation (MT) outputs are prone to hallucination, authors say . lack of high-quality benchmarks for halluciation detection has hindered MT deployments .
Approach:	They propose a dataset that provides annotated hallucination distributions and benchmarks . they use 350,959 span-level annotations across 38 language pairs to analyze hallucis a MT output .
Outcome:	The proposed dataset provides high-quality benchmarks for hallucination detection in machine translation . the dataset includes 350,959 span-level annotated samples across 38 language pairs .

Copied to clipboard

Kai Yin, Xiangjue Dong, Chengkai Liu, Allen Lin, Lingfeng Shi, Ali Mostafavi, James Caverlee

Language Models Don’t Know What You Want: Evaluating Personalization in Deep Research Needs Real Users (2026.acl-long)

Challenge:	Existing models fail to handle the varied search intents inherent to disaster management scenarios, resulting in inconsistent and unreliable performance.
Approach:	They propose a new series of dense retrieval models tailored for disaster management that train on a three-stage framework with unsupervised contrastive pre-training and difficulty-aware progressive instruction fine-tuning.
Outcome:	The proposed model outperforms baseline models by 13.3 times and 33 times over baselines with only 7.6% of their parameters.

Copied to clipboard

Nishant Balepur, Malachi Hamada, Varsha Kishore, Sergey Feldman, Amanpreet Singh, Pao Siangliulue, Joseph Chee Chang, Eunsol Choi, Jordan Lee Boyd-Graber, Aakanksha Naik

LAD-RAG: Layout-aware Dynamic RAG for Visually-Rich Document Understanding (2026.acl-long)

Challenge:	Earlier research used real users to push personalization, but easy-to-use judges have been criticized for not adopting online studies.
Approach:	They propose a personalized action-following tool that infers a user's research interests and proposes personalized actions for a query.
Outcome:	The proposed tool beats baselines in citation metrics and personalized action-following with an online version of MySQA.

Copied to clipboard

Zhivar Sourati, Zheng Wang, Marianne Menglin Liu, Yazhe Hu, Mengqing Guo, Sujeeth Bharadwaj, Kyu J. Han, Tao Sheng, Sujith Ravi, Morteza Dehghani, Dan Roth

Among Us: Measuring and Mitigating Malicious Contributions in Model Collaboration Systems (2026.acl-long)

Challenge:	Conventional retrieval-augmented generation (RAG) methods encode content in isolated chunks during ingestion, losing structural and cross-page dependencies, and retrieve a fixed number of pages at inference.
Approach:	They propose a Layout-Aware Dynamic RAG framework that encodes content in isolated chunks during ingestion and retrieves a fixed number of pages at inference.
Outcome:	Experiments on MMLongBench-Doc, LongDocURL, DUDE, and MP-DoxVQA show that LAD-RAG improves retrieval, achieving over 90% perfect recall on average without any top-k tuning, and outperforming baseline retrievers by up to 20% in recall at comparable noise levels.

Copied to clipboard

Ziyuan Yang, Wenxuan Ding, Shangbin Feng, Yulia Tsvetkov

Not All Tokens Matter: Towards Efficient LLM Reasoning via Token Significance in Reinforcement Learning (2026.acl-long)

Challenge:	Existing research is leveraging multiple language models with diverse skills and strengths to collaborate.
Approach:	They propose mitigation strategies to mitigate the impact of malicious models by employing external supervisors to disable/mask them out to reduce their influence.
Outcome:	The proposed mitigation strategies recover 95.31% of initial performance while making model collaboration systems fully resistant to malicious models remains an open question.

Copied to clipboard

Hanbing Liu, Lang Cao, Yuanyi Ren, Mengyu Zhou, Haoyu Dong, Xiaojun Ma, Shi Han, Dongmei Zhang

Query-Efficient Agentic Graph Extraction Attacks on GraphRAG Systems (2026.acl-long)

Challenge:	Large language models (LLMs) often produce unnecessarily long explanations that reduce efficiency.
Approach:	They propose a length-aware reward that selectively penalizes insignificance tokens . they also propose 'dynamic length control' that encourages more detailed reasoning .
Outcome:	The proposed method reduces response length while maintaining correctness, the authors show . it selectively penalizes insignificance tokens while maintaining accuracy .

Copied to clipboard

Shuhua Yang, Jiahao Zhang, Yilong Wang, Dongwon Lee, Suhang Wang

Function Words as Statistical Cues for Language Learning (2026.acl-long)

Challenge:	Existing attacks exploit leakage of retrieved subgraphs, leaving the security implications of structured knowledge representations unexplored.
Approach:	They propose a framework that leverages a novelty-guided exploration–exploitation strategy and external graph memory modules to extract a latent entity–relation graph.
Outcome:	The proposed framework outperforms baselines on medical, agriculture, and literary datasets under identical query budgets while maintaining high precision.

Copied to clipboard

Xiulin Yang, Heidi R. Getz, Ethan Gotlieb Wilcox

Challenge:	Existing studies have argued that function words aid learning abstract grammatical knowledge from linear input.
Approach:	They examine the statistical distribution of function words and their properties . they show that function words are reliable, diverse, and informative .
Outcome:	The results show that function words preserve high frequency, reliable syntactic association, phrase-boundary alignment and are informative to structural dependency.

Evaluating Robustness of Large Language Models Against Multilingual Typographical Errors (2026.acl-long)

Copied to clipboard

Raoyuan Zhao, Yihong Liu, Lena Altinger, Hinrich Schuetze, Michael A. Hedderich

Listening Like Humans: Semantics-Guided Noise-Robust Multimodal Speech Recognition (2026.acl-long)

Challenge:	Large language models (LLMs) are increasingly deployed in multilingual, real-world applications where user inputs introduce typographical errors.
Approach:	They propose a multilingual typo generation algorithm that simulates human-like errors based on language-specific keyboard layouts and typing behavior.
Outcome:	The proposed model can generate the correct answer ("500") under typos in English, German, and Russian.

Copied to clipboard

Yan Fang, Jun Chen, Yian Yao, Shuxin Zhong, Min Sun, Kaishun Wu

Measuring What Matters!! Assessing Therapeutic Principles in Mental-Health Conversation (2026.acl-long)

Challenge:	Severe acoustic degradation results in unreliable ASR outputs . et al., 2024b): critical concerns regarding reliability and fairness of ASR .
Approach:	They propose a multimodal framework that reframes ASR as semantics-guided speech reconstruction.
Outcome:	The proposed framework achieves an average reduction in WER while also attaining 98.71% BERTScore and 96.7% USE over advanced baselines.

Copied to clipboard

Abdullah Mazhar, Het Riteshkumar Shah, Aseem Srivastava, Smriti Joshi, Md Shad Akhtar

ParaCodex: A Profiling-Guided Autonomous Coding Agent for Reliable Parallel Code Generation and Translation (2026.acl-long)

Challenge:	Recent systems exhibit conversational competence but lack structured mechanisms to evaluate adherence to core therapeutic principles.
Approach:	They propose a framework to evaluate therapist-like responses for clinically grounded appropriateness and effectiveness using an ordinal scale.
Outcome:	The proposed framework achieves an F-1 score of 63.34 versus the baseline Qwen3 score of 38.56 .

Copied to clipboard

Erel Kaplan, Tomer Bitan, Lian Ghrayeb, Le Chen, Tom Yotam, Niranjan Hasabnis, Gal Oren

CoLA: A Choice Leakage Attack Framework to Expose Privacy Risks in Subset Training (2026.acl-long)

Challenge:	Parallel programming is central to modern highperformance computing, but producing parallel implementations that are correct and fast remains arduous.
Approach:	They propose a workflow that turns a Codex-based agent into an autonomous OpenMP GPU offload system . they use staged hotspot analysis, explicit data planning, correctness gating, profiling-guided refinement .
Outcome:	The proposed system outperforms a Codex-based agent on HeCBench, Rodinia, and NAS.

Copied to clipboard

Qi Li, Cheng-Long Wang, Yinzhi Cao, Di Wang

MediEval: A Unified Medical Benchmark for Patient-Contextual and Knowledge-Grounded Reasoning in LLMs (2026.acl-long)

Challenge:	Existing threat models underestimate subset-training privacy risks because of the scale of modern datasets.
Approach:	They propose a unified framework for analyzing privacy leakage in subset selection based on side-channel metadata from the subset process or via the outputs of the target model.
Outcome:	The proposed framework analyzes privacy leakage in subset selection based on two different scenarios .

Copied to clipboard

Zhan Qu, Michael Färber

The Path Not Taken: Duality in Reasoning about Program Execution (2026.acl-long)

Challenge:	Existing evaluations test factual medical knowledge in isolation or assess patient-level reasoning without verifying correctness, leaving a critical gap.
Approach:	They propose a benchmark that links MIMIC-IV EHRs to a unified knowledge base built from UMLS and other biomedical vocabularies.
Outcome:	The proposed model improves by +16.4 macro-F1 points over the base model and eliminates truth inversion errors.

Copied to clipboard

Eshgin Hasanov, Md. Mahadi Hassan, Santu Karmaker, Aashish Yadavally

Accommodation and Epistemic Vigilance: A Pragmatic Account of Why LLMs Fail to Challenge Harmful Beliefs (2026.acl-long)

Challenge:	Existing benchmarks focus on predicting program properties tied to specific inputs, resulting in a narrow view of dynamic code reasoning and data contamination.
Approach:	They instantiate dual-path reasoning in a benchmark and evaluate 13 LLMs.
Outcome:	The proposed model provides a robust and discriminative proxy for dynamic code understanding.

Copied to clipboard

Myra Cheng, Robert D. Hawkins, Dan Jurafsky

Every Response Counts: Quantifying Uncertainty of LLM-based Multi-Agent Systems through Tensor Decomposition (2026.acl-long)

Challenge:	Recent studies show that large language models fail to challenge users’ harmful beliefs in domains ranging from medical advice to social reasoning.
Approach:	They propose to examine whether pragmatic factors influence LLM accommodation and epistemic vigilance in humans.
Outcome:	The proposed model can be understood and addressed as having excessive accommodation and insufficient epistemic vigilance.

Copied to clipboard

Tiejin Chen, Huaiyuan Yao, Jia Chen, Evangelos E. Papalexakis, Hua Wei

Uncertainty Quantification in LLM Agents: Foundations, Emerging Challenges, and Opportunities (2026.acl-long)

Challenge:	Existing methods for MAS fail to address the unique complexities of multi-step reasoning . Existing uncertainty quantification methods struggle with cascading uncertainty .
Approach:	They propose a framework that quantifies uncertainty through tensor decomposition . they show that MATU effectively estimates holistic and robust uncertainty .
Outcome:	The proposed framework disentangles and quantifies distinct sources of uncertainty . it is generalizable across different agent structures and can be used for scientific discovery, education, healthcare and transportation.

Copied to clipboard

Changdae Oh, Seongheon Park, To Eun Kim, Jiatong Li, Wendi Li, Samuel Yeh, Sean Du, Hamed Hassani, Paul Bogdan, Dawn Song, Sharon Li

Challenge:	Uncertainty quantification (UQ) for large language models is a key building block for daily applications.
Approach:	They propose a general formulation of agent UQ that subsumes broad classes of existing UQ setups.
Outcome:	The proposed framework is based on the first general formulation of agent UQ that subsumes broad classes of existing setups.

Guided by Gut: Efficient Test-Time Scaling with Reinforced Intrinsic Confidence (2026.acl-long)

Copied to clipboard

Amirhosein Ghasemabadi, Keith G. Mills, Baochun Li, Di Niu

Knowing When Not to Answer: Lightweight KB-Aligned OOD Detection for Safe RAG (2026.acl-long)

Challenge:	Guided by Gut (GG) is an efficient self-guided TTS framework for Large Language Models (LLMs) that performs step-by-step reasoning at a low cost without any reward models or verifiers.
Approach:	They propose a self-guided TTS framework that enables LLMs to perform step-by-step reasoning at a low cost without any reward models or verifiers.
Outcome:	Empirical evaluations show that GG performs better than TTS with PRMs while reducing GPU memory usage by up to 10.

Copied to clipboard

Ilias Triantafyllopoulos, Renyi Qu, Salvatore Giorgi, Brenda Curtis, Lyle Ungar, João Sedoc

Guidelines as Environments: A World Model Approach to Rule Following (2026.acl-long)

Challenge:	Retrieval-Augmented Generation (RAG) systems are increasingly deployed in high-stakes domains where safety depends on how a system answers . out-of-domain (OOD) queries can impair performance and safety .
Approach:	They propose to use lightweight, KB-aligned OOD detection as an always-on gate for RAG systems.
Outcome:	The proposed method scores queries in a compact subspace selected either by explained-variance retention (EVR) or by a separability-driven -test ranking.

Copied to clipboard

Haiqing Li, Wenliang Zhong, Yinhao Wu, Hehuan Ma, Yuzhi Guo, Thao M. Dang, Junzhou Huang

Cross-Modal Taxonomic Generalization in (Vision-) Language Models (2026.acl-long)

Challenge:	Existing models for guideline-following are a poor fit for ambiguous, text-defined constraints.
Approach:	They propose a Rule-Grounded Causal World Model that builds an explicit state space from guideline text itself.
Outcome:	Experiments show that the proposed model can be used to model rule execution with an explicit state space from the guideline text itself.

Copied to clipboard

Tianyang Xu, Marcelo Sandoval-Castañeda, Karen Livescu, Greg Shakhnarovich, Kanishka Misra

Biomedical Question Answering via Multi-Level Summarization on a Local Knowledge Graph (2026.acl-long)

Challenge:	Existing studies have shown that language models learn from surface form to learn from more grounded evidence.
Approach:	They propose to use a vision-language model to learn hypernyms from images . they find that the model can recover this knowledge and generalize even when there is no hypernomia in the image.
Outcome:	The proposed model can recover this knowledge and generalize even when the model receives no evidence of hypernyms during training.

Copied to clipboard

Lingxiao Guan, Yuanhao Huang, Jie Liu

Sigmoid Head for Quality Estimation under Language Ambiguity (2026.acl-long)

Challenge:	Existing work on how to effectively capture multi-document relationships remains an open question . Existing techniques to mitigate this problem include hierarchical summarization of semantically related chunks or integrating Knowledge Graphs (KGs).
Approach:	They propose a method which constructs a local knowledge graph from retrieved documents . they use propositional claims to construct a knowledge graph and contextualize a small language model .
Outcome:	The proposed method outperforms RAG on biomedical benchmarks and is generalizable and effective.

Copied to clipboard

Tu Anh Dinh, Jan Niehues

RExBench: Can coding agents autonomously implement AI research extensions? (2026.acl-long)

Challenge:	Language model (LM) probability is not reliable quality estimator, as natural language is ambiguous.
Approach:	They propose to train a language model (LM) probability module on top of pre-trained LMs to address these limitations.
Outcome:	The proposed module is an extra unembedding head with sigmoid activation to tackle the first limitation.

Copied to clipboard

Nicholas Edwards, Yukyung Lee, Yujun Audrey Mao, Yulu Qin, Sebastian Schuster, Najoung Kim

Challenge:	Existing large language model (LLM) agents are not capable of performing research extension tasks autonomously.
Approach:	They propose a benchmark to evaluate LLM agents' ability to extend existing AI research . they use extensions of 12 recently published research papers accompanied by domain expert-written instructions .
Outcome:	The proposed benchmark evaluates 12 LLM agents implemented using aider and OpenHands.

Narrative License and Model Sycophancy in LLM Summaries of Scientific Work (2026.acl-long)

Copied to clipboard

Calvin Isch, Grace Jennings

Summarization

Challenge:	Large language models (LLMs) are increasingly used to summarize academic work . however, they can exaggerate or mischaracterize findings .
Approach:	They examine how Narrative License (NL) emerges in large language models summaries . authors find that stated stances and user personas produce predictable shifts .
Outcome:	The proposed models can exaggerate or mischaracterize findings in scholarly articles . the authors show that the models' "sycophancy" can reduce NL .

AlignUSER: Human-Aligned LLM Agents via World Models for Recommender System Evaluation (2026.acl-long)

Copied to clipboard

Nicolas Bougie, Gian Maria Marconi, Xiaotong Ye, Narimawa Watanabe

Domain Generalizable AI Guardrails with Augmented Policy Training (2026.acl-long)

Challenge:	Existing evaluation practices for recommender systems rely on few-shot prompting and offline metrics are often misaligned with online behavior.
Approach:	They propose a framework that learns world-model-driven agents from human interactions.
Outcome:	The proposed framework enables agents to express rich preferences and feedback in natural language and interact with recommender systems in a simulation.

Copied to clipboard

Minqian Liu, Ioana Baldini, David Rabinowitz, David S Rosenberg, Sebastian Gehrmann, Mark Dredze

Challenge:	Current guardrails overfit the training policies, preventing adaptation to new domains and policies.
Approach:	They propose a training recipe that uses a suite of policy perturbation strategies to reduce overfitting and increase generalization to guardrails.
Outcome:	The proposed training recipe reduces overfitting and increases generalization on unseen policies and achieves comparable or better performance than existing 8B guardrails on unsen policies.

APEX-MEM: Agentic Semi-Structured Memory with Temporal Reasoning for Long-Term Conversational AI (2026.acl-long)

Copied to clipboard

Pratyay Banerjee, Masud Moshtaghi, Shivashankar Subramanian, Amita Misra, Ankit Chadha

emg2speech: synthesizing speech from electromyography using self-supervised speech models (2026.acl-long)

Challenge:	Large language models struggle with reliable long-term conversational memory . enlarging context windows or applying nave retrieval often introduces noise .
Approach:	They propose a conversational memory system that uses domain-agnostic ontology to structure conversations as temporally grounded events in an entity-centric framework.
Outcome:	APEX-MEM outperforms state-of-the-art retrieval methods in accuracy and time resolution.

Copied to clipboard

Harshavardhana T Gowda, Daniel C Comstock, Lee M. Miller

CoAct: Co-Active LLM Preference Learning with Human-AI Synergy (2026.acl-long)

Challenge:	a neuromuscular speech interface translates electromyographic (EMG) signals recorded from orofacial muscles during speech articulation directly into audio.
Approach:	They propose a neuromuscular speech interface that translates electromyographic (EMG) signals recorded from orofacial muscles during speech articulation directly into audio.
Outcome:	The proposed system synthesizes speech without explicit modeling or vocoder training.

Copied to clipboard

Ruiyao Xu, Mihir Parmar, Tiankai Yang, Zhengyu Hu, Yue Zhao, Kaize Ding

Label and Explanation Variation in LLM-Based Annotation: a Case Study in Natural Language Inference (2026.acl-long)

Challenge:	Existing methods to learn from preference-based feedback are expensive and scarce.
Approach:	They propose a framework that synergistically combines self-rewarding and active learning through human-AI collaboration.
Outcome:	The proposed framework outperforms existing methods on three reasoning benchmarks and achieves average improvements of +13.25% on GSM8K, +8.19% on MATH, and +13.16% on WebInstruct.

Copied to clipboard

Artur Kulmizev, Erika Lombart, Patrick Watrin, Marie-Catherine de Marneffe

Text-Attributed Knowledge Graph Enrichment with Large Language Models for Medical Concept Representation (2026.acl-long)

Challenge:	Large language models (LLMs) have shown considerable promise for annotation purposes, but questions remain about their ability to capture human label variation (HLV) label variation is genuine disagreement between annotators observed across NLP tasks.
Approach:	They investigate how label and explanation variation manifests within and across LLMs with respect to the Natural Language Inference task.
Outcome:	The proposed models generate label distributions similar to humans but exhibit distinct, idiosyncratic judgments and disagreement patterns.

Copied to clipboard

Mohsen Nayebi Kerdabadi, Arya Hadizadeh Moghaddam, Chen Chen, Dongjie Wang, Zijun Yao

ViDoRe V3: A Comprehensive Evaluation of Retrieval Augmented Generation in Complex Real-World Scenarios (2026.acl-long)

Challenge:	eHRs encode a patient's medical history as a high-dimensional and sparse sequence of diagnosis, medication, and procedure concepts . robust concept representation learning is hindered by key challenges, authors say . clinically important cross-type dependencies are often missing or incomplete in existing ontology resources .
Approach:	They propose a graph learning framework that integrates semantics with medical concepts to improve prediction performance.
Outcome:	The proposed framework improves prediction performance and integrates semantics with graph structure.

Evolving Agents (2026.acl-long)

Copied to clipboard

Leonardo Ranaldi

Challenge:	Current models are static entities incapable of compressing complexity of real world into generalisable concepts . authors: lack of endogenous mechanism for representation updating renders models vulnerable to domain mismatch and catastrophic forgetting .
Approach:	a meta-control system distils on-the-fly abstract representations of states, actions, goals . authors propose a paradigm for autonomous learning driven by pseudo-symbolic abstraction .
Outcome:	a meta-control system distils on-the-fly abstract representations of states, actions, goals . a novel approach resolves the domain mismatch problem and lays the groundwork for truly autonomous AI models .

Copied to clipboard

António Loison, Quentin Macé, Antoine Edy, Victor Xing, Tom Balough, Gabriel de Souza P. Moreira, Bo Liu, Manuel Faysse, Celine Hudelot, Gautier Viaud

Why LLM Safety Guardrails Collapse After Fine-tuning: A Similarity Analysis Between Alignment and Fine-tuning Datasets (2026.acl-long)

Challenge:	Existing benchmarks focus on textual data, single-document comprehension, or evaluating retrieval and generation in isolation.
Approach:	They propose a multimodal RAG benchmark featuring multi-type queries over visually rich document corpora.
Outcome:	The proposed benchmark outperforms existing benchmarks in visual retrieval and human-verified queries.

Copied to clipboard

Lei Hsiung, Tianyu Pang, Yung-Chen Tang, Linyue Song, Tsung-Yi Ho, Pin-Yu Chen, Yaoqing Yang

TabEmb: Joint Semantic-Structure Embedding for Table Annotation (2026.acl-long)

Challenge:	Existing mitigation strategies focus on reactively addressing jailbreak incidents after safety guardrails have been compromised.
Approach:	They investigate the degradation of safety guardrails through the lens of representation similarity between upstream alignment datasets and downstream fine-tuning tasks.
Outcome:	The proposed model reduces harmfulness score by 10.33% when compared to baseline models.

Copied to clipboard

Ehsan Hoseinzade, Ke Wang, Anandharaju Durai Raju

Text-to-Distribution Prediction with Quantile Tokens and Neighbor Context (2026.acl-long)

Challenge:	Existing tables learn by linearizing the 2D table into a 1D token sequence and encoding it with pretrained language models (PLMs) such as BERT, but this leads to limited semantic quality and weaker generalization to unseen or rare values compared to modern LLMs.
Approach:	They propose a table annotation module called TabEmb which decouples semantic encoding from structural modeling by creating semantically rich embeddings for each column.
Outcome:	Experiments show that TabEmb outperforms baselines on different table annotation tasks.

Copied to clipboard

Yilun Zhu, Yuan Zhuang, Nikhita Vedula, Dushyanta Dhyani, Shaoyuan Xu, Mohsen Bayati, Bryan Wang, Shervin Malmasi

Challenge:	Existing methods for text regression lack local grounding and rely on shared representations.
Approach:	They propose a distributional regression model with quantile tokens that insert dedicated quantiles into the input sequence.
Outcome:	The proposed method outperforms baseline models on the inside Airbnb and StackSample datasets.

Check Your Work: Structured Checklist Feedback for Improving Large Language Models (2026.acl-long)

Copied to clipboard

Jonathan Cook, Tim Rocktäschel, Jakob Nicolaus Foerster, Dennis Aumiller, Alex Wang

LOKA: Conflict-Aware LLM Knowledge Update with Adaptive Knowledge Memory (2026.acl-long)

Challenge:	Recent advances in Large Language Models have been driven by verifiable feedback in deterministic domains like mathematics and code.
Approach:	They propose to decompose granular, prompt-specific checklists into a scalar reward and use them to transform them into skalar rewards.
Outcome:	The proposed approach yields an 11.8% win-rate improvement on AlpacaEval 2.0 using Qwen3-8B, outperforming holistic reward models and existing checklist baselines.

Copied to clipboard

Binchi Zhang, Zhengzhang Chen, Zaiyi Zheng, Jundong Li, Haifeng Chen

Student Guides Teacher: Weak-to-Strong Inference via Spectral Orthogonal Exploration (2026.acl-long)

Challenge:	Existing approaches that tackle unlearning and learning separately encounter task conflicts and knowledge management issues when applied to comprehensive knowledge updates.
Approach:	They propose a conflict-aware framework for Large language mOdel Knowledge updAtes that integrates updated knowledge across multiple memory units during training and integrates it with original LLM.
Outcome:	The proposed framework is based on theoretical analysis and empirical evidence and validates the proposed framework with empirical and theoretical evidence.

Copied to clipboard

Dayu Wang, Jiaye Yang, Weikang Li, Jiahui Liang, Yang Li, Deguo Xia, Jizhou Huang

When LLMs Read Tables Carelessly: Measuring and Reducing Data Referencing Errors (2026.acl-long)

Challenge:	Existing Large Language Models suffer from "Reasoning Collapse" on mathematical reasoning tasks where stochastic sampling produces lexical variations of the same erroneous logic rather than genuine semantic exploration.
Approach:	They propose a geometric inference framework that uses a spectral orthogonal probe to introduce semantically heterogeneous reasoning signals into the teacher's orthogonale complement of its dominant subspace.
Outcome:	The proposed framework improves accuracy and sampling efficiency over baseline methods on logic and code generation benchmarks.

Copied to clipboard

Yuqing Yang, Qi Zhu, Zhen Han, Boran Han, Zhengyuan Shen, Shuai Wang, Vassilis N. Ioannidis, Huzefa Rangwala

PubMed Reasoner: Dynamic Reasoning-based Retrieval for Evidence-Grounded Biomedical Question Answering (2026.acl-long)

Challenge:	Large language models (LLMs) perform well on table tasks, but they still make data referencing errors (DREs) prior studies have only offered limited, small-scale analyses.
Approach:	They propose inference-time strategies and lightweight critics to mitigate data referencing errors.
Outcome:	The proposed model achieves an average F1 score of 78.2% in detecting both in-distribution and out-of-difference DREs and assists inference for larger models.

Copied to clipboard

Yiqing Zhang, Xiaozhong Liu, Fabricio Murai

Decoupling Task-Solving and Output Formatting in LLM Generation (2026.acl-long)

Challenge:	Existing approaches to QA provide inaccurate answers but lack mechanisms to iteratively refine poor queries.
Approach:	They propose a biomedical question answering agent that performs self-critic query refinement . they propose re-reflection methods that kick in only after full retrieval is completed .
Outcome:	a biomedical question answering agent achieves 78.32% accuracy on PubMedQA . the proposed approach provides practical assistance to clinicians and biomedically researchers .

Copied to clipboard

Haikang Deng, Po-Nien Kung, Nanyun Peng

GS-Quant: Granular Semantic and Generative Structural Quantization for Knowledge Graph Completion (2026.acl-long)

Challenge:	Recent studies suggest intertwining task and format instructions with strict formatting requirements can negatively impact LLMs' reasoning capabilities.
Approach:	They propose a decoding framework that explicitly decouples format adherence from problem solving.
Outcome:	Experiments show that Deco-G consistently gains over prompting and structured generation baselines, with guaranteed format compliance.

Copied to clipboard

Qizhuo Xie, Yunhui Liu, Yu Xing, Qianzi Hou, Xudong Jin, Tao Zheng, Tieke He

Mind the Gap in Cultural Alignment: Task-Aware Culture Management for Large Language Models (2026.acl-long)

Challenge:	Existing quantization-based approaches to knowledge Graph Completion (KGC) are incomplete.
Approach:	They propose a framework that generates semantically coherent discrete codes for KG entities . they introduce a Granular Semantic Enhancement module that injects hierarchical knowledge into the codebook .
Outcome:	The proposed framework outperforms existing text-based and embedding-based baselines in the KGC domain.

Copied to clipboard

Binchi Zhang, Xujiang Zhao, Jundong Li, Haifeng Chen, Zhengzhang Chen

Analyzing and Internalizing Complex Policy Documents for LLM Agents (2026.acl-long)

Challenge:	Existing cultural alignment approaches fail to align LLMs’ broad cultural values with the specific goals of downstream tasks and suffer from cross-cultural interference.
Approach:	They propose a novel pipeline for task-specific cultural alignment that synthesizes task-aware cultural data in line with target task formats.
Outcome:	Experiments across five national cultures and ten culture-sensitive tasks show consistent improvements over prompt-based and fine-tuning baselines.

Copied to clipboard

Jiateng Liu, Zhenhailong Wang, Xiaojiang Huang, Yingjie Li, Xiang Li, Chenlei Guo, Xing Fan, Ruhi Sarikaya, Heng Ji

In-Context Representation Hijacking (2026.acl-long)

Challenge:	Large language model agents rely on in-context policy documents to act as effective user assistants.
Approach:	They propose an agentic benchmark generator with Controllable Complexity in agent policy across four levels to evaluate agents under increasing complexity.
Outcome:	The proposed method outperforms the baseline in data-sparse and high-complexity settings.

Copied to clipboard

Itay Yona, Amir Sarid, Michael Karasik, Yossi Gandelsman

Quantifying Metric and Model Agreement in Bias Evaluation of Large Language Models (2026.acl-long)

Challenge:	In-context representation hijacking attacks on large language models are insufficient to prevent harm, yet the mechanisms underlying this behavior remain unclear.
Approach:	They propose a simple in-context representation hijacking attack that replaces a harmful keyword with a benign token, and embeds the harmful semantics under a euphemism.
Outcome:	The proposed attack is optimized for open-source and achieves strong success rates on closed-source systems, reaching 74% on Llama-3.3-70B-Instruct with a single-sentence context override.

Copied to clipboard

Arash Asgari, Huan Wu, Amirreza Naziri, Mojtaba Kolahdouzi, Laleh Seyyed-Kalantari

MultiFinBen: Benchmarking Large Language Models for Multilingual and Multimodal Financial Application (2026.acl-long)

Challenge:	a systematic way to measure agreement across bias metrics and models is lacking . a lack of agreement between metrics and model results may be a problem .
Approach:	They introduce Metric Agreement Score and Model Agreement Score to measure agreement across bias metrics and models.
Outcome:	The proposed measures show that metrics within the same category behave independently of each other.

Copied to clipboard

Xueqing Peng, Lingfei Qian, Yan Wang, Ruoyu Xiang, Yueru He, Yang Ren, Mingyang Jiang, Vincent Jim Zhang, Yuqing Guo, Jeff Zhao, Huan He, Yi Han, Yun Feng, Yuechen Jiang, Yupeng Cao, Haohang Li, Yangyang Yu, Xiaoyu Wang, Penglei Gao, Shengyuan Lin, Keyi Wang, Shanshan Yang, Yilun Zhao, Zhiwei Liu, Peng Lu, Jerry Huang, Suyuchen Wang, Triantafillos Papadopoulos, Polydoros Giannouris, Efstathia Soufleri, Nuo Chen, Zhiyang Deng, Heming Fu, Yijia Zhao, Mingquan Lin, Meikang Qiu, Kaleb E Smith, Arman Cohan, Xiao-Yang Liu, Jimin Huang, Guojun Xiong, Alejandro Lopez-Lira, Xi Chen, Junichi Tsujii, Jian-Yun Nie, Sophia Ananiadou, Qianqian Xie

Bidirectional LMs are Better Knowledge Memorizers? A Benchmark for Real-world Knowledge Injection (2026.acl-long)

Challenge:	Existing evaluations of LLMs in finance are text-only, monolingual, and largely saturated by current models.
Approach:	They propose a multilingual and multimodal benchmark for evaluating LLMs in real financial contexts.
Outcome:	The first expert-annotated multilingual and multimodal benchmark is released . it evaluates 21 leading LLMs and shows they perform better in multilingual settings .

Copied to clipboard

Yuwei Zhang, Wenhao Yu, Shangbin Feng, Yifan Zhu, Letian Peng, Jayanth Srinivasa, Gaowen Liu, Jingbo Shang

Mapping the Circumplex of Affect: Geometric Analysis of Emotion Representations via Hyperspherical Contrastive Learning (2026.acl-long)

Challenge:	Existing knowledge injection benchmarks for large language models lack standardized testing grounds.
Approach:	They propose a knowledge injection benchmark that leverages recently-added and expert-curated facts from Wikipedia’s “Did You Know...” entries.
Outcome:	The proposed framework improves reliability accuracy by 29.1%.

Copied to clipboard

Yusuke Yamauchi, Akiko Aizawa

Do LLMs Really Need 10+ Thoughts for “Find the Time 1000 Days Later”? Towards Structural Understanding of LLM Overthinking (2026.acl-long)

Challenge:	Existing methods to induce circular emotion representations in language models are limited . elucidates trade-offs involved in applying circumplex models to deep learning architectures .
Approach:	They propose a method to induce circular emotion representations within language models via contrastive learning on a hypersphere.
Outcome:	The proposed method underperforms in high-dimensional settings and fine-grained classification.

Copied to clipboard

Xinliang Frederick Zhang, Anhad Mohananey, Alexandra Chronopoulou, Pinelopi Papalampidi, Somit Gupta, Tsendsuren Munkhdalai, Lu Wang, Shyam Upadhyay

Anchor: Branch-Point Data Generation for GUI Agents (2026.acl-long)

Challenge:	Existing studies on LLMs' thought processes are limited to superficial, profiling-based observations, failing to delve into their inner workings.
Approach:	They propose a utility-based definition of overthinking that moves beyond length-based metrics and provides a more insightful understanding of LLMs' thought progression.
Outcome:	The proposed model decomposes the LLM thought process into minimally complete sub-thoughts and identifies common thinking patterns for topically similar queries.

Copied to clipboard

Jinbiao Wei, Yilun Zhao, Kangqi Ni, Arman Cohan

Challenge:	Existing GUI agents for real desktop environments require large amounts of high-quality interaction data, but collecting human demonstrations is expensive.
Approach:	They propose a framework that bootstraps scalable desktop supervision from seed demonstrations.
Outcome:	Experiments on standard desktop benchmarks show that the framework improves on zero-shot agents and representative synthesis baselines.

When One LLM Drools, Multi-LLM Collaboration Rules (2026.acl-long)

Copied to clipboard

Shangbin Feng, Wenxuan Ding, Alisa Liu, Zifeng Wang, Weijia Shi, Yike Wang, Shannon Zejiang Shen, Xiaochuang Han, Hunter Lang, Chen-Yu Lee, Tomas Pfister, Yejin Choi, Yulia Tsvetkov

Challenge:	a single general-purpose LLM is not enough to produce a reliable output, argues this paper . a multi-LLM collaboration approach addresses reliability, democratization, and pluralism .
Approach:	They argue that a single general-purpose LLM is not enough to produce a reliable output . they organize existing multi-LLM collaboration methods into a hierarchy based on access and information exchange .
Outcome:	The proposed method addresses reliability, democratization, and pluralism challenges a single LLM fails to produce a reliable output.

From Query to Counsel: Structured Reasoning with a Multi-Agent Framework and Dataset for Legal Consultation (2026.acl-long)

Copied to clipboard

Mingfei Lu, Yi Zhang, Mengjia Wu, Yue Feng

Putting Captions to the Test: Evaluating Video Caption Quality through Multiple-Choice Question Answering (2026.acl-long)

Challenge:	Legal consultation question answering presents unique challenges compared to traditional legal QA tasks .
Approach:	They propose a framework that converts queries into a legal element graph . jurisMA supports dynamic routing, statutory grounding, and stylistic optimization .
Outcome:	The proposed framework outperforms general-purpose and legal-domain LLMs across multiple lexical and semantic metrics.

Copied to clipboard

Zizhen Wang, Bo Feng, Zhengfeng Lai, Shiyu Li, Yang Lu, Meng Cao, Ping Huang, Xiaoming Simon Wang

Subject-level Inference for Realistic Text Anonymization Evaluation (2026.acl-long)

Challenge:	Existing metrics for video captioning are based on text-based comparisons with ground-truth references.
Approach:	They propose a reference-free benchmark that assesses video captions based on their utility . they will release the benchmark to facilitate reproducible research .
Outcome:	The proposed benchmark improves on human-verified, fine-grained questions . it correlates significantly better with human judgments than existing metrics .

Copied to clipboard

Myeong Seok Oh, Dong-Yun Kim, Hanseok Oh, Chaean Kang, Joeun Kang, Xiaonan Wang, Hyunjung Park, Young Cheol Jung, Hansaem Kim

SceneAlign: Aligning Multimodal Reasoning to Scene Graphs in Complex Visual Scenes (2026.acl-long)

Challenge:	Existing text anonymization evaluations assume only a single data subject, ignoring multi-subject scenarios.
Approach:	They propose a benchmark that shifts the unit of evaluation from text spans to individuals . they show that subject-level inference protection drops as low as 33% when masked .
Outcome:	The proposed benchmark reduces the amount of protection available when PII spans are masked.

Copied to clipboard

Chuhan Wang, Xintong Li, Jennifer Yuntong Zhang, Junda Wu, Chengkai Huang, Lina Yao, Julian McAuley, Jingbo Shang

Select Before Use: On the Importance of Reference Model Selection in Preference Alignment (2026.acl-long)

Challenge:	Existing preference-based approaches fail to address this challenge by exploiting language priors to bypass visual grounding.
Approach:	They propose a framework that leverages scene graphs as structured visual information to perform controllable structural interventions.
Outcome:	The proposed framework improves answer accuracy and reasoning faithfulness across seven visual reasoning benchmarks.

Copied to clipboard

Muyang Li, Runze Wu, Xiangyu Zhao, Bo Han, Daoyi Dong, Tongliang Liu

Challenge:	Supervised Fine-Tuning (SFT) is used as the initialization and reference model for subsequent preference alignment.
Approach:	They propose to use RewardRank to estimate initial implicit alignment between reference model and preference objective to ensure LLMs generate safe, helpful, and instruction-aligned content.
Outcome:	Empirical evidence shows that using the selected model as reference can gain up to 67.6% relative increase on length-controlled win rate compared to baselines.

Lost in Translation, and Found: Detecting and Interpreting Translation Effects (2026.acl-long)

Copied to clipboard

Shira Wein, Anna Serbina, Jiyuan Ji, Nathan Wolf, Jason DeGraaff, Prajakta Kini, Maria Leonor Pacheco

From Isolation to Entanglement: When Do Interpretability Methods Identify and Disentangle Known Concepts? (2026.acl-long)

Challenge:	Translationese refers to the statistical patterns that distinguish translated texts from original texts.
Approach:	They analyze linguistic features which enable our model to achieve high accuracy by a collection of linguistic characteristics and pretrained neural models pick up these features without any fine-tuning.
Outcome:	The proposed model achieves high accuracy with a set of linguistic features that correspond to translationese theories and pretrained neural models pick up these features without any fine-tuning.

Copied to clipboard

Aaron Mueller, Andrew Lee, Shruti Joshi, Ekdeep Singh Lubana, Dhanya Sridhar, Patrik Reizinger

Paraphrasing as Zero-shot Translation with Feature-guided Diversity Enhancement (2026.acl-long)

Challenge:	Existing methods to evaluate features disentangle concepts from activations of neural networks are limited by their quality . current methods for concept identification and steering are sparse autoencoders, but they are not reliable.
Approach:	They propose to evaluate how well featurization methods disentangle one concept from another . they use sentiment, domain, voice, and tense to steer these features .
Outcome:	The proposed evaluations show that featurization methods are insufficient to establish steering selectivity . the results suggest that steering a feature affects many concepts despite a near absence of interaction effects.

Copied to clipboard

Ziyue Yan, Hongying Zan, Xinglin Lyu, Hongfei Xu

VOYAGER: A Training Free Approach for Generating Diverse Datasets using LLMs (2026.acl-long)

Challenge:	Existing studies use parallel corpora for training, which results in less diverse paraphrases.
Approach:	They train a bidirectional multilingual neural machine translation model on a bilingual parallel corpus and use it as a paraphrasing model.
Outcome:	The proposed method generates paraphrases with higher semantic consistency, literal fluency and sentential diversity than existing parabanks and LLMs.

Copied to clipboard

Avinash Amballa, Yashas Malur Saidutta, Chi-Heng Lin, Vivek Kulkarni, Srinivas Chappidi

Challenge:	Large language models (LLMs) are used to generate synthetic datasets but lack diversity . prior work has noted that such generated data lacks diversity - a problem that requires domain expertise.
Approach:	They propose a principled approach that optimizes a mathematical quantity that optimize the diversity of the dataset using determinantal point processes.
Outcome:	The proposed method improves diversity by 1.5-3 times compared to baseline approaches.

CaseFacts: A Benchmark for Legal Fact-Checking and Precedent Retrieval (2026.acl-long)

Copied to clipboard

Akshith Reddy Putta, Jacob Devasier, Chengkai Li

SpecAgent: A Speculative Retrieval and Forecasting Agent for Code Completion (2026.acl-long)

Challenge:	Automated Fact-Checking has largely focused on verifying general knowledge against static corpora.
Approach:	They propose a benchmark to verify colloquial legal claims against Supreme Court precedents . the benchmark leverages large language models to synthesize claims from expert case summaries . they say the benchmark is a step forward in the field of legal fact verification .
Outcome:	The proposed benchmark bridges the gap between layperson assertions and technical jurisprudence while accounting for temporal validity.

Copied to clipboard

George Ma, Anurag Koul, Qi Chen, Yawen Wu, Sachit Kuhar, Yu Yu, Aritra Sengupta, Varun Kumar, Murali Krishna Ramanathan

From Inheritance to Saturation: Disentangling the Evolution of Visual Redundancy for Architecture-Aware MLLM Inference Acceleration (2026.acl-long)

Challenge:	Large Language Models (LLMs) excel at code-related tasks but struggle in real software repositories.
Approach:	They propose a large-scale agent that injects repository context at inference time to improve both latency and code-generation quality by proactively exploring repository files during indexing and constructing speculative context.
Outcome:	Experiments show that SpecAgent achieves 9–11% relative performance gains compared to baselines while significantly reducing inference latency.

Copied to clipboard

Jiaqi Shi, Xulong Zhang, Yuechan Li, Xiaoyang Qu, Jianzong Wang

Your Reasoning Model is Secretly a Reward Model - Optimization-Free Verification from Experience (2026.acl-long)

Challenge:	Existing acceleration strategies suffer from severe "backbone dependency" Existing strategies such as token pruning or layer sparsity suffer from this .
Approach:	They propose a framework that decouples visual redundancy into IVR and architecture-dependent secondary saturation redundancies.
Outcome:	The proposed framework outperforms existing frameworks on Qwen25-VL and Qwa25-LL.

Copied to clipboard

Zhenwen Liang, Ruosen Li, Yujun Zhou, Linfeng Song, Dian Yu, Xinya Du, Haitao Mi, Dong Yu

What Do LLMs Learn First? Asymmetric Learning Dynamics of Input Complexity and Output Ambiguity in Preference Alignment (2026.acl-long)

Challenge:	Existing verifiers operate on the surface text or on confidence proxies derived from token probabilities, which can be brittle.
Approach:	They propose a training-free, non-parametric verifier that summarizes each reasoning trace by an activation delta and compares it to two class centroids computed from labeled experience.
Outcome:	The proposed model improves selection and reranking on large and less-calibrated models.

Copied to clipboard

Mengyang Li, Jingwen Wang, Pinlong Zhao

Challenge:	Existing methods treat all preference pairs uniformly during training.
Approach:	They propose a training framework that maintains separate, adaptive pacing schedules for each dimension.
Outcome:	The proposed training framework outperforms curriculum baselines by 2.1% and 0.21 points . it achieves 42.3% length-controlled win rate on AlpacaEval 2.0 and 7.66 on MT-Bench .

Multi-Agent-as-Judge: Aligning LLM-Agent-Based Automated Evaluation with Multi-Dimensional Human Evaluation (2026.acl-long)

Copied to clipboard

Jiaju Chen, Yuxuan Lu, Xiaojie Wang, Huimin Zeng, Jing Huang, Jiri Gesi, Ying Xu, Bingsheng Yao, Dakuo Wang

Challenge:	Existing "LLM-as-a-judge" evaluation frameworks are limited by persona descriptions and are not generalizable to other tasks.
Approach:	They propose a framework that can automatically construct multiple evaluator personas with distinct dimensions from relevant text documents and instantiate LLM agents with the persona.
Outcome:	The proposed framework can believably simulate human evaluators . it extracts stakeholders' diverse perspectives from the provided research papers and constructs personas for the agents .

OpenRubrics: Towards Scalable Synthetic Rubric Generation for Reward Modeling and LLM Alignment (2026.acl-long)

Copied to clipboard

Tianci Liu, Ran Xu, Tony Yu, Ilgee Hong, Carl Yang, Tuo Zhao, Haoyu Wang

Challenge:	Existing reward models rely on scalar or pairwise judgments that fail to capture multifaceted nature of human preferences.
Approach:	They propose a rubric-based reward model that uses a large collection of prompt, rubric pairs to generate a scalar score or preference label for each response.
Outcome:	The proposed model surpasses strong size-matched baselines by 8.4% across multiple benchmarks.

Challenging the Boundaries of Reasoning: An Olympiad-Level Math Benchmark for Large Language Models (2026.acl-long)

Copied to clipboard

Haoxiang Sun, Yingqian Min, Zhipeng Chen, Xin Zhao, Ji-Rong Wen

LENS: LLM-Enabled Narrative Synthesis for Mental Health by Aligning Multimodal Sensing with Language Models (2026.acl-long)

Challenge:	Existing evaluation frameworks for large reasoning models are saturated by a lack of reliable and verifiable benchmarks.
Approach:	They propose a rigorously curated, Olympiad-level math benchmark comprising 350 problems, each with parallel English and Chinese versions.
Outcome:	The proposed benchmark unifies two evaluation paradigms and offers 150 problems formalized in Lean 4 for rigorous process-level evaluation.

Copied to clipboard

Wenxuan Xu, Arvind Pillai, Subigya Nepal, Amanda C. Collins, Daniel M Mackin, Michael V. Heinz, Tess Z Griffin, Nicholas C. Jacobson, Andrew Campbell

FedProxy: Federated Fine-Tuning of LLMs via Proxy SLMs and Heterogeneity-Aware Fusion (2026.acl-long)

Challenge:	Current LLMs cannot natively ingest long-duration sensor streams and paired sensor–text datasets are scarce.
Approach:	They propose a framework that aligns multimodal sensing data with language models to generate clinically grounded mental-health narratives.
Outcome:	The proposed framework outperforms baselines on NLP metrics and task-specific measures of symptom severity and clinically meaningful narratives.

Copied to clipboard

Tao Fan, Guoqiang Ma, Yuanfeng Song, Lixin Fan, Kai Chen, Qiang Yang

DiZiNER: Disagreement-guided Instruction Refinement via Simulating Pilot Annotation for Zero-shot Named Entity Recognition (2026.acl-long)

Challenge:	Existing methods for fine-tuning Large Language Models (LLMs) suffer from a performance bottleneck . Existing approaches like Offsite-Tuning (OT) secure the LLMs IP .
Approach:	They propose a framework that replaces weak adapters with a unified, powerful Proxy Small Language Model (SLM) they propose 'resource-friendly' compression and 'robust optimization' to handle data heterogeneity.
Outcome:	Experiments show that FedProxy outperforms OT and centralized fine-tuning methods.

Copied to clipboard

Siun Kim, Hyung-Jin Yoon

A Data-Centric Approach to Generalizable Speech Deepfake Detection (2026.acl-long)

Challenge:	Large language models have advanced information extraction (IE) by enabling zero-shot and few-shot named entity recognition (NER) but their outputs still show persistent and systematic errors.
Approach:	They propose a framework that simulates the pilot annotation process and employs LLMs as both annotators and supervisors to refine model disagreements.
Outcome:	Using a pilot annotation process, the proposed framework outperforms its supervisor model on 18 benchmarks.

Copied to clipboard

Wen Huang, Yuchen Mao, Yanmin Qian

Free Energy-Driven Reinforcement Learning with Adaptive Advantage Shaping for Unsupervised Reasoning in LLMs (2026.acl-long)

Challenge:	Speech deepfake detection (SDD) is a critical research area as speech synthesis technologies become more sophisticated.
Approach:	They propose a data-centric approach to generalize SDD data from two perspectives . they propose naive aggregation strategies for mixing heterogeneous data and diversity-optimized sampling strategy for a single dataset and multiple datasets.
Outcome:	The proposed approach outperforms the naive aggregation baseline on a 12k-hour data pool while using only 3% of the total available data.

Copied to clipboard

Yiming Huang, Zhenbo Shi, Xin-Cheng Wen, Jichuan Zeng, Cuiyun Gao, Peiyi Han, Chuanyi Liu

PaperRegister: Boosting Flexible-grained Paper Search via Hierarchical Register Indexing (2026.acl-long)

Challenge:	Existing unsupervised reinforcement learning methods lack the capacity to adapt to the model’s evolving reasoning capabilities during training.
Approach:	They propose an unsupervised reinforcement learning algorithm that adapts rewards to balance consensus and exploration based on the Free Energy Principle.
Outcome:	Empirical evaluations on nine datasets show that FREIA outperforms baseline methods on reasoning tasks.

Copied to clipboard

Zhuoqun Li, Xuanang Chen, Hongyu Lin, Yaojie Lu, Xianpei Han, Shanshan Jiang, Bin Dong, Le Sun

ETR: Entropy Trend Reward for Efficient Chain-of-Thought Reasoning (2026.acl-long)

Challenge:	Existing paper search systems lack detailed information to support finer-grained queries.
Approach:	They propose a paper-based index that transforms abstract-based corpus index into hierarchical index tree and offline can support paper search queries.
Outcome:	The proposed system achieves the SOTA performance and excels in fine-grained scenarios.

Copied to clipboard

Xuan Xiong, Huan Liu, Li Gu, Zhixiang Chi, Yue Qiu, Yuanhao YU, Yang Wang

All Changes May Have Invariant Principles: Improving Ever-Shifting Harmful Meme Detection via Design Concept Reproduction (2026.acl-long)

Challenge:	Existing methods to shorten CoTs use length penalties or global entropy reduction . Existing approaches to CoT reasoning have significant practical drawbacks .
Approach:	They propose a method that shortens CoTs by length penalties or global entropy reduction . they integrate ETR into Group Relative Policy Optimization and evaluate it .
Outcome:	The proposed objective improves accuracy–efficiency trade-off by +9.9% while reducing CoT length by 67% across four benchmarks.

Copied to clipboard

Ziyou Jiang, Mingyang Li, Junjie Wang, Yuekai Huang, Jie Huang, Zhiyuan Chang, Zhaoyang Li, Qing Wang

Activation-Guided Local Editing for Jailbreaking Attacks (2026.acl-long)

Challenge:	Existing methods for harmful meme detection only learn the combination of harmful elements and lack understanding of these implicit expressions.
Approach:	They propose a method that detects harmful memes by replicating the design concept of malicious users.
Outcome:	The proposed method achieves the highest accuracy with 81.1% and has slight accuracy decreases when generalized to type-shifting and temporal-evolving memes.

Copied to clipboard

Jiecong Wang, Haoran Li, Hao Peng, Ziqian Zeng, Zihao Wang, Haohua Du, Zhengtao Yu

Controlling Distributional Bias in Multi-Round LLM Generation via KL-Optimized Fine-Tuning (2026.acl-long)

Challenge:	Existing methods for jailbreaking Large Language Models (LLMs) are limited and produce incoherent or unreadable inputs.
Approach:	They propose a two-stage framework that performs a one-shot, scenario-based generation of context and rephrases the original malicious query to obscure its harmful intent.
Outcome:	The proposed framework achieves state-of-the-art Attack Success Rate, with gains of up to 37.74% over the strongest baseline, and excellent transferability to black-box and large-scale models.

Copied to clipboard

Yanbei Jiang, Amr Keleg, Ryandito Diandaru, Jey Han Lau, Lea Frermann, Biaoyan Fang, Fajri Koto

On the Continued Value of Universal Dependencies in the Era of Large Language Models (2026.acl-long)

Challenge:	Existing evaluation methods focus on single-round inference, but this view is problematic in real-world applications.
Approach:	They propose a framework that couples Steering Token Calibration with Semantic Alignment to ensure that LLMs are correctly aligned across gender, race, and sentiment.
Outcome:	The proposed framework outperforms baseline methods in achieving precise distributional control in attribute generation tasks.

Copied to clipboard

Wenxi Li, Jingyu Peng

Adaptive Text2GQL: Integrating Structural Twig Linking and Evolutionary In-Context Learning (2026.acl-long)

Challenge:	a growing belief that explicit linguistic representations are no longer necessary is questioned in large language models . a recent study examines whether and in what ways this cross-lingual syntactic framework can still benefit LLMs .
Approach:	They use Universal Dependencies (UD) to examine whether and in what ways it can still benefit LLMs.
Outcome:	The proposed model outperforms its syntax-agnostic counterparts in a cross-lingual evaluation task.

Copied to clipboard

Fang Niu, Chaokun Wang, Hang Zhang, Songyao Wang

XToM: Exploring the Multilingual Theory of Mind for Large Language Models (2026.acl-long)

Challenge:	Existing approaches struggle with structural hallucinations and lack adaptability in cold-start scenarios.
Approach:	They propose a unified, training-free framework for translating natural language into Graph Query Languages.
Outcome:	The proposed framework improves accuracy and executability over baselines in Graph2GQLs.

Copied to clipboard

Chunkit Chan, Yauwai Yim, Hongchuan Zeng, Zhiying Zou, Xinyuan Cheng, Zhifan Sun, Zheye Deng, Kawai Chung, Yuzhuo Ao, Fan Yixiang, Cheng Jiayang, Ercong Nie, Ginny Wong, Helmut Schmid, Hinrich Schuetze, Simon See, Yangqiu Song

Cognitive Policy-Driven LLM for Diagnosis and Intervention of Cognitive Distortions in Emotional Support Conversation (2026.acl-long)

Challenge:	Existing evaluations of ToM in LLMs are limited to English, neglecting the linguistic diversity that shapes human cognition.
Approach:	They propose a multilingual benchmark that evaluates ToM across five languages . they find that models excel in multilingual language understanding, but their ToM performance varies across languages.
Outcome:	The proposed benchmark evaluates LLMs across five languages and incorporates diverse task scenarios.

Copied to clipboard

Lin Zhong, Renjin Zhu, Shujuan Ma, Jinhao Cui, Lingzhi Wang, Hao Chen, Qing Liao

SciFlow-Bench: Evaluating Structure-Aware Scientific Diagram Generation via Inverse Parsing (2026.acl-long)

Challenge:	Existing models for ESC ignore cognitive distortions in help-seekers' expressions . current models provide basic emotional comfort, rather than helping help- seekers address psychological distress at a deeper cognitive level.
Approach:	They propose a Large Language Model framework to enhance LLMs' ability to diagnose and intervene cognitive distortions in help-seekers.
Outcome:	The proposed framework outperforms 15 state-of-the-art baselines in terms of distortion diagnosis accuracy, intervention strategy effectiveness, and safety risk control.

Copied to clipboard

Tong Zhang, Honglin Lin, Zhou Liu, Chong Chen, Wentao Zhang

Immediate Inference: The Missing Foundation in Large Language Model Logical Reasoning (2026.acl-long)

Challenge:	Existing benchmarks for scientific diagram generation rely on image-centric metrics or evaluation of intermediate symbolic representations rather than final rendered images.
Approach:	They propose a structure-first benchmark for evaluating scientific diagram generation from pixel-level outputs.
Outcome:	The proposed benchmark evaluates scientific diagram generation directly from pixel-level outputs.

Copied to clipboard

Sihang Jiang, Zhiyu Lu, Keyi Wang, Jiaqing Liang, Yanghua Xiao, Xiaojun Meng, Jiansheng Wei

Tracing Logit Trajectories Across Layer Depth: Dataset-Level Explainability for Language Models (2026.acl-long)

Challenge:	Recent work on LLMs has focused on fine-grained skill decomposition and consistency probing at the propositional level.
Approach:	They propose a benchmark evaluating immediate inference that evaluates elemental operations over categorical propositions and proposes a model that uses immediate inferential reasoning.
Outcome:	The proposed benchmark demonstrates that models lack robust operator grounding, oscillating between structural reasoning and surface pattern matching, inconsistent handling of quantifiers and negation.

Copied to clipboard

Jeesu Jung, Sangkeun Jung

REZE: Representation Regularization for Domain-adaptive Text Embedding Pre-finetuning (2026.acl-long)

Challenge:	Sentence-level explanations miss the bigger picture of how a black-box model behaves across data . a dataset-level analysis that traces the intermediate structure of decision formation is needed .
Approach:	They propose a method that aggregates logit updates into a reproducible dataset-level trajectory pattern.
Outcome:	The proposed model enables depth-wise explainability across 6 languages and 5 NLP tasks.

Copied to clipboard

Seungmin Lee, Jeonghwan Lee, Hyunkuk Lim, Sejoon Kim, Mingi Sung

TAMA: Target-Aware Multilingual Abuse Detection by Cascaded Conditional Multi-Task Learning (2026.acl-long)

Challenge:	Recent text embedding models often introduce task-induced bias alongside domain knowledge, leading to performance degradation.
Approach:	They propose a representation regularization framework that explicitly controls representation shift during embedding pre-finetuning.
Outcome:	The proposed framework outperforms standard pre-finetuning and isotropy-oriented post-hoc regularization in most settings.

Copied to clipboard

JiYan Liu, Youzheng Liu, Taihang Wang, Yimin Wang, Ye Jiang, Diana Maynard

Reference Attack: A New Cross-Modal Jailbreaking Attack against Multimodal Large Language Models (2026.acl-long)

Challenge:	Existing models for protecting public figures from online abuse ignore who is targeted and how.
Approach:	They propose a target-aware multi-task framework that conditions downstream predictions on upstream beliefs via three lightweight modules: Cross-Task Feature Fusion (CTF), Task-Adaptive Gating (TAG), and Label-Guided Span Detection (LGSD).
Outcome:	The proposed framework yields higher average F1 than single-task training and standard multi-task learning.

Copied to clipboard

Yulong Wang, Yifei Fu, Jiayi Gao

POWSM: A Phonetic Open Whisper-Style Speech Foundation Model (2026.acl-long)

Challenge:	Large Language Models (LLMs) have raised significant safety concerns about generated content, drawing attention from both academia and industry.
Approach:	They propose a reference-guided cross-modal jailbreak method that enhances existing prompt-to-image injection attacks by exploiting MLLMs’ semantic reconstruction capabilities.
Outcome:	The proposed method achieves an attack success rate of over 93% on leading MLLMs including ChatGPT, Gemini, Claude, and the widely used open-source LLaMA model.

Copied to clipboard

Chin-Jou Li, Kalvin Chang, Shikhar Bharadwaj, Eunjung Yeo, Kwanghee Choi, Jian Zhu, David R. Mortensen, Shinji Watanabe

LCR-RAG: Enhancing Logical Consistency in Retrieval-Augmented Generation via Neuro-symbolic Reinforcement Learning (2026.acl-long)

Challenge:	Phone-level modeling of speech is a common approach to speech recognition, but it relies on task-specific architectures and datasets.
Approach:	They propose a phonetic framework capable of performing multiple phone-related tasks . they propose 'Phonetic Open Whisper-style Speech Model' that can perform these tasks together .
Outcome:	The proposed model outperforms or matches specialized PR models of similar size while supporting G2P, P2G, and ASR.

Copied to clipboard

Wenxiang Zheng, Guo Tang, Shixin Jiang, Liangyu Huo, Xiyuan Zhang, Jian Xie, Ming Liu

CIA: Inferring the Communication Topology from LLM-based Multi-Agent Systems (2026.acl-long)

Challenge:	Retrieval-Augmented Generation (RAG) is widely used to ground large language models in external knowledge and improve factual accuracy.
Approach:	They propose a framework that integrates neuro-symbolic verification with reinforcement learning to optimize logical consistency.
Outcome:	The proposed framework outperforms strong RAG baselines on hotpotQA, ASQA, and TriviaQA.

Copied to clipboard

Yongxuan Wu, Xixun Lin, He Zhang, Nan Sun, Kun Wang, Chuan Zhou, Shirui Pan, Yanan Cao

Efficiently Learning To Reason or Not to Reason: Root-token Policy Optimization for Adaptive Thinking (2026.acl-long)

Challenge:	LLM-based multi-agent systems (MAS) have demonstrated remarkable capabilities in solving complex tasks.
Approach:	They propose a communication inference attack that constructs new adversarial queries to induce intermediate agents’ reasoning outputs and models their semantic correlations through the global bias disentanglement and LLM-guided weak supervision.
Outcome:	The proposed attack achieves an average AUC of 0.87 and a peak AUC up to 0.99, revealing the privacy risk in MAS.

Copied to clipboard

Taehyeon Kim, Hyunsoo Lee, Youngsoo Jang, Moontae Lee

SAGE: Synergistic Adaptive Gating of Experts for Hateful Video Detection (2026.acl-long)

Challenge:	Large reasoning models (LRMs) externalize explicit reasoning traces before producing the answer, yet suffer from overthinking challenge.
Approach:	They propose a framework that enables large reasoning models to self-determine when to reason by training only the initial root token via group relative reward and group-wise advantages.
Outcome:	The proposed framework reduces training overhead and VRAM usage by focusing on the root token . it learns difficulty-aware adaptive thinking at just 2% of the training compute of prior methods.

Copied to clipboard

Jie Huang, Xin Liao, Junjie Wang, Mingyang Li, Wenshuo Wang, Ziyou Jiang, Shoubin Li, Qing Wang

HiKEY: Hierarchical Multimodal Retrieval for Open-Domain Document Question Answering (2026.acl-long)

Challenge:	Existing methods for hateful video detection rely on multimodal feature fusion . existing methods rely only on blind feature mixing, which leads to feature dilution .
Approach:	They propose a framework that shifts from blind feature mixing to decision-level arbitration . it instantiates disentangled experts to rigorously preserve modality-specific semantics .
Outcome:	The proposed framework outperforms state-of-the-art methods on HateMM and MultiHateClip benchmarks.

Copied to clipboard

Joongmin Shin, Gyuho Shim, Jeongbae Park, Jaehyung Seo, Heuiseok Lim

Towards Efficient and Effective Diffusion Language Model Inference via Semantic-Aware Adaptive Denoising (2026.acl-long)

Challenge:	Existing approaches to document-based Opendomain Question Answering (ODQA) use flat text chunks or page-level images to locate the correct document.
Approach:	They propose a hierarchical tree-based multimodal retrieval framework that elevates document hierarchy to a first-class retrieval signal.
Outcome:	The proposed framework outperforms page- and chunk-based baselines on ODQA benchmarks and improves retrieval recall by 12.9% and end-to-end QA performance by 6.8%.

Copied to clipboard

Fan Li, Yu Gu, Zhigang Wang, Fangling Leng, Zhenghao Liu, Ge Yu

AdvancedIF: Rubric-Based Benchmarking and Reinforcement Learning for Advancing LLM Instruction Following (2026.acl-long)

Challenge:	Existing acceleration works cannot accurately detect semantically stabilized tokens and then skip computation, leading to sub-optimal speedup in practice.
Approach:	They propose a semantic-aware adaptive denoising framework that encodes scalar confidence scores into an evolution-awful feature vector and clusters vectors proactively and adaptively identify semantically converged tokens.
Outcome:	The proposed framework outperforms the SOTA competitor in speed and quality . it can detect semantically stabilized tokens and skip computation, resulting in sub-optimal speedup .

Copied to clipboard

Yun He, Wenzhe Li, Hejia Zhang, Songlin Li, Karishma Mandyam, Sopan Khosla, Yuanhao Xiong, Nanshu Wang, Xiaoliang Peng, Beibin Li, Shengjie Bi, Shishir G Patil, Qi Qi, Shengyu Feng, Julian Katz-Samuels, Richard Yuanzhe Pang, Sujan Kumar Gonugondla, Hunter Lang, Yue Yu, Yundi Qian, Maryam Fazel-Zarandi, Licheng Yu, Amine Benhalloum, Hany Hassan Awadalla, Manaal Faruqui

Improving Autoformalization Using Direct Dependency Retrieval (2026.acl-long)

Challenge:	Recent advances in large language models (LLMs) have shown impressive performance on a range of tasks, yet advanced instruction following (IF) remains a significant challenge.
Approach:	They propose a benchmark that features over 1,600 prompts and expert-curated rubrics that assess LLMs’ ability to follow complex, multi-turn, and system-level instructions.
Outcome:	The proposed framework improves instruction-following abilities of large language models, achieving a 6.7% gain on AdvancedIF and strong results on public benchmarks.

Copied to clipboard

Shaoqi Wang, Lu Yu, Siwei Lou, Feng Yan, Chunjie Yang, Qing Cui, Jun Zhou

Long-Chain Reasoning Distillation via Adaptive Prefix Alignment (2026.acl-long)

Challenge:	Existing methods for hallucinate formal dependencies lack scalability and precision to leverage ever-growing public datasets.
Approach:	They propose a retrieval-augmented framework based on Direct Dependency Retrieval to generate formal dependencies from natural-language mathematical descriptions and verify their existence via an efficient Suffix Array Check (SAC).
Outcome:	The proposed framework outperforms state-of-the-art methods in retrieval precision and recall and can be used to validate formal representations in a public dataset.

Copied to clipboard

Zhenghao Liu, Zhuoyang Wu, Xinze Li, Yukun Yan, Shuo Wang, Zulong Chen, Yu Gu, Ge Yu, Maosong Sun

TRN-R1-Zero: Text-rich Network Reasoning via LLMs with Reinforcement Learning Only (2026.acl-long)

Challenge:	Large Language Models (LLMs) have demonstrated remarkable reasoning capabilities, especially in solving complex mathematical problems.
Approach:	They propose a framework that exploits teacher CoTs for distillation through adaptive prefix alignment.
Outcome:	The proposed framework outperforms baseline models on multiple mathematical reasoning benchmarks by over 3%.

Copied to clipboard

Yilun Liu, Ruihong Qiu, Zi Huang

Threshold Differential Attention for Sink-Free, Ultra-Sparse, and Non-Dispersive Language Modeling (2026.acl-long)

Challenge:	Recent large language model-based approaches often overlook graph context or depend on distillation from larger models, limiting generalisation.
Approach:	They propose a framework for zero-shot reasoning on text-rich networks . they use a Neighbour-aware Group Relative Policy Optimisation objective .
Outcome:	The proposed framework optimises base LLMs using a Neighbour-aware group relative policy optimisation objective based on a novel margin gain metric for the informativeness of neighbouring signals .

Copied to clipboard

Xingyue Huang, Xueying Ding, Mingxuan Ju, Yozen Liu, Neil Shah, Tong Zhao

PRiSM: Benchmarking Phone Realization in Speech Models (2026.acl-long)

Challenge:	a strict sum-to-one constraint forces attention sinks on irrelevant tokens, while probability mass disperses as sequence lengths increase.
Approach:	They propose a sink-free attention mechanism that achieves ultra-sparsity and improved robustness at longer sequence lengths without the computational overhead of projection methods.
Outcome:	The proposed mechanism produces >99 % exact zeros and eliminates attention sinks while maintaining competitive performance on standard and long-context benchmarks.

Copied to clipboard

Shikhar Bharadwaj, Chin-Jou Li, Yoonjae Kim, Kwanghee Choi, Eunjung Yeo, Ryan Soh-Eun Shim, Hanyu Zhou, Brendon Boldt, Karen Rosero, Kalvin Chang, Darsh Agrawal, Keer Xu, Chao-Han Huck Yang, Jian Zhu, Shinji Watanabe, David R. Mortensen

Your Reasoning Benchmark May Not Test Reasoning: Revealing Perception Bottleneck in Abstract Reasoning Benchmarks (2026.acl-long)

Challenge:	Existing evaluations of phone recognition systems only measure surface-level transcription accuracy.
Approach:	They propose to standardize transcription-based evaluation and assess downstream utility in clinical, educational, and multilingual settings with transcription and representation probes.
Outcome:	The proposed system outperforms LALMs in clinical, educational, and multilingual settings.

Copied to clipboard

Xinhe Wang, Jin Huang, Xingjian Zhang, Tianhao Wang, Jiaqi W. Ma

Model-Based Imaginative Planning for Embodied Agents (2026.acl-long)

Challenge:	Abstraction and Reasoning Corpus and ARC-AGI are widely used to assess progress in artificial intelligence.
Approach:	They propose a two-stage pipeline that separates perception and reasoning . they propose to test this pipeline against standard end-to-end one-stage evaluation .
Outcome:	The proposed pipeline separates perception and reasoning, and isolates reasoning from bottlenecks.

Copied to clipboard

Junru Song, Hengzhe Jin, Yucong Huang, Tingsong Jiang, Weien Zhou, Feifei Wang, Yang Yang, Ying Wen, Wen Yao

Persona-Grounded Safety Evaluation of AI Companions in Multi-Turn Conversations (2026.acl-long)

Challenge:	a lightweight world model converts raw pixels into object-centric symbolic states amenable to language-based reasoning . IMPLEMENT is a framework for grounding language agents in visual embodied environments .
Approach:	They propose a model-based reasoning framework that enables frozen large language models to perform imaginative planning.
Outcome:	The proposed framework can be used to ground language agents in visual embodied environments.

Copied to clipboard

Prerna Juneja, Lika Lomidze

Beyond Chunking: Discourse-Aware Hierarchical Retrieval for Long Document Question Answering (2026.acl-long)

Challenge:	Existing safety evaluations rely on self-reported user data or interviews . a recent study evaluated how Replika responds to high-risk user groups .
Approach:	They propose a framework for controlled simulation and safety evaluation of multi-turn interactions with AI companion applications.
Outcome:	The proposed framework evaluates how Replika responds to high-risk user groups . it incorporates emotion modeling and LLM-assisted utterance-and harm-level classification .

Copied to clipboard

Huiyao Chen, Yi Yang, Yinghui Li, Meishan Zhang, Baotian Hu, Min Zhang

Beyond Value Benchmarks: Measuring Value-Structure Alignment in Large Language Models via Symmetric Q-Sorts (2026.acl-long)

Challenge:	Existing long document question answering systems process texts as flat sequences or use heuristic chunking, which overlooks the discourse structures that guide human comprehension.
Approach:	They propose a discourse-aware hierarchical framework that leverages rhetorical structure theory for long document question answering.
Outcome:	The proposed framework exhibits strong robustness across diverse document types and linguistic settings.

Copied to clipboard

Jingting Zheng, Yuqi Ren, Linhao Yu, Yongqi Leng, Deyi Xiong

Would LLMs be Good Historical Linguists and Chinese Dialect Learners? (2026.acl-long)

Challenge:	Existing evaluations of Large Language Models (LLMs) focus on item-level behavioral metrics without capturing how models prioritize competing values as a whole.
Approach:	They propose a symmetric human-LLM evaluation framework to measure value-structure alignment . they evaluate 12 LLMs across four model families via 240 replicated Q-sorts .
Outcome:	The proposed framework measures value-structure alignment across four model families.

Copied to clipboard

Yicheng Liu, Shumin Shi, Youchao Zhou, Xingchen Zhang

ProActor: Timing-Aware Reinforcement Learning for Proactive Task Scheduling Agents (2026.acl-long)

Challenge:	Large language models struggle with low-resource Chinese dialects due to substantial phonological divergence.
Approach:	They propose to incorporate Middle Chinese, the common historical ancestor of modern Chinese dialects, into LLMs to improve dialectal pronunciation modeling.
Outcome:	The proposed approach improves on standard Chinese but struggles with low-resource Chinese dialects . the proposed model improves over baselines while revealing variation across dialects.

Copied to clipboard

Lei Ding, Bin He, Chenguang Wang, Yang Liu

Seeing No Evil: Blinding Large Vision-Language Models to Safety Instructions via Adversarial Attention Hijacking (2026.acl-long)

Challenge:	Existing approaches to measuring and optimizing proactive task-oriented agents lack generalizable end-to-end solutions.
Approach:	They propose a framework for conversational task scheduling that integrates proactiveness reinforcement learning with a domain-agnostic annotation methodology.
Outcome:	The proposed framework enables scalable proactiveness reinforcement learning (RL) Experiments on two newly auto-annotated datasets demonstrate significant improvements in proactive timing while maintaining action consistency comparable to state-of-the-art baselines.

Copied to clipboard

Jingru Li, Wei Ren, Tianqing Zhu

CADMate: Generating CAD Assembly Plan with Geometric Chain-of-Thought and Spatial Physical Rewards (2026.acl-long)

Challenge:	Existing attacks optimize image perturbations to maximize harmful output likelihood, but suffer from slow convergence due to gradient conflict between adversarial objectives and the model’s safety-retrieval mechanism.
Approach:	They propose a push-pull approach which suppresses attention to system-prompt tokens and anchors generation on adversarial image features to avoid collisions.
Outcome:	The proposed approach reduces gradient conflict by 45% and achieves 94.4% attack success rate on Qwen-VL (vs. 68.8% baseline) with 40% fewer iterations.

Copied to clipboard

Jiali Chen, DingBa Fu, Xusen Hei, Yuhang Liu, Yiyang Chen, Jiayuan Xie, Wenqi Fan, Yi Cai

Semantic-Aware Logical Reasoning via a Semiotic Framework (2026.acl-long)

Challenge:	Computer-aided design (CAD) is crucial in prototyping complex 3D objects . designers manually define assembly sequences for individual CAD parts .
Approach:	They propose a framework for computer-aided design that predicts actions for CAD parts . they use a reference design image and disassembled parts to generate 6-DoF transformations .
Outcome:	The proposed framework outperforms existing MLLMs in the design of CAD assemblies.

Copied to clipboard

Yunyao Zhang, Xinglang Zhang, Junxi Sheng, Wenbing Li, Junqing Yu, Yi-Ping Phoebe Chen, Wei Yang, Zikai Song

Calibration-Aware Policy Optimization for Reasoning LLMs (2026.acl-long)

Challenge:	Existing studies largely overlook the interplay between logical complexity and semantic complexity, limiting their robustness under abstract propositions, ambiguous contexts, and conflicting stances.
Approach:	They propose a semiotic-square-guided framework that integrates automated deduction with reflective verification to manage logical complexity across deeper reasoning chains.
Outcome:	The proposed framework achieves state-of-the-art performance on RepublicQA with 6.25% average gain, and generalizes well to four mainstream logical reasoning benchmarks with an additional 7.05% improvement.

Copied to clipboard

Ziqi Wang, Xingzhou Lou, Meiqi Wu, Zhengqi Wen, Junge Zhang

GBV-SQL: Guided Generation and SQL2Text Back-Translation Validation for Multi-Agent Text2SQL (2026.acl-long)

Challenge:	Existing approaches to model calibration are limited or sacrifice gains in reasoning accuracy.
Approach:	They propose a method that improves calibration by 15% while boosting accuracy by 5% . they propose GRPO-style algorithms that misalign uncertainty-agnostic advantage estimation .
Outcome:	The proposed approach improves calibration by 15% while achieving comparable to or better than GRPO on multiple mathematical reasoning benchmarks.

Copied to clipboard

Daojun Chen, Xi Wang, Shenyuan Ren, Qingzhi Ma, Pengpeng Zhao, An Liu

Branch-and-Browse: Efficient and Controllable Web Exploration with Tree-Structured Reasoning and Action Memory (2026.acl-long)

Challenge:	Existing frameworks for Text2SQL generation still have a critical semantic gap . a dedicated validator translates generated SQL back into natural language and checks whether its logic is aligned with the original question.
Approach:	They propose a framework that introduces Guided Generation with SQL2Text Back-translation Validation . dedicated validator translates generated SQL back into natural language and checks whether logic is aligned with original question .
Outcome:	The proposed framework achieves 63.23% execution accuracy on the BIRD benchmark and 90.42% on repaired BIDR dev.

Copied to clipboard

Shiqi He, Yue Cui, Xinyu Ma, Yaliang Li, Bolin Ding, Mosharaf Chowdhury

Beyond End-to-End: Dynamic Chain Optimization for Private LLM Adaptation on the Edge (2026.acl-long)

Challenge:	Existing methods for embodied reasoning are coarse-grained and expensive . branch-and-browse framework enables fine-grounded, memory-guided, and efficient multi-branch reasoning.
Approach:	They propose a framework that unifies structured reasoning-acting, contextual memory, and efficient execution.
Outcome:	The proposed framework achieves task success rate of 35.8% and reduces execution time by up to 40.4% relative to state-of-the-art methods.

Copied to clipboard

Yebo Wu, Jingguang Li, Chunlin Tian, KaHou Tam, Zhijiang Guo, Li Li

TalkLoRA: Communication-Aware Mixture of Low-Rank Adaptation for Large Language Models (2026.acl-long)

Challenge:	Large Language Models (LLMs) are revolutionizing mobile intelligence, but their implementation on mobile devices is severely bottlenecked by the prohibitive resource demands of LLMs.
Approach:	They propose a paradigm that forgoes end-to-end updates in favor of a sequential, layer-by-layer manner.
Outcome:	Extensive experiments on multiple benchmarks demonstrate the superiority of ChainFed over existing methods, boosting average accuracy by up to 46.46%.

Copied to clipboard

Lin Mu, Haiyang Wang, Li Ni, Lei Sang, Zhize Wu, Peiquan Jin, Yiwen Zhang

TARE: Lightweight Token-Aware Representation Editing for Fine-tuning Transformer-like Models (2026.acl-long)

Challenge:	Existing LoRA methods assume that experts operate independently, leading to unstable routing, expert dominance.
Approach:	They propose a communication-aware MoELoRA framework that relaxes this assumption by introducing expert-level communication prior to routing.
Outcome:	The proposed framework outperforms vanilla LoRA and MoELoRA on diverse language understanding tasks while maintaining expert dominance.

Copied to clipboard

Yulong Wang, Siyu Zhao

Soft Orthogonal Low-Rank Adaptation for Knowledge Sharing in Large Language Model Continual Learning (2026.acl-long)

Challenge:	Existing PEFT methods can be costly and underfit token-level contexts.
Approach:	They propose a PEFT method that performs fine-grained, token-specific edits with a small additional inference overhead and minimal tuning.
Outcome:	The proposed method outperforms state-of-the-art methods in 8 tasks and GLUE with a minimal tuning overhead and inference overhead.

Copied to clipboard

Yitong Wang, Xue Han, WenChun Gao, Qian Hu, Jiahui Wang, Ziqing Wang, Lijun Mei, Junlan Feng

ChartAgent: A Multimodal Agent for Visually Grounded Reasoning in Complex Chart Question Answering (2026.acl-long)

Challenge:	Existing methods for continual learning (CL) are designed to mitigate catastrophic forgetting while neglecting knowledge sharing across tasks.
Approach:	They propose a framework that facilitates knowledge transfer while mitigating catastrophic forgetting by assigning task-specific parameter subspaces to new tasks . they then leverage attribution scores to evaluate task similarity and employ soft orthogonality between task- specific subspace .
Outcome:	The proposed framework facilitates knowledge transfer while mitigating catastrophic forgetting.

Copied to clipboard

Rachneet Kaur, Nishan Srishankar, Zhen Zeng, Sumitra Ganesh

Curing Miracle Steps in LLM Mathematical Reasoning with Rubric Rewards (2026.acl-long)

Challenge:	Recent multimodal LLMs have shown promise in chart-based visual question answering, but their performance declines sharply on unannotated charts.
Approach:	They propose a novel agentic framework that explicitly performs visual reasoning directly within the chart’s spatial domain.
Outcome:	The proposed framework achieves state-of-the-art accuracy on the ChartBench and ChartX benchmarks surpassing prior methods by up to 16.07% absolute gain overall and 17.31% on numerically intensive queries.

Copied to clipboard

Youliang Yuan, Qiuyang Mang, Jingbang Chen, Hong Wan, Xiaoyuan Liu, Junjielong Xu, Jen-tse Huang, Wenxuan Wang, Wenxiang Jiao, Pinjia He

Adaptive Planning for Multi-Attribute Controllable Summarization with Monte Carlo Tree Search (2026.acl-long)

Challenge:	Existing models are susceptible to reward hacking, leading to a substantial overestimation of a model's reasoning ability.
Approach:	They propose a Rubric Reward Model that rewards the entire reasoning trajectory against problem-specific rubrics.
Outcome:	The proposed model outperforms outcome-only supervision on four math benchmarks and boosts Verified Pass@1024 from 26.7% to 62.6% and reduces the incidence of Miracle Steps by 71%.

Copied to clipboard

Sangwon Ryu, Heejin Do, Yunsu Kim, Gary Lee, Jungseul Ok

CPT-Agent: A Cognitive Process Theory-driven Framework for Student Simulation in Writing Development (2026.acl-long)

Challenge:	Controllable summarization is a form of outputs that tailors summaries to user-specified attributes.
Approach:	They propose an adaptive planning framework that reframes the task as planning the order of sequential attribute control with a customized Monte Carlo Tree Search.
Outcome:	The proposed framework surpasses LLM-based self-planning models and fine-tuned baselines in multi-attribute controllable summarization.

Copied to clipboard

Yuhan Chen, Zizhuo Shen, Miaomiao Cheng, Xu Han, Jiefu Gong, Shijin Wang, Wei Song

Team-Based Self-Play With Dual Adaptive Weighting for Fine-Tuning LLMs (2026.acl-long)

Challenge:	Existing LLMs model overly capable learners who over-apply feedback, resulting in pedagogically implausible behavior.
Approach:	They propose a framework that decouples cognitive ability from writing proficiency and models their interaction during writing and revision.
Outcome:	The proposed model produces distinguishable proficiency levels and is consistent with instructional theories.

Copied to clipboard

Wu Li, Yigeng Zhou, Zesheng Shi, Yequan Wang, Min Zhang, Jing Li

SearchGym: Bootstrapping Real-World Search Agents via Cost-Effective and High-Fidelity Environment Simulation (2026.acl-long)

Challenge:	Recent self-training approaches have reduced reliance on human-labeled data, which limits their scalability.
Approach:	They propose a team-based self-play algorithm that iteratively refines alignment without additional human supervision.
Outcome:	The proposed algorithm outperforms baselines and LLM benchmarks in the self-supervised setting.

Copied to clipboard

Xichen Zhang, Ziyi He, Yinghao Zhu, Sitong Wu, Shaozuo Yu, Meng Chu, Wenhu Zhang, Haoru Tan, Jiaya Jia

Learning to Select: Query-Aware Adaptive Dimension Selection for Dense Retrieval (2026.acl-long)

Challenge:	Search agents are a pivotal paradigm for solving open-ended, knowledge-intensive reasoning tasks.
Approach:	They propose a search agent simulation environment that bootstraps robust search agents using Reinforcement Learning.
Outcome:	The proposed model outperforms the web-enhanced ASearcher model by 10.6%.

Copied to clipboard

Zhanyu Wu, Richong Zhang, Zhijie Nie

MeepleLM: A Virtual Playtester Simulating Diverse Subjective Experiences (2026.acl-long)

Challenge:	Existing methods for dense retrieval use pseudo-relevance feedback to model dimension importance . however, they learn global transformations shared across queries and do not model dimension-aware dimension importance.
Approach:	They propose a Query-Aware Adaptive Dimension Selection framework that learns to predict per-dimension importance directly from query embedding.
Outcome:	The proposed framework improves retrieval effectiveness over the full-dimensional and PRF-based models.

Copied to clipboard

Zizhen Li, Chuanhao Li, Yibin Wang, Jianwen Sun, Yukang Feng, Jiaxin Ai, Fanrui Zhang, Mingzhu Sun, Yifei Huang, Kaipeng Zhang

MSMO-ABSA: Multi-Scale and Multi-Objective Optimization for Cross-Lingual Aspect-Based Sentiment Analysis (2026.acl-long)

Challenge:	Recent advances in large language models have expanded the role of board games as creative co-designers . however, current systems lack the capacity to offer constructive critique grounded in the emergent user experience .
Approach:	They propose a large language model that internalizes persona-specific reasoning patterns to accurately simulate the subjective feedback of diverse player archetypes.
Outcome:	The proposed model outperforms commercial models in community alignment and critique quality.

Copied to clipboard

Chengyan Wu, Bolei Ma, Ningyuan Deng, He Yanqing, Yun Xue, Liu Xiaoyong

FineSteer: A Unified Framework for Fine-Grained Inference-Time Steering in Large Language Models (2026.acl-long)

Challenge:	Aspect-based sentiment analysis (ABSA) has seen success with English texts, but real-world social media interactions often involve multiple languages.
Approach:	They propose a framework for cross-lingual ABSA that incorporates code-switched bilingual sentences into the language discriminator and consistency training modules to enhance cross-linguistic alignment.
Outcome:	The proposed framework achieves cross-lingual sentence-level and aspect-level alignment, aligning features of aspect terms in different contextual environments.

Copied to clipboard

Zixuan Weng, Jinghuai Zhang, Kunlin Cai, Ying Li, Peiran Wang, Yuan Tian

Is the Attention Matrix Really the Key to Self-Attention in Multivariate Long-Term Time Series Forecasting? (2026.acl-long)

Challenge:	Existing methods for inference-time steering fail to be effective, utility-preserving and training-efficient due to rigid, one-size-fits-all designs and limited adaptability.
Approach:	They propose a steering framework that decomposes inference-time steering into two stages . they propose 'conditional steering' mechanism that preserves model utility by avoiding unnecessary steering . a 'mixture-of-Steering-Experts' mechanism captures multimodal nature of desired steering behaviors .
Outcome:	The proposed framework outperforms the state-of-the-art methods on safety and truthfulness benchmarks.

Copied to clipboard

Xinyu Li, Kexi Chen, Jiajie Shen, Ying Zheng, Hong Lu, Jin Zhao, Xin Wang

Difficulty-Controllable Cloze Question Distractor Generation (2026.acl-long)

Challenge:	In multivariate long-term time series forecasting, it is widely believed that the effectiveness of self-attention arises from its attention matrix.
Approach:	They propose a multi-branch MLP that isolates the ‘multi-brain mapping with element-wise operation’ structure from the Transformer and shows that it achieves competitive performance.
Outcome:	The proposed model outperforms three classic and three latest Transformer models and shows that it achieves competitive performance.

Copied to clipboard

Seokhoon Kang, Yejin Jeon, Seonjeong Hwang, Gary Lee

Challenge:	Existing methods for generating high-quality distractors lack adaptability and control over difficulty levels.
Approach:	They propose a two-way distractor generation process to generate plausible distractors using an ensemble QA system and a multitask learning strategy to train a difficulty-controllable generation model.
Outcome:	The proposed method significantly outperforms GPT-4o in aligning distractor difficulty with human perception.

ToolPRM: Fine-Grained Inference Scaling of Structured Outputs for Function Calling (2026.acl-long)

Copied to clipboard

Jianghao Lin, Yuanyuan Shi, Xin Peng, Renjie Ding, Hairui Wang, Yuxuan Peng, Bizhe Bai, Weixi Song, Fengshuo Bai, Huacan Chai, Weinan Zhang, Fei Huang, Ying Wen

Challenge:	Existing research on inference scaling focuses on unstructured output generation tasks, such as mathematical problems.
Approach:	They propose an inference-scaling framework that combines fine-grained beam search with ToolPRM, a process reward model scoring each intra-call decision.
Outcome:	The proposed framework outperforms outcome and coarse-grained reward models in predictive accuracy and yields consistent test-time gains on multiple function-calling benchmarks.

GTA: Generating Long-horizon Tasks for Web Agents at Scale (2026.acl-long)

Copied to clipboard

Tenghao Huang, Kung-Hsiang Huang, Prafulla Kumar Choubey, Yilun Zhou, Muhao Chen, Jonathan May, Chien-Sheng Wu

Closing the Modality Reasoning Gap for Speech Large Language Models (2026.acl-long)

Challenge:	Existing benchmarks provide only coarse start–goal annotations without intermediate trajectories . Existing frameworks provide no supervision over the agent's latent decision process .
Approach:	They propose a framework that integrates crawling, retrieval-based seeding, in-context generation and automated quality control to produce realistic tasks paired with executable trajectories.
Outcome:	The proposed framework decouples crawling from generation for greater efficiency and ensures dense supervision through deterministic replays and systematic validation.

Copied to clipboard

Chaoren Wang, Heng Lu, Xueyao Zhang, Shujie Liu, Yan Lu, Jinyu Li, Zhizheng Wu

Logical Phase Transitions: Understanding Collapse in LLM Logical Reasoning (2026.acl-long)

Challenge:	Recent advances in Speech Large Language Models have a modality reasoning gap that is not addressed by prior work.
Approach:	They propose a reinforcement-learning framework that aligns text-conditioned and speech-conditioned trajectories through an asymmetric reward design.
Outcome:	Experiments on MMSU and OBQA show that the proposed framework narrows the modality reasoning gap and achieves state-of-the-art performance among 7B-scale Speech LLMs.

Copied to clipboard

Xinglang Zhang, Yunyao Zhang, ZeLiang Chen, Junqing Yu, Wei Yang, Zikai Song

SpecCache: Speculative KV Cache Reuse for Efficient RAG Serving (2026.acl-long)

Challenge:	Symbolic logical reasoning is a critical yet underexplored capability of large language models (LLMs).
Approach:	They propose a framework that aligns natural language with logical symbols to establish a shared representation and reshapes training dynamics around phase-transition boundaries to progressively strengthen reasoning at increasing logical depths.
Outcome:	The proposed framework mitigates logical reasoning collapse at high complexity while improving generalization to unseen logical compositions.

Copied to clipboard

Zijian Wen, Tao Zhang, Shuangwu Chen, Shenghao Ye, Yu Guo, Qirui Chen, Jingxian Shuai, Yunpeng Hou, Huasen He, null Jianyang

REG: Retrieval via Emotion Similarity for Guiding Empathetic Dialogue Generation (2026.acl-long)

Challenge:	Retrieval-Augmented Generation (RAG) improves LLMs but faces high prefill latency during long contexts.
Approach:	They propose a method that uses deep-layer hidden-state norms to guide token selection . they propose to use deep-layered hidden-status norms as a proxy to guide the token selection.
Outcome:	The proposed SpecCache outperforms state-of-the-art (SOTA) benchmarks.

Copied to clipboard

Xu Wang, Bo Wang, Yang Xiang, Yihong Tang, Dongming Zhao, null Yuzifei, Yuexian Hou

TellWhisper: Tell Whisper Who Speaks When (2026.acl-long)

Challenge:	Empathy relies on the cognitive capacity to relate to similar past experiences. Existing methods prioritize semantic similarity over emotion characteristics, leading to unempathetic responses.
Approach:	They propose a framework that integrates four Emotion Attributes into the retrieval process to ensure explicit emotional alignment.
Outcome:	Empirical results show that REG significantly outperforms baselines, offering a robust solution for empathetic generation.

Copied to clipboard

Yifan Hu, Peiji Yang, Zhisheng Wang, Yicheng Zhong, Rui Liu

Taming System Complexity: Demystifying Software Engineering Agents in Diagnosing Linux Kernel Faults (2026.acl-long)

Challenge:	Existing approaches decouple temporal modeling and speaker modeling when addressing 'when' and 'who' . a new framework that couples temporal structure with speaker dynamics is proposed to address these limitations .
Approach:	They propose a framework that couples temporal and speaker identity within the speech encoder . they propose TS-RoPE, a time-speaker rotary positional encoding that partitions Query/Key channels into temporal, speaker subspaces and applies region-specific rotations to align "when" and "who" cues in selfattention.
Outcome:	The proposed framework couples temporal structure with speaker dynamics in speech encoder . it uses frame-level speaker activity to estimate speaker-activity estimates .

Copied to clipboard

Zhenhao Zhou, Zhuochen Huang, Yike He, Chong Wang, Jiajun Wang, Yijian Wu, Xin Peng, Yiling Lou

Challenge:	Existing LLM agents struggle with identifying bugs in the Linux kernel . bugs can affect billions of users, affecting the Linux Foundation's research on the topic .
Approach:	They propose a LinuxFLBench benchmark to measure the accuracy of LLM agents on the Linux kernel.
Outcome:	The proposed framework improves FL accuracy with minimal costs.

DisCo_Speech: Controllable Zero-Shot Speech Generation with A Disentangled Speech Codec (2026.acl-long)

Copied to clipboard

Tao Li, Wenshuo Ge, Zhichao Wang, Zihao Cui, Yong Ma, Yingying Gao, Chao Deng, Shilei Zhang, Junlan Feng

WildReward: Learning Reward Models from In-the-Wild Human Interactions (2026.acl-long)

Challenge:	DisCo-Speech is a zero-shot controllable text-to-speech framework . standard codecs entangle timbre and prosody, which hinders independent control in continuation-based LMs.
Approach:	They propose a disentangled speech codec and an LM-based generator to solve this problem . they propose fusion and reconstruction that merges content and prosody into unified tokens .
Outcome:	DisCo-Speech achieves competitive voice cloning and superior zero-shot prosody control.

Copied to clipboard

Hao Peng, Yunjia Qi, Xiaozhi Wang, Zijun Yao, Lei Hou, Juanzi Li

Challenge:	Prior work focused on collecting preference pairs, requiring substantial annotation efforts.
Approach:	They propose a pipeline to extract reliable human feedback from in-the-wild interactions . they propose to use WildChat as an interaction source to train the model .
Outcome:	The proposed model achieves comparable or even superior performance compared to conventional models with improved calibration and cross-sample consistency.

Mitigating Catastrophic Forgetting in Target Language Adaptation of LLMs via Source-Shielded Updates (2026.acl-long)

Copied to clipboard

Atsuki Yamaguchi, Terufumi Morishita, Aline Villavicencio, Nikolaos Aletras

SEE: Signal Embedding Energy for Quantifying Noise Interference in Large Audio Language Models (2026.acl-long)

Challenge:	Large language models underperform in languages absent or underrepresented in training data, creating barrier to equitable access for speakers worldwide.
Approach:	They propose a selective parameter update strategy that proactively preserves source knowledge by identifying critical parameters critical to maintaining source abilities.
Outcome:	Experiments in five typologically diverse languages show that SSU mitigates catastrophic forgetting.

Copied to clipboard

Yuanhe Zhang, Jiayu Tian, Yibo Zhang, Shilinlu Yan, Liang Lin, Zhenhong Zhou, Li Sun, Sen Su

Know the Known and the Unknown: Reasonable Answer Generation with Knowledge-Informed Citations (2026.acl-long)

Challenge:	Existing studies on noise lack quantitative analysis and rely on intuition and empirical observation, thus failing to understand practical robustness.
Approach:	They propose a method for quantifying the impact of noise intensity on LALM inputs by using a structured activation subspace derived from the model's internal representations.
Outcome:	The proposed method outperforms existing denoising methods and demonstrates that noise is perceived more accurately than raw audio features.

Copied to clipboard

Yichi Zhang, Zhuo Chen, Lingbing Guo, Jun Xu, Mengshu Sun, Zhizhen Liu, Lei Liang, Wen Zhang, Huajun Chen

FocalOrder: Focal Preference Optimization for Reading Order Detection (2026.acl-long)

Challenge:	Existing approaches focus on generating multi-level citations linked to specific references, making it verifiable and trustworthy.
Approach:	They propose a new data construction pipeline and a benchmark to improve citation granularity and awareness of unknown information.
Outcome:	The proposed model improves on the existing benchmark and data construction pipeline and provides citation granularity and awareness of unknown information.

Copied to clipboard

Fuyuan Liu, Dianyu Yu, He Ren, Nayu Liu, Xiaomian Kang, Delai Qiu, Fa Zhang, Genpeng Zhen, Shengping Liu, Liang Jiaen, null Weihuang, Yining Wang, Junnan Zhu

Challenge:	Existing methods for document comprehension rely on uniform supervision, resulting in a performance degradation in the intermediate sections.
Approach:	They propose a framework driven by Focal Preference Optimization to detect reading order in document layouts.
Outcome:	The proposed framework outperforms competing baselines and surpasses large-scale general VLMs.

TLSA: LLM-Guided Text-Label Space Alignment with Contrastive Learning for Generalized Category Discovery (2026.acl-long)

Copied to clipboard

Wenxi Xu, Chuan Qin, Xi Chen, Chuyu Fang, Yuanchun Zhou, Hengshu Zhu

Hierarchical Token Prepending: Enhancing Information Flow in Decoder-based LLM Embeddings (2026.acl-long)

Challenge:	Existing methods for generalized category discovery suffer from weak text–label alignment, inconsistent objectives across known and novel categories, and poor discrimination of semantically similar clusters.
Approach:	They propose a unified framework that enforces contrastive alignment between text and label representations within a shared semantic space.
Outcome:	The proposed framework outperforms state-of-the-art methods on four benchmark datasets.

Copied to clipboard

Xueying Ding, Xingyue Huang, Mingxuan Ju, Liam Collins, Yozen Liu, Leman Akoglu, Neil Shah, Tong Zhao

Suggest-Verify-Revise: A Three-Stage Document-Level Event Causality Identification with Narrative Consistency (2026.acl-long)

Challenge:	Large language models produce powerful text embeddings, but their causal attention mechanism restricts the flow of information from later to earlier tokens, harming performance.
Approach:	They propose a method that prepending a single summary token to reduce attention-level compression by partitioning the input into blocks and prepending blocks to subsequent blocks.
Outcome:	The proposed method achieves consistent performance gains across 11 retrieval datasets and 30 general embedding benchmarks.

Copied to clipboard

Ya Su, Hu Zhang, Dan Qiao, YuJie Wang, Yunxiao Zhao, Yue Fan, Shike Li, Ru Li, Hongye Tan

Challenge:	Existing methods for document-level Event Causality Identification rely on local semantic similarity for independent event-pair discrimination . Existing approaches ignore the influence of the overall narrative backbone in the propagation of causal dependencies and the role differentiation of events within multi-cause/multi-effect structures.
Approach:	They propose a suggest-verify-revise approach for document-level Event Causality Identification with narrative consistency (SVRECI) they integrate heuristic causal suggestions generated by an LLM with structural suggestions derived from hypergraph modeling .
Outcome:	The proposed approach outperforms existing methods on event-storylines and Causal-TimeBank datasets.

StoryCoder: Narrative Reformulation for Structured Reasoning in LLM Code Generation (2026.acl-long)

Copied to clipboard

Geonhui Jang, Dongyoon Han, YoungJoon Yoo

ODL-TempLLM: Ontology-Guided and Description Logic-Reasoned Temporal Reasoning with LLMs (2026.acl-long)

Challenge:	Existing approaches augment reasoning steps or inject specific structure into how models think, but leave scattered problem conditions unchanged.
Approach:	They propose a narrative reformulation framework that transforms code generation questions into coherent natural language narratives.
Outcome:	The proposed framework improves the performance of 11 code generation models on HumanEval, LiveCodeBench, and CodeForces.

Copied to clipboard

Jinshuo Liu, Cheng Bi, Meng Wang, Juan Deng, Donghong Ji, Jeff Z. Pan

FACTrial: Factorized Clinical Contrastive Training for Scalable Patient-Trial Retrieval (2026.acl-long)

Challenge:	Temporal reasoning is crucial for large language models to understand event concurrency and complex temporal interactions in natural language.
Approach:	They propose an ontology-guided and description logic–constrained temporal reasoning paradigm that shifts focus from internal inference to the explicit modeling of temporal structure.
Outcome:	The proposed method outperforms state-of-the-art methods by 2.07–31.83 F1 points and 1.00–30.73 EM points, exhibiting strong generalization and highlighting the potential of explicit temporal reasoning.

Copied to clipboard

Xuanren Chen, Chongyang Tao, Tao Shen, Shuai Ma

Your Students Don’t Use LLMs Like You Wish They Did (2026.acl-long)

Challenge:	Existing approaches to patient–trial retrieval rely on generic semantic matching and zero-shot transfer.
Approach:	They propose a factorized contrastive training framework that synthesizes diagnosis-aware supervision for scalable patient–trial retrieval.
Outcome:	Experiments show that the proposed framework improves quality and recall coverage.

Copied to clipboard

Sebastian Kobler, Matthew Clemson, Angela Sun, Jonathan K. Kummerfeld

Cut Your Losses! Learning to Prune Paths Early for Efficient Parallel Reasoning (2026.acl-long)

Challenge:	Educational NLP systems are evaluated using engagement metrics and satisfaction surveys . authors identify a fundamental misalignment between pedagogical design and student usage patterns .
Approach:	They propose a computational framework for measuring behaviour in student-AI dialogue . they validate their framework by analysing 12,650 messages from four courses .
Outcome:	The proposed metrics outperform surveys and satisfaction surveys on student-AI dialogues.

Copied to clipboard

Jiaxi Bi, Tongxu Luo, Wenyu Du, Zhengyang Tang, Benyou Wang

Discourse Coherence and Response-Guided Context Rewriting for Multi-Party Dialogue Generation (2026.acl-long)

Challenge:	Parallel reasoning enhances Large Reasoning Models but incurs prohibitive costs due to futile paths caused by early errors.
Approach:	They propose a systematic taxonomy of path pruning to categorize methods by signal source and learnability.
Outcome:	The proposed model improves LRMs but incurs prohibitive costs due to futile paths caused by early errors.

Copied to clipboard

Zhiyu Cao, Peifeng Li, Qiaoming Zhu

EvolvR: Self-Evolving Pairwise Reasoning for Story Evaluation to Enhance Generation (2026.acl-long)

Challenge:	Existing research on multi-party dialogue generation has focused on structural information inherent in dialogues, but colloquial expressions and incomplete utterances often impede comprehension and weaken the fidelity of dialogue structure representations.
Approach:	They propose a framework to improve multi-party dialogue generation through dialogue context rewriting using two complementary feedback signals to construct preference data for both context & response generation.
Outcome:	The proposed framework improves multi-party dialogue generation through dialogue context rewriting.

Copied to clipboard

Xinda Wang, Zhengxu Hou, Yangshijie Zhang, null Yanbingren, Jialin Liu, ChenZhuo Zhao, Zhibo Yang, Bin-Bin Yang, Feng Xiao

Double: Breaking the Acceleration Limit via Double Retrieval Speculative Parallelism (2026.acl-long)

Challenge:	Existing methods for story evaluation lack reasoning capabilities for open-source models . evolvR framework provides high-fidelity evaluators for story generation tasks .
Approach:	They propose a framework that self-synthesizes chain-of-thought data via a multi-persona strategy . they propose evolvR to provide a reward model for story generation .
Outcome:	The proposed framework achieves state-of-the-art performance on three evaluation benchmarks . it also enhances the quality of generated stories, validating the superiority of the framework .

Copied to clipboard

Yuhao Shen, Tianyu Liu, Junyi Shen, Jinyang Wu, Quan Kong, Huan Li, Cong Wang

BWLA: Breaking the Barrier of W1AX Post-Training Quantization for LLMs (2026.acl-long)

Challenge:	Parallel Speculative Decoding (PSD) has limitations due to speedup limits and high computational waste . a novel synchronous mechanism solves the Retrieval Precision-Efficiency Dilemma .
Approach:	They propose a framework that combines a draft-verification-based approach with a synchronous mechanism to solve the Retrieval Precision-Efficiency Dilemma.
Outcome:	The proposed framework breaks speedup limits for Speculative Decoding by overlapping draft generation with verification.

Copied to clipboard

Zhixiong Zhao, Zukang Xu, Dawei Yang

Challenge:	Large language models have driven major progress in NLP, but memory and compute requirements hinder practical deployment.
Approach:	They propose a framework that preserves high accuracy while achieving 1-bit weight quantization . the orthogonal-kronecker transformation learns an orthogonale mapping via EM minimization - a new approach to quantization is proposed .
Outcome:	The proposed framework achieves 1-bit weight quantization with low activations with low-bit activations.

Progressive Multimodal Search and Reasoning for Knowledge-Intensive Visual Question Answering (2026.acl-long)

Copied to clipboard

Changin Choi, Wonseok Lee, Jungmin Ko, Wonjong Rhee

Dynamics of Cognitive Heterogeneity: Investigating Behavioral Biases in Multi-Stage Supply Chains with LLM-Based Simulation (2026.acl-long)

Challenge:	Existing approaches to knowledge-intensive visual question answering lack mechanisms to revise misdirected reasoning.
Approach:	They propose a framework that progressively constructs a structured reasoning trajectory . they use dual-scope queries to retrieve diverse knowledge from heterogeneous knowledge bases .
Outcome:	The proposed framework improves retrieval recall and end-to-end answer accuracy.

Copied to clipboard

Jiuyun Jiang, Yuecheng Hong, Bo Yang, Jin Yang, Guangxin Jiang, Xiaomeng Guo, Guang Xiao

Triviality Corrected Endogenous Reward (2026.acl-long)

Challenge:	Recent advances in Large Language Models (LLMs) allow repeatable experiments in which individual characteristics can be precisely defined.
Approach:	They propose a scalable experimental paradigm using Large Language Models to simulate multi-stage supply chain dynamics.
Outcome:	The proposed model systematically replicates and validates the results of a behavioral simulation on agents in multi-stage supply chain dynamics.

Copied to clipboard

Xinda Wang, Zhengxu Hou, Yangshijie Zhang, null Yanbingren, Jialin Liu, ChenZhuo Zhao, Zhibo Yang, Bin-Bin Yang, Feng Xiao

Identity-Robust Language Model Generation via Content Integrity Preservation (2026.acl-long)

Challenge:	Recent work on unsupervised reinforcement learning for mathematical reasoning using confidence-based endogenous rewards focuses on open-ended text generation, requiring either annotated data or powerful closed-source models.
Approach:	They propose a method that rewards the relative information gain between a specialist and a generalist reference policy, modulated by a probability-dependent correction mechanism.
Outcome:	The proposed model improves on multiple writing benchmarks and model architectures without external supervision and validates generality across different generation tasks.

Copied to clipboard

Miao Zhang, Kelly Chen, Mehrab Tanjim, Rumi Chunara

RFS-Guard: Detecting Reasoning Hallucinations via Cross-Phase Routing Focus in Large Reasoning Models (2026.acl-long)

Challenge:	Existing studies show that Large Language Model outputs vary across sociodemographic attributes . this causes disparities in factual accuracy, utility, and safety, even for objective questions .
Approach:	They propose a lightweight framework for identity-robust generation that neutralizes non-critical identity information while preserving semantically essential attributes.
Outcome:	The proposed framework reduces identity-dependent generation bias by 66.3% over vanilla prompting and outperforms existing prompt-based defenses.

Copied to clipboard

Zihang Liu, Fang Zhouhua, Hui Liu, Zhiwei Liu, Yong Li, Haishuai Wang

Causal-ESC: Reliable Policy Learning for Emotional Support Conversation via Causal Inference (2026.acl-long)

Challenge:	Large reasoning models (LRMs) generate intermediate reasoning traces before the final answer, yet they remain vulnerable to reasoning hallucinations such as subtle arithmetic errors.
Approach:	They propose a Routing Focus Score (RFS) that measures how strongly cross-step attention routing aligns with semantic proximity derived from hidden-state cosine similarity.
Outcome:	The proposed framework detects and localizes hallucinations without external tools or repeated sampling.

Copied to clipboard

Xv Wang, Wang Zhenyu, Guanyu Zheng, Rui Zhang

Chain-of-Thought Compression Should Not Be Blind: V-Skip for Efficient Multimodal Reasoning via Dual-Path Anchoring (2026.acl-long)

Challenge:	Existing approaches to Emotional Support Conversation (ESC) are mechanistically opaque and lacks a causal mechanism between dialogue features and effective empathic strategies.
Approach:	They propose a framework that uses Doubly Robust learning to model causal effects of utterance features on strategy selection.
Outcome:	The proposed framework outperforms state-of-the-art baselines in empathy and helpfulness and provides a theoretically grounded, interpretable solution to the mechanistic interpretability dilemma in affective computing.

Copied to clipboard

Dongxu Zhang, Yiding Sun, Cheng Tan, Wenbiao Yan, Ning Yang, Jihua Zhu, Haijun Zhang

CodeJudgeBench: Benchmarking LLM-as-a-Judge for Coding Tasks (2026.acl-long)

Challenge:	Existing efforts to mitigate this via token compression fail due to its autoregressive nature . linguistically redundant tokens are erroneously pruned, leading to hallucinations .
Approach:	They propose a method that reformulates token pruning as a Visual-Anchored Information Bottleneck (VA-IB) optimization problem.
Outcome:	Experiments on Qwen2-VL and Llama-3.2 families show that the proposed model achieves a speedup with negligible accuracy loss.

Copied to clipboard

Hongchao Jiang, Yiming Chen, Yushi Cao, Hung-yi Lee, Robby T. Tan

Towards a Mechanistic Understanding of Large Reasoning Models: A Survey of Training, Inference, and Failures (2026.acl-long)

Challenge:	Large Language Models (LLMs) are increasingly used to judge code, but their reliability remains poorly understood.
Approach:	They propose a benchmark to evaluate Large Language Models as code judges . they find that small reasoning models outperform larger non-reasoning models .
Outcome:	The proposed benchmark evaluates LLM-as-a-Judge models across three coding tasks.

Copied to clipboard

Yi Hu, Jiaqi Gu, Ruxin Wang, Zijun Yao, Hao Peng, Xiaobao Wu, Jianhui Chen, Muhan Zhang, Liangming Pan

Modeling LLM Unlearning as an Asymmetric Two-Task Learning Problem (2026.acl-long)

Challenge:	Recent research has shown that reinforcement learning can elicit intriguing emergent reasoning behaviors.
Approach:	They propose a comprehensive survey of the mechanistic understanding of large reasoning models . they organize findings into three core dimensions: 1) training dynamics, 2) reasoning mechanisms, and 3) unintended behaviors.
Outcome:	This paper synthesizes the mechanistic understanding of large reasoning models into three dimensions . authors outline a roadmap for future studies including improved interpretability and methodologies .

Copied to clipboard

Zeguan Xiao, Siqing Li, Yong Wang, Xuetao Wei, Jian Yang, Yun Chen, Guanhua Chen

MVP: Enhancing Video Large Language Models via Self-supervised Masked Video Prediction (2026.acl-long)

Challenge:	Large language models (LLMs) are inherently dual-use and can be leveraged for both beneficial and harmful purposes.
Approach:	They propose a retention-prioritized gradient synthesis framework that decouples task-specific gradient extraction from conflict-aware combination.
Outcome:	The proposed method achieves tighter alignment on WMDP Bio and RWKU benchmarks.

Copied to clipboard

Xiaokun Sun, Zezhong Wu, Zewen Ding, Linli Xu

Knowledge Poisoning Attacks on Medical Multi-Modal Retrieval-Augmented Generation (2026.acl-long)

Challenge:	Recent research has attempted to transfer reinforcement learning paradigms to Video Large Language Models (MLLMs) but these methods lack explicit supervision for intrinsic temporal coherence and inter-frame correlations.
Approach:	They propose a novel post-training objective: Masked Video Prediction (MVP) that requires the model to reconstruct a masked continuous segment from a set of challenging distractors and employs Group Relative Policy Optimization (GRPO) with a fine-grained reward function to enhance the model's understanding of video context and temporal properties.
Outcome:	The proposed model improves video reasoning capabilities by reinforcing temporal reasoning and causal understanding.

Copied to clipboard

Peiru Yang, Haoran Zheng, Tong Ju, Shiting Wang, Wanchun Ni, Jiajun Liu, Shangguang Wang, Yongfeng Huang, Tao Qi

Maximizing Local Entropy Where It Matters: Prefix-Aware Localized LLM Unlearning (2026.acl-long)

Challenge:	Existing studies have investigated knowledge poisoning attacks in medical RAG systems . knowledge poison attacks can disrupt model outputs and undermine system reliability .
Approach:	They propose a knowledge poisoning framework that injects misinformation into textual data . they propose to use paired visual data as a query-agnostic trigger to promote retrieval .
Outcome:	The proposed framework produces clinically plausible but incorrect generations on five LLMs and datasets.

Copied to clipboard

Naixin Zhai, Pengyang Shao, Binbin Zheng, Yonghui Yang, Fei Shen, Long Bai, Xun Yang

Challenge:	Existing approaches to machine unlearning treat all tokens indiscriminately and enforce uncertainty over the entire vocabulary.
Approach:	They propose a framework that targets the prefix in a response and minimizes uncertainty in the critical subspace.
Outcome:	The proposed framework achieves superior forgetting efficacy and utility preservation compared to baselines.

When Bigger Isn’t Better: A Comprehensive Fairness Evaluation of Political Bias in Multi-News Summarisation (2026.acl-long)

Copied to clipboard

Nannan Huang, Iffat Maab, Junichi Yamagishi

Language Acquisition Device in Large Language Models (2026.acl-long)

Challenge:	Existing models that deal with multiple sources can exhibit political biases, causing unequal representation of viewpoints and underrepresentation of minority voices.
Approach:	They examine how large language models handle sources with varying political leanings using a dataset with political orientation labels.
Outcome:	The proposed model outperforms larger models and offers the best balance of fairness and efficiency.

Copied to clipboard

Masato Mita, Taiga Someya, Ryo Yoshida, Yohei Oseki

CAML: A Conflict-Aware Molecular Language Model Merging Framework for Multi-Constraint Molecular Generation (2026.acl-long)

Challenge:	Large Language Models (LLMs) are less data-efficient than humans, and pre-pretraining on synthetic languages has been proposed to close this gap.
Approach:	They propose to pre-pretrain on MP-STRUCT, a formal language whose strings encode hierarchical composition, feature-based dependencies, and long-distance displacement via MERGE, AGREE, and MOVE.
Outcome:	The proposed model outperforms k-Shuffle Dyck despite not being definable in C-RASP despite being hierarchically expressive and circuit-theoretically learnable .

Copied to clipboard

Xuanbai Ren, Luoda Tan, Pei Liu, Tengfei Ma, Xiangzheng Fu, Longyue Wang, Yiping Liu, Xiangxiang Zeng

Stratagem: Learning Transferable Reasoning via Trajectory-Modulated Game Self-Play (2026.acl-long)

Challenge:	Existing paradigms struggle with this challenge due to catastrophic forgetting or gradient conflicts.
Approach:	They propose a conflict-aware molecular language model merging framework that generates multiple constraints moleculaire as a cooperative game among property-specific fine-tune models.
Outcome:	The proposed framework generates multiple constraints molecular as a cooperative game among property-specific fine-tune models (expert models) it minimizes conflicts among properties by exploring the optimal combination of the importance of the task parameter and relative fusion weights of each expert (fusion coefficient).

Copied to clipboard

Xiachong Feng, Deyi Yin, Xiaocheng Feng, Yi Jiang, Libo Qin, Yangfan Ye, Lei Huang, Weitao Ma, Qiming Li, Yuxuan Gu, Bing Qin, Lingpeng Kong

NOSE: Neural Olfactory-Semantic Embedding with Tri-Modal Orthogonal Contrastive Learning (2026.acl-long)

Challenge:	Existing self-play approaches to developing general reasoning in language models rely on terminal game outcomes.
Approach:	They propose a game-based reasoning transfer model that addresses two barriers to reasoning transfer.
Outcome:	The proposed model improves mathematical reasoning, general reasoning, and code generation benchmarks.

Copied to clipboard

Yanyi Su, Hongshuai Wang, Zhifeng Gao, Jun Cheng

ReFL: Reflective Feedback Learning for Hallucination Detection of Large Language Models (2026.acl-long)

Challenge:	Existing representation methods fail to fully capture olfactory pathway . current approaches focus on isolated segments of the olefactory pathways .
Approach:	They propose a representation learning framework that aligns molecular structure, receptor sequence, and natural language description.
Outcome:	The proposed framework achieves state-of-the-art and excellent zero-shot generalization . it decouples contributions of molecular structure, receptor sequence, and natural language description .

Copied to clipboard

Cunhang Fan, Jun Zhang, Xue Zhang, Shuai Zhang, Zhao Lv, Jianhua Tao, Zhengqi Wen

Fine-Mem: Fine-Grained Feedback Alignment for Long-Horizon Memory Management (2026.acl-long)

Challenge:	Existing methods for detecting hallucinations depend on external knowledge sources, incurring high computational costs and limiting real-time applicability, or extract the model’s internal states, leading to poor generalization.
Approach:	They propose a hallucination detection framework that leverages corrective in-context learning to guide LLMs to recognize their own prediction errors and adjust internal representations, critically without updating model weights.
Outcome:	The proposed framework outperforms existing methods on two benchmark datasets and achieves state-of-the-art performance.

Copied to clipboard

Weitao Ma, Xiaocheng Feng, Lei Huang, Xiachong Feng, Zhanyu Ma, Jun Xu, Jiuchong Gao, Jinghua Hao, Renqing He, Bing Qin

NaturalSloth: Revisiting Denial-of-Service Attacks on Large Language Models (2026.acl-long)

Challenge:	Existing approaches to memory management rely on final task performance as the primary reward, resulting in severe reward sparsity and ineffective credit assignment.
Approach:	They propose a framework for fine-grained feedback alignment using a Chunk-level step reward and Evidence-Anchored Reward Attribution to redistribute global rewards based on memory items utilized as evidence in reasoning.
Outcome:	The proposed framework outperforms baselines and supports generalization across different model configurations and backbones.

Copied to clipboard

Yiming Chen, Zexin Li, Xianghu Yue, Robby T. Tan, Haizhou Li

HiGoE: Hierarchical Graph of Evidence to Enhance Retrieval-Augmented Generation for Long-context Summarization (2026.acl-long)

Challenge:	Longer generations consume more GPU time, increase latency, and reduce throughput in multi-tenant systems.
Approach:	They propose an adversarial dataset of natural instruction-based DoS prompts to scale the dataset while preserving malicious intent and increasing semantic diversity.
Outcome:	The proposed framework scales with a human-curated seed set of natural instruction-based DoS prompts while preserving malicious intent and increasing semantic diversity.

Copied to clipboard

Long Yuan, Kaiwen Tian, Zi Chen, Bolong Zheng, Chuan Ma

Region-Grounded Report Generation for 3D Medical Imaging: A Fine-Grained Dataset and Graph-Enhanced Framework (2026.acl-long)

Challenge:	Existing methods for long-context summarization fail to capture high-level thematic structures and long-range dependencies.
Approach:	They propose a hierarchical Graph of Evidence to reduce hallucination and attention dilution by replacing unreliable chunk-based methods with a filtered proposition–evidence graph.
Outcome:	Experiments show that HiGoE surpasses baselines in quality and efficiency.

Copied to clipboard

Cong Huy Nguyen, Son Dinh Nguyen, Guanlin Li, Tuan Dung Nguyen, Aditya Narayan Sankaran, Mai Huy Thong, Thanh Trung Nguyen, Mai Hong Son, Reza Farahbakhsh, Phi Le Nguyen, Noel Crespi

UI-Copilot: Advancing Long-Horizon GUI Automation via Tool-Integrated Policy Optimization (2026.acl-long)

Challenge:	Current methods map whole volumes to reports, ignoring the clinical workflow of analyzing localized Regions of Interest (RoIs) Current models exhibit suboptimal accuracy and are prone to significant hallucinations.
Approach:	They propose a framework that mimics the professional radiologist diagnostic workflow by employing graph-based relational modules to capture dependencies between RoI attributes.
Outcome:	The proposed framework surpasses existing models by 19.7% in BLEU and 4.7% in ROUGE-L while achieving a 45.8% improvement in clinical metrics.

Copied to clipboard

Zhengxi Lu, Fei Tang, Guangyi Liu, Jin Ma, Kaitao Song, Xu Tan, Wenqi Zhang, Weiming Lu, Jun Xiao, Yueting Zhuang, Yongliang Shen

STK-Adapter: Incorporating Evolving Graph and Event Chain for Temporal Knowledge Graph Extrapolation (2026.acl-long)

Challenge:	Experimental results show that UI-Copilot-7B achieves state-of-the-art performance on challenging MemGUI-Bench, outperforming strong 7B-scale GUI agents such as GUI-Owl-7B and UITARS-1.5-7B.
Approach:	They propose a collaborative framework where the GUI agent focuses on task execution while a lightweight copilot provides on-demand assistance for memory retrieval and numerical computation.
Outcome:	The proposed framework outperforms GUI-Owl-7B and UI-TARS-1.5-7B on MemGUI-Bench and delivers 17.1% improvement on AndroidWorld over the base Qwen model.

Copied to clipboard

Shuyuan Zhao, Wei Chen, Weijie Zhang, Xinrui Hou, Junfeng Shen, Boyan Shi, Shengnan Guo, Youfang Lin, Huaiyu Wan

Learning What Matters: Dynamic Dimension Selection and Aggregation for Interpretable Vision-Language Reward Modeling (2026.acl-long)

Challenge:	Temporal Knowledge Graphs (TKGs) store dynamic facts in the real world.
Approach:	They propose a Spatial-Temporal Knowledge Adapter which integrates the evolving graph encoder and the LLM to facilitate TKG reasoning.
Outcome:	The proposed method outperforms state-of-the-art methods on benchmark datasets and exhibits strong generalization capabilities in cross-dataset task.

Copied to clipboard

Qiyuan Chen, Hongsen Huang, Jiahe Chen, Qian Shao, Jintai Chen, Hongxia Xu, Renjie Hua, Ren Chuan, Jian Wu

DeepGuard: Secure Code Generation via Multi-Layer Semantic Aggregation (2026.acl-long)

Challenge:	Existing multimodal reward models are interpretable but slow, while discriminative ones are opaque "black boxes."
Approach:	They propose a framework that dynamically decomposes evaluation into granular, interpretable dimensions.
Outcome:	The proposed framework outperforms open-source reward models on benchmarks like VL-RewardBench.

Copied to clipboard

Li Huang, Zhongxin Liu, Yifan Wu, Tao Yin, Dong li, Jichao Bi, Nankun Mu, Hongyu Zhang, Meng Yan

Distillation Traps and Guards: A Calibration Knob for LLM Distillability (2026.acl-long)

Challenge:	Large Language Models (LLMs) can replicate insecure patterns from training data.
Approach:	They propose a framework that leverages distributed security-relevant cues by aggregating representations from multiple upper layers via an attention-based module.
Outcome:	Experiments show that the framework improves the secure-and-correct generation rate by 11.9% over baselines.

Copied to clipboard

Weixiao Zhan, Yongcheng Jing, Leszek Rutkowski, Dacheng Tao

DVMap: Fine-Grained Pluralistic Value Alignment via High-Consensus Demographic-Value Mapping (2026.acl-long)

Challenge:	Knowledge distillation (KD) transfers capabilities from large language models (LLMs) to smaller students, yet it can fail unpredictably and also underpins model leakage risks.
Approach:	They propose a method that allows teachers to control their distillability via reinforcement fine-tuning (RFT) they propose to use tail noise, off-policy instability, and the teacher–student gap to improve KD.
Outcome:	The proposed method outperforms SFT and KD baselines and can be used to protect teachers and students from bottlenecks.

Copied to clipboard

Pengyun Zhu, Yuqi Ren, Zhen Wang, Lei Yang, Deyi Xiong

Challenge:	Current Large Language Models (LLMs) rely on coarse-grained national labels for pluralistic value alignment.
Approach:	They propose a framework for fine-grained pluralistic value alignment using demographic constraints.
Outcome:	The proposed framework can identify groups with predictable, high-consensus value preference . it achieves 48.6% accuracy, surpassing open-source LLM DeepSeek-v3.2 .

HTMR: Hybrid Token Masking Reinforcement Learning with Verifiable Rewards for Event Argument Extraction with Multi-Perspective Reasoning (2026.acl-long)

Copied to clipboard

Jianwen Luo, Yongkang Jin, Yu Hong, Jianmin Yao

IS-CoT: Breaking the Long-form Generation Collapse via Interleaved Structural Thinking (2026.acl-long)

Challenge:	Recent work formulates EAE with large language models as a structured conditional generation task and applies Reinforcement Learning with Verifiable Rewards (RLVR) to optimize sequence-level event structures.
Approach:	They propose a method that selectively updates policy gradients on high-entropy forking tokens and event-critical tokens that define event structure.
Outcome:	The proposed method outperforms full-token and high-entropy only methods and transfers effectively as a plug-and-play approach to other tasks such as named entity recognition and relation classification.

Copied to clipboard

Zechen Sun, Yuyang Sun, Zecheng Tang, Juntao Li, Wenpeng Hu, Wenliang Chen, Zhunchen Luo, Guotong Geng, Min Zhang

Seeing the Whole Elephant: A Benchmark for Failure Attribution in LLM-based Multi-Agent Systems (2026.acl-long)

Challenge:	Existing models with reasoning capabilities suffer from a severe length collapse in open-ended writing .
Approach:	They propose a framework that embeds a dynamic plan-write-reflect cycle into the generation process and train a model with interleaved reasoning traces.
Outcome:	The proposed framework achieves state-of-the-art performance on long-form benchmarks compared to other models on the same dataset.

Copied to clipboard

Mengzhuo Chen, Junjie Wang, Fangwen Mu, Yawen Wang, Zhe Liu, Huanxiang Feng, Qing Wang

MTAVG-Bench: A Diagnostic Benchmark for Multi-Talker Dialogue-Centric Audio-Video Generation (2026.acl-long)

Challenge:	Existing benchmarks rely on partially observable traces that capture only agent outputs . lack of full execution traces obscures many failure causes, authors argue .
Approach:	They propose a benchmark that allows attribution under full execution observability . they find full traces improve attribution accuracy by up to 76.5% over a partial-observation counterpart .
Outcome:	The proposed benchmark improves attribution accuracy by up to 76.5% over a partial-observation counterpart.

Copied to clipboard

Yanghao Zhou, Haitian Li, Rexar Lin, Heyan Huang, Jinxing Zhou, Changsen Yuan, Tian Lan, Ziqin Zhou, Yudong Li, Jiajun Xu, Jingyun Liao, YiMing Cheng, Xuefeng Chen, Xian-Ling Mao, Yousheng Feng

Why LLMs Hallucinate on Structured Knowledge: A Mechanistic Analysis of Reasoning over Linearized Representations (2026.acl-long)

Challenge:	Existing evaluation benchmarks for text-to-audio-video (T2AV) generation are largely designed for human-recorded videos or single-speaker settings.
Approach:	They propose a failure-driven diagnostic benchmark for multi-talker dialogue-centric audio-video generation.
Outcome:	The benchmark evaluates multi-speaker dialogue generation at four levels: audio-visual signal fidelity, temporal attribute consistency, social interaction, and cinematic expression.

Copied to clipboard

Shanghao Li, Jinda Han, Yibo Wang, Yuanjie Zhu, Zihe Song, Langzhou He, Kenan Kamel A Alghythee, Philip S. Yu

SubTokenTest: A Practical Benchmark for Real-World Sub-token Understanding (2026.acl-long)

Challenge:	Existing literature primarily addresses this problem through external interventions such as retrieval augmentation and prompt engineering at the input or output level.
Approach:	They find that LLMs can still produce hallucinated outputs when using structured external knowledge.
Outcome:	The proposed models fail to ground the provided knowledge, causing the model to revert to parametric memory.

Copied to clipboard

Shuyang Hou, Yi Hu, Muhan Zhang

Benchmarking and Enabling Efficient Chinese Medical Retrieval via Asymmetric Encoders (2026.acl-long)

Challenge:	Existing benchmarks have highlighted character-level tasks as lacking practical relevance . many real-world applications rely heavily on precise sub-token understanding .
Approach:	They propose a benchmark that assesses sub-token understanding through practical tasks . they examine the impact of test-time scaling on sub-word reasoning .
Outcome:	The proposed benchmark assesses sub-token understanding through practical tasks . it includes ten tasks across four domains and isolates tokenization-related failures .

Copied to clipboard

Angqing Jiang, Jianlyu Chen, null Zhefang, Yongcan Wang, Xinpeng Li, Keyu Ding, Defu Lian

Merlin’s Whisper: Enabling Efficient Reasoning in Large Language Models via Black-box Persuasive Prompting (2026.acl-long)

Challenge:	Effective medical text retrieval requires high accuracy and low latency.
Approach:	They propose a benchmark for medical text retrieval in Chinese using a symmetric architecture . CARE is a lightweight encoder with an LLM-based encoder for offline document encoding .
Outcome:	The proposed benchmark surpasses state-of-the-art symmetric models on CMedTEB . it matches high retrieval quality without increasing latency, and it performs well on a single GPU .

Copied to clipboard

Heming Xia, Cunxiao Du, Rui Li, Chak Tou Leong, Yongqi Li, Wenjie Li

CAKE: Causal-Guided Adaptive Knowledge Editing for LLMs (2026.acl-long)

Challenge:	Large reasoning models (LRMs) have demonstrated proficiency in tackling complex tasks through step-by-step thinking.
Approach:	They propose a black-box persuasive prompting framework that generates concise responses without compromising accuracy.
Outcome:	The proposed framework reduces token usage while preserving performance.

Copied to clipboard

Shuxin Liu, Jianhao Zhang

Challenge:	Existing knowledge editing methods rely on fixed-layer selection and uniform residual assignment, ignoring heterogeneous causal efficacy of different layers.
Approach:	They propose a method that allows for a causally-guided adaptive knowledge editing that combines causal tracing scores with a constrained quadratic optimization problem.
Outcome:	The proposed method achieves comparable performance with comparable overhead.

Preconditioned Test-Time Adaptation for Out-of-Distribution Debiasing in Narrative Generation (2026.acl-long)

Copied to clipboard

Hanwen Shen, Ting Ying, Jiajie Lu, Shanshan Wang

Backdoor Collapse: Eliminating Unknown Threats Via Known Backdoor Aggregation In Language Models (2026.acl-long)

Challenge:	Debiased large language models excel at handling known or low-bias prompts, but fail on unfamiliar and high-biased prompts.
Approach:	They propose a debiasing framework that detects high-bias prompts and triggers context-aware LoRA updates only when a bias-risk score exceeds a threshold.
Outcome:	The proposed framework reduces toxicity/bias score with significantly lower latency than standard optimization methods.

Copied to clipboard

Liang Lin, Miao Yu, Moayad Aloqaily, Zhenhong Zhou, Kun Wang, Linsey Pang, Prakhar Mehrotra, Qingsong Wen

Less is More: Improving LLM Reasoning with Minimal Test-Time Intervention (2026.acl-long)

Challenge:	Existing defenses rely on impractical assumptions about trigger settings to mitigate backdoor attacks . a recent study found that small amounts of training data can systematically induce harmful behaviors in large language models.
Approach:	They propose a backdoor defense framework that requires no prior knowledge of trigger settings . they use a two-stage process to aggregate backdoor representations and fine-tune recovery .
Outcome:	The proposed defense reduces the average Attack Success Rate to 4.41% across multiple benchmarks . the proposed framework generalizes across different types of backdoors, confirming its robustness in practical deployment scenarios.

Copied to clipboard

Zhen Yang, Mingyang Zhang, Feng Chen, Ganggui Ding, Liang Hou, Xin Tao, Ying-Cong Chen

League of LLMs: A Benchmark-Free Paradigm for Mutual Evaluation of Large Language Models (2026.acl-long)

Challenge:	Recent advances in large language models (LLMs) have focused on test-time scaling to improve reasoning quality but at the cost of efficiency.
Approach:	They propose a training-free framework that enhances reasoning accuracy and stability with minimal overhead.
Outcome:	The proposed framework yields consistent gains across general, coding, and STEM tasks while remaining highly efficient.

Copied to clipboard

Qianhong Guo, Wei Xie, Xiaofang Cai, Enze Wang, Shuoyoucheng Ma, Xiaobing Sun, Tian Xia, Kai Chen, Xiaofeng Wang, Baosheng Wang

RARE: Redundancy-Aware Retrieval Evaluation Framework for High-Similarity Corpora (2026.acl-long)

Challenge:	Large language models (LLMs) have shown exceptional capabilities across a wide range of tasks, but reliable evaluation remains a challenge due to data contamination, opaque operation, and subjective preferences.
Approach:	They propose a benchmark-free evaluation paradigm that organizes multiple LLMs into a self-governed league for multi-round mutual evaluation.
Outcome:	Experiments on eight mainstream LLMs in mathematics and programming show that the proposed model can distinguish capabilities while maintaining high internal ranking stability.

Copied to clipboard

Hanjun Cho, Jay-Yoon Lee

Leibniz: Theory-of-Mind Driven Neuro-Symbolic Logical Reasoning via Multi-Agent Collaboration (2026.acl-long)

Challenge:	Existing QA benchmarks assume distinct documents with minimal overlap, yet RAG systems operate on corpora where information is highly redundant and documents exhibit strong inter-document similarity.
Approach:	They propose a framework for constructing realistic benchmarks by decomposing documents into atomic facts and enhancing LLM-based data generation with CRRF.
Outcome:	The proposed framework can be applied to finance, legal, and patent corpora with high redundancy and similarity.

Copied to clipboard

Yue Fan, Hu Zhang, Yunxiao Zhao, Guangjun Zhang, Hao Zhan, Ru Li, Hongye Tan, Yuanlong Wang

The Paradox of Outcome Optimization: A Causal Information-Theoretic Bound on Reasoning Shortcuts in LLMs (2026.acl-long)

Challenge:	Existing methods for logical reasoning with large language models suffer from insufficient rule semantic grounding and weak rule application mechanisms.
Approach:	They propose a theory-of-mind driven neuro-symbolic reasoning framework that integrates natural language and symbolic representations throughout the reasoning process.
Outcome:	The proposed model surpasses state-of-the-art models in reasoning accuracy and flexibility.

Copied to clipboard

Zihan Chen, Yiming Zhang, Wenxiang Geng, Zenghui Ding, Yining Sun

SPENCE: A Syntactic Probe for Detecting Contamination in NL2SQL Benchmarks (2026.acl-long)

Challenge:	Large Language Models (LLMs) aligned via outcome-based Reinforcement Learning (RL) exhibit a critical failure mode: they exhibit brittle reasoning capabilities on out-of-distribution tasks.
Approach:	They propose a framework bridging Structural Causal Models and the Information Bottleneck principle to explain this paradox.
Outcome:	The proposed framework bridges the framework between SCM and IB principles to explain the problem.

Copied to clipboard

Mohammadtaher Safarzadeh, Hitesh Laxmichand Patel, Afshin Oroojlooy, Graham Horwood, Dan Roth

PaTaRM: Bridging Pairwise and Pointwise Signals via Preference-Aware Task-Adaptive Reward Modeling (2026.acl-long)

Challenge:	Large language models (LLMs) have achieved strong performance on natural language to SQL (NL2SQL) benchmarks, yet their reported accuracy may be inflated by contamination from benchmark queries or structurally similar patterns seen during training.
Approach:	They propose a syntactic probing framework for detecting and quantifying such contamination in large language models.
Outcome:	The proposed framework generates syntactic variants of test queries for four widely used NL2SQL datasets.

Copied to clipboard

Ai Jian, Jingqing Ruan, Xing Ma, Dailin Li, Weipeng Zhang, Ke Zeng, Xunliang Cai

Challenge:	Existing reward models lack generative and reasoning capabilities, resulting in poor performance.
Approach:	They propose a reward-aware task-adaptive reward model that enables pointwise training using readily available pairwise data via a novel Preference-Aware Reward mechanism.
Outcome:	The proposed reward model achieves an average relative improvement of 8.7% over the base models on RewardBench and RMBench.

Uncovering Sentiment Analysis Circuit in Large Language Model (2026.acl-long)

Copied to clipboard

Shichen Li, Zhouyang Wang, Zhongqing Wang, Peifeng Li

Think Before Writing: Feature-Level Multi-Objective Optimization for Generative Citation Visibility (2026.acl-long)

Challenge:	Prior work has shown that sentiment is encoded linearly in LLM representations, but their ability to utilize this information remains fragile to prompt variations.
Approach:	They propose a simple inference-time intervention method that amplifies circuit features to compensate for insufficient activation.
Outcome:	The proposed method improves on a sentiment analysis circuit with sparse autoencoders and circuit-level analysis.

Copied to clipboard

Zikang Liu, Peilan Xu

Zero-Shot Multimodal Retrieval with Multi-Scale Contextual Representations (2026.acl-long)

Challenge:	Existing generative engine optimization approaches rely on token-level text rewriting, offering limited interpretability and weak control over the trade-off between visibility and content quality.
Approach:	They propose a feature-level, multi-objective optimization framework that abstracts webpages into interpretable structural, content, and linguistic properties.
Outcome:	The proposed framework outperforms token-level methods in citation visibility and content quality on three generative engines.

Copied to clipboard

Sourajit Saha, Tejas Gokhale

ReEfBench: Quantifying the Reasoning Efficiency of LLMs (2026.acl-long)

Challenge:	Existing approaches to multimodal information retrieval (MMIR) lack generalization across different modalities and require annotated training data.
Approach:	They propose a fine-tuning-free, two-stage MMIR approach that couples efficient candidate filtering with fine-grained multimodal re-ranking.
Outcome:	The proposed approach outperforms supervised methods on 23 datasets.

Copied to clipboard

Zhizhang Fu, Yuancheng Gu, Chenkai Hu, Hanmeng Liu, Yue Zhang

Beyond Ranking: Fine-Grained Diagnostics and Self-Improvement for MLLMs (2026.acl-long)

Challenge:	Existing methods for Chain-of-Thought evaluations do not distinguish between genuine reasoning and mere verbosity.
Approach:	They propose a framework for the non-intrusive, comprehensive process-centric evaluation of reasoning grounded in First-Order Logic.
Outcome:	The proposed framework identifies four distinct behavioral prototypes and diagnoses the failure modes.

Copied to clipboard

Mingze Xu, Zijing Zhao, Qiming Peng, Houwen Peng, Han Hu, Zhanhui Kang, Yuxing Han

Bridging Internal Consistency and External Alignment: A Causal and Dynamic Interpretability Framework for LLM Generation (2026.acl-long)

Challenge:	Current paradigms rely on holistic scoring and static leaderboards to disentangle fine-grained competencies.
Approach:	They propose a framework to shift the focus from ranking to fine-grained diagnosis.
Outcome:	The proposed framework surpasses the strongest baseline by 7.92%.

Copied to clipboard

Shuyao Xiao, Shengling Wang, Ke Chao

Beyond the Last Frame: Process-aware Evaluation for Generative Video Reasoning (2026.acl-long)

Challenge:	Existing interpretability methods focus on internal and external aspects of the model . existing explanations often focus on surface correlations or static dependencies .
Approach:	They propose a causal and dynamic interpretability framework for Large Language Models . they characterize backdoor-adjusted causal effects of generated prefix and prompt .
Outcome:	The proposed framework provides a unified causal view of internal consistency and external alignment in LLM generation dynamics.

Copied to clipboard

Yifan Li, YuKai Gu, Yingqian Min, Zikang Liu, Yifan Du, Kun Zhou, Min Yang, Xin Zhao, Minghui Qiu

THOR: A Theta-Gamma Hierarchical Oscillatory Reasoning Framework for Multi-hop QA (2026.acl-long)

Challenge:	Existing evaluation frameworks often rely on single-frame assessments, which can lead to outcome-hacking.
Approach:	They propose a process-aware evaluation paradigm that uses a hierarchical rubric to evaluate the validity of the intermediate steps and the final result.
Outcome:	The proposed model achieves POC@1.0 only about 20% and exhibits significant outcome-hacking.

Copied to clipboard

Ziyang Ling, Ronald X. Xu, Mingzhai Sun

How Should We Enhance the Safety of Large Reasoning Models: An Empirical Study (2026.acl-long)

Challenge:	Current large language models (LLMs) have shown a powerful ability of reasoning in understanding question requirement, retrieving the supporting fact and generating a precise answer.
Approach:	They propose a brain-inspired Theta-Gamma hierarchical oscillatory reasoning framework which decouples attention between global planning and local retrieval.
Outcome:	Extensive comparative experiments and specific validation experiments on multi-hop QA benchmarks show that THOR improves answer accuracy and robustness while mitigating limitations.

Copied to clipboard

Zhexin Zhang, Xian Qi Loye, Victor Shea-Jay Huang, Junxiao Yang, Qi Zhu, Shiyao Cui, Fei Mi, Lifeng Shang, Yingkang Wang, Hongning Wang, Minlie Huang

HSCodeComp: A Realistic and Expert-level Agent Benchmark for Hierarchical Rule Application (2026.acl-long)

Challenge:	Large Reasoning Models have achieved remarkable success on reasoning-intensive tasks, but their enhanced reasoning capabilities do not translate to improved safety performance.
Approach:	They propose to use supervised fine tuning to enhance the safety of Large Reasoning Models.
Outcome:	The proposed method improves the safety of large reasoning models on reasoning-intensive tasks.

Copied to clipboard

Tian Lan, Yiqian Yang, Qianghuai Jia, Li Zhu, Hui Jiang, Hang Zhu, Weihua Luo, Longyue Wang

AnchorSeg: Language Grounded Query Banks for Reasoning Segmentation (2026.acl-long)

Challenge:	Existing agent benchmarks neglect hierarchical rule application in real-world domains . a critical gap persists in numerous real-life professional domains where decision-making is governed by expert-written rules.
Approach:	They propose a benchmark requiring agents to assign a unique 10-digit Harmonized System (HS) Code to products by aligning their fuzzy attributes with strict tariff classification rules.
Outcome:	The proposed benchmarks lack hierarchical rule application capability in real-world domains . the proposed benchmark is based on e-commerce and is open-source .

Copied to clipboard

Rui Qian, Chuanhang Deng, Qiang Huang, Jian Xiong, Mingxuan Li, Yingbo Zhou, Wei Zhai, Jintao Chen, Dejing Dou

R^3AG: Retriever Routing for Retrieval-Augmented Generation (2026.acl-long)

Challenge:	Existing models rely on a single segmentation token whose hidden state implicitly encodes both semantic reasoning and spatial localization . Existing methods rely only on SEG>, which encodes semantic reasoning, limiting the model's ability to explicitly disentangle what to segment from where to segment.
Approach:	They propose a method which reformulates reasoning segmentation as a structured conditional generation process over image tokens conditioned on language grounded query banks.
Outcome:	The proposed model bridges token-level predictions and pixel-level supervision by decoupling spatial grounding from semantic reasoning through structured language grounded query banks.

Copied to clipboard

Tong Zhao, Yutao Zhu, Yucheng Tian, Zhicheng Dou

ClusterRAG: Cluster-Based Collaborative Filtering for Personalized Retrieval-Augmented Generation (2026.acl-long)

Challenge:	Retrieval-augmented generation (RAG) is often bottlenecked by the “one-size-fits-all” retrieval paradigm, as different queries exhibit distinct preferences for different retrievers.
Approach:	They propose a novel routing framework that explicitly models the dynamic alignment between queries and retriever capabilities and decomposes retriever capability into two learnable dimensions: retrieval quality and generation utility.
Outcome:	Experiments on knowledge-intensive tasks show that R3AG outperforms both the best individual retrievers and state-of-the-art static routing methods.

Copied to clipboard

Gibson Nkhata, Uttamasha Anjally Oyshi, Quan Mai, Susan Gauch

Toward Robust In-Context Learning: Leveraging Out-of-distribution Proxies for Target Inaccessible Demonstration Retrieval (2026.acl-long)

Challenge:	Existing approaches to Personalized Retrieval-Augmented Generation (RAG) ignore long-term user information and inter-user relationships when constructing retrieval contexts, limiting personalization and the ability to leverage analogous users' knowledge for improved generation quality.
Approach:	They propose a Cluster-Based Collaborative Filtering for Personalized Retrieval-Augmented Generation that organizes users into semantically coherent clusters and performs retrieval at both the cluster and document levels via cluster-level similarity and fine-grained ranking.
Outcome:	Extensive experiments on the LaMP benchmark show that ClusterRAG integrates seamlessly with different dense retrievers and rankers, and remains effective when paired with both fine-tuned and zero-shot language models.

Copied to clipboard

Hao Xu, Rite Bo, Fausto Giunchiglia, Yingji Li, Rui Song

AutoSchemaKG: Autonomous Knowledge Graph Construction through Dynamic Schema Induction from Web-Scale Corpora (2026.acl-long)

Challenge:	Large language models (LLMs) perform well on Out-of-Distribution tasks, but performance degrades as distribution shift becomes more severe.
Approach:	They propose a demonstration search framework that incorporates an OOD proxy to approximate the inaccessible target domain and guide the retrieval process.
Outcome:	The proposed framework enhances robustness in out-of-distribution tasks by incorporating an OOD proxy to approximate the inaccessible target domain and guide the retrieval process.

Copied to clipboard

Jiaxin Bai, Wei Fan, Qi Hu, Qing Zong, Chunyang Li, Hong Ting Tsang, Hongyu Luo, Yauwai Yim, Haoyu Huang, Xiao Zhou, Feng Qin, Tianshi Zheng, Xi Peng, Xin Yao, Huiwen Yang, Leijie Wu, JI Yi, Gong Zhang, Renhai Chen, Yangqiu Song

Aligned Multi-View Scripts for Universal Chart-to-Code Generation (2026.acl-long)

Challenge:	Existing knowledge graph construction frameworks require predefined schemas, limiting their scalability and domain coverage.
Approach:	They propose a framework for fully autonomous knowledge graph construction that eliminates the need for predefined schemas.
Outcome:	The proposed framework outperforms state-of-the-art models on multi-hop QA tasks and enhances LLM factuality.

Copied to clipboard

Zhihan Zhang, Lizi Liao

SpeakerSleuth: Can Large Audio-Language Models Judge Speaker Consistency across Multi-turn Dialogues? (2026.acl-long)

Challenge:	Existing methods for chart-to-code generation are largely Python-centric, limiting practical use and overlooking a critical source of supervision.
Approach:	They propose a chart-to-code generation tool that converts a graph image into an executable plotting script.
Outcome:	The proposed method outperforms existing systems and is competitive with proprietary systems.

Copied to clipboard

Jonggeun Lee, Junseong Pyo, Gyuhyeon Seo, Yohan Jo

Breaking Block Boundaries: Anchor-based History-stable Decoding for Diffusion Large Language Models (2026.acl-long)

Challenge:	Large Audio-Language Models (LALMs) are a popular approach for evaluating speech quality, yet their ability to assess speaker consistency across multi-turn dialogues remains unexplored.
Approach:	They construct 1,818 human-verified evaluation instances across four datasets spanning synthetic and real speech, with controlled acoustic difficulty.
Outcome:	The proposed model performs better in comparing and ranking acoustic variants, demonstrating inherent acustic discrimination capabilities.

Copied to clipboard

Shun Zou, Yong Wang, Zehui Chen, Lin Chen, Chongyang Tao, Feng Zhao, Xiangxiang Chu

HiSVD: Principled Low-Rank Approximation of LLMs via Hierarchical Modeling of Information Capacity and Spectral Structure (2026.acl-long)

Challenge:	Semi-autoregressive (Semi-AR) decoding suffers from inherent block constraints . naive lookahead decoding is unreliable, token stability closely correlates with convergence trend, and historical information is isolated.
Approach:	They propose a training-free, plug-and-play dynamic decoding strategy that monitors the stability of tokens in real time through dynamic anchors.
Outcome:	The proposed approach reduces decoding steps by 80% while improving performance by 3.67% on the BBH benchmark.

Copied to clipboard

Zhuo Chen, Minghao Li, Xiaoqian Ma, Siqi Fan, Xiusheng Huang, Zhang Liujie, Weihang Chen

Unveiling the Unknown: Open-Set Entity Typing via Two-Stage Generation (2026.acl-long)

Challenge:	Existing methods for generating layer importance ignore the fine-grained influence of spectral distribution shape.
Approach:	They propose a hierarchical rank allocation framework with two stages to address this gap . they propose SVD-based lowrank approximation that exploits spectral heterogeneity .
Outcome:	Experiments show that HiSVD outperforms state-of-the-art methods on LLMs .

Copied to clipboard

Hu Chen, Binhan Yang, Wei Shen

Don’t Adapt Small Language Models for Tools; Adapt Tool Schemas to the Models (2026.acl-long)

Challenge:	Existing fine-grained entity typing models are susceptible to misclassify unknown-type instances . manual collection and annotation of large unknown-Type instances is time-consuming and labor-intensive in open environments.
Approach:	They propose a novel task that uses open-set entity typing to classify entities of unknown types . they propose 'two-stage generation model' that automatically produces large-scale pseudo unknown-type instances .
Outcome:	The proposed framework surpasses baselines on two newly established benchmark datasets.

Copied to clipboard

Jonggeun Lee, Woojung Song, Jongwook Han, Haesung Pyun, Yohan Jo

Language Reconstruction with Brain Predictive Coding from fMRI Data (2026.acl-long)

Challenge:	Small language models struggle with tool-use tasks, particularly in selecting appropriate tools and identifying correct parameters.
Approach:	They propose a training-free method that leverages peakedness to align schemas with pretraining knowledge to rename tool components.
Outcome:	Experiments on MetaTool and RoTBench show that PA-Tool significantly improves tool-use accuracy without retraining.

Copied to clipboard

Congchi Yin, Ziyi Ye, Piji Li

Chaining the Evidence: Robust Reinforcement Learning for Deep Search Agents with Citation-Aware Rubric Rewards (2026.acl-long)

Challenge:	Existing studies have shown that the perception of speech can be decoded from brain signals and subsequently reconstructed as continuous language.
Approach:	They propose to use FMRI-to-text decoding with Predictive coding to generate a main network and a side network to generate brain predictive representations from related regions of interest.
Outcome:	The proposed model outperforms current decoding models on several evaluation metrics on two naturalistic language comprehension fMRI datasets.

Copied to clipboard

Jiajie Zhang, Xin Lv, Ling Feng, Lei Hou, Juanzi Li

Think Natively: Unlocking Multilingual Reasoning with Consistency-Enhanced Reinforcement Learning (2026.acl-long)

Challenge:	Existing methods for reinforcement learning (RL) rely on binary outcome rewards that fail to capture the comprehensiveness and factuality of agents’ reasoning process.
Approach:	They propose a reward framework that emphasizes reasoning comprehensiveness, factual grounding, and evidence connectivity.
Outcome:	The proposed framework outperforms standard outcome-based RL baselines across multiple deep search benchmarks and shows that it discourages shortcut exploitation and promotes comprehensive, evidence-grounded reasoning.

Copied to clipboard

Xue Zhang, Yunlong Liang, Fandong Meng, Songming Zhang, Kaiyu Huang, Yufeng Chen, Xu Jinan, Jie Zhou

Mem2Evolve: Towards Self-Evolving Agents via Co-Evolutionary Capability Expansion and Experience Distillation (2026.acl-long)

Challenge:	Current Large Reasoning Models exhibit two critical limitations when processing non-English languages: (1) They struggle to maintain input-output language consistency; (2) They generally perform poorly with wrong reasoning paths and lower answer accuracy compared to English.
Approach:	They propose a language-consistency reward and a cross-lingual thinking alignment reward to improve the model's interpretability and accuracy.
Outcome:	The proposed model achieves nearly 100% language consistency and superior performance on two multilingual benchmarks (MMATH and PolyMath).

Copied to clipboard

Zihao Cheng, Zeming Liu, Yingyu Shan, Xinyi Wang, Xiangrong Zhu, Yunpu Ma, Hongru Wang, Yuhang Guo, Wei Lin, Yunhong Wang

VideoCuRL: Video Curriculum Reinforcement Learning with Orthogonal Difficulty Decomposition (2026.acl-long)

Challenge:	Existing frameworks that focus on static tools and static assets are ineffective for self-evolving agents.
Approach:	They propose a paradigm of co-evolutionary Capability Expansion and Experience Distillation that leverages accumulated experience to guide dynamic creation of assets.
Outcome:	The proposed framework improves performance in single-task and cross-task settings by 18.53% over standard LLMs, 11.80% over agents evolving solely through experience, and 6.46% over those evolving solelly through asset creation.

Copied to clipboard

Hongbo Jin, Kuanwei Lin, Wenhao Zhang, Yichen Jin, Ge Li

Visual Inception: Compromising Long-term Planning in Agentic Recommenders via Multimodal Memory Poisoning (2026.acl-long)

Challenge:	Reinforcement Learning (RL) is crucial for Video-LLMs with complex spatiotemporal reasoning.
Approach:	They propose a framework that decomposes difficulty into two axes in video understanding . they employ efficient, training-free proxies to map data onto a 2D curriculum grid .
Outcome:	The proposed framework surpasses strong RL baselines on reasoning and perception tasks.

Copied to clipboard

Jiachen Qian

Language on Demand, Knowledge at Core: Composing LLMs with Encoder-Decoder Translation Models for Extensible Multilinguality (2026.acl-long)

Challenge:	Existing research focuses on prompt injection or immediate adversarial misclassification of user-uploaded images.
Approach:	They propose a dual-process defense framework inspired by human cognition to mitigate this vulnerability by injecting triggers into user-uploaded images that act as "sleeper agents"
Outcome:	The proposed framework achieves about 85% Goal-Hit Rate (GHR) while reducing the risk to 10% with configurable latency trade-offs.

Copied to clipboard

Mengyu Bu, Yang Feng

GeoArena: Evaluating Open-World Geographic Reasoning in Large Vision-Language Models (2026.acl-long)

Challenge:	Large language models exhibit strong general intelligence, yet their multilingual performance remains imbalanced.
Approach:	They propose a compositional encoder-LLM-decoder architecture that offloads multilingual understanding to external pretrained translation models while preserving the LLM as an English-centric core for general knowledge processing and reasoning.
Outcome:	The proposed architecture outperforms baseline models on four large language models across understanding, reasoning, and generation.

Copied to clipboard

Pengyue Jia, Yingyi Zhang, Xiangyu Zhao, Sharon Li

Fisher-Driven Adaptive Locating for Knowledge Editing in Large Language Models (2026.acl-long)

Challenge:	Existing evaluation paradigms for geographic reasoning are outcome-centric and focus on label matching, leaving the underlying linguistic reasoning chains as unexamined black boxes.
Approach:	They propose a dynamic, human-preference-based evaluation framework for benchmarking open-world geographic reasoning.
Outcome:	The proposed framework reframes evaluation as a pairwise reasoning alignment task on in-the-wild images, where human judges compare model-generated explanations based on reasoning quality, evidence synthesis, and plausibility.

Copied to clipboard

Chenghao Xu, Jiexi Yan, Guangtao Lyu, Qi Liu, Muli Yang, Cheng Deng

From AR to Diffusion: Efficiently Adapting Large Language Models with Strictly Causal and Elastic Horizons (2026.acl-long)

Challenge:	Existing methods for locating and editing static knowledge are costly and risk catastrophic forgetting or error.
Approach:	They propose a Fisher-driven adaptation-aware locating strategy that dynamically identifies which model components should be edited for a given knowledge update.
Outcome:	Experiments on standard benchmarks show that FiDAL improves editing effectiveness and knowledge preservation across multiple editing methods.

Copied to clipboard

Xiangyu Ma, Teng Xiao, Zuchao Li, Lefei Zhang

ExPerT: Personalizing LLM Responses to Users’ Domain Expertise via Query-Wise Semantic and Keystroke Behavioral Cues (2026.acl-long)

Challenge:	Autoregressive (AR) models rely on bidirectional attention, creating a structural mismatch with pre-trained Autoregression models.
Approach:	They propose a framework that efficiently adapts autoregressive (AR) models to the diffusion paradigm.
Outcome:	The proposed framework reduces training costs by orders of magnitude while maintaining state-of-the-art performance.

Copied to clipboard

Yeji Park, Jiwon Tark, Taesik Gong

Challenge:	Existing personalization methods relying on static profiles or text-only signals fail to capture query-specific expertise variation.
Approach:	They propose a query-wise personalization framework that adapts LLM responses to query domain expertise by combining semantic and behavioral cues.
Outcome:	ExPerT reduces expertise inference error by 65.7% compared to the strongest baseline and improves response satisfaction by 17.52% on a 5-point Likert scale.

Measure Twice, Click Once: Co-evolving Proposer and Visual Critic via Reinforcement Learning for GUI Grounding (2026.acl-long)

Copied to clipboard

Wenkai Wang, Xiyun Li, Hongcan Guo, Wenhao Yu, Tianqing Fang, Haitao Mi, Dong Yu, Shengyu Zhang

Retrieving to Recover: Towards Incomplete Audio-Visual Question Answering via Semantic-consistent Purification (2026.acl-long)

Challenge:	Graphical User Interface (GUI) grounding requires mapping natural language instructions to precise pixel coordinates due to visually homogeneous elements and dense layouts.
Approach:	They propose to replace static consistency strategies with a learnable selection mechanism that selects the optimal target by critiquing its own proposals rendered on the screenshot.
Outcome:	The proposed model significantly improves both grounding and critiquing capabilities over 6 benchmarks.

Copied to clipboard

Jiayu Zhang, Shuo Ye, Qilang Ye, Zihan Song, Jiajian Huang, Zitong YU

ChatHLS: Towards Systematic Design Automation and Optimization for High-Level Synthesis (2026.acl-long)

Challenge:	Recent audio-visual question answering methods lack effective mechanisms for handling missing modalities, leading to performance degradation in real-world scenarios with data interruptions.
Approach:	They propose a framework that shifts the paradigm of missing modality handling to retrieval-based recovery . they leverage cross-modal retrieval via unified semantic embeddings to acquire missing domain-specific knowledge.
Outcome:	The proposed framework improves AVQA and enhances robustness in modal-incomplete scenarios.

Copied to clipboard

Runkai Li, Jia Xiong, Xiuyuan He, Jieru Zhao, Jiaqi Lv, Haowen Fang, Lei Qi, Xi Wang

Cultural Benchmarking of LLMs in Standard and Dialectal Arabic Dialogues (2026.acl-long)

Challenge:	High-Level Synthesis (HLS) is a hardware design tool that can be used to design hardware from C-like languages, but its widespread adoption is limited by strict coding constraints and design-specific optimizations.
Approach:	They propose a multi-agent HLS design framework that leverages specialized LLMs for automated debugging and directive tuning.
Outcome:	The proposed framework outperforms Gemini-3-pro in debugging and speedups across various HLS kernels and neural network accelerators.

Copied to clipboard

Muhammad Dehan Al Kautsar, Saeed Almheiri, Momina Ahsan, Bilal Elbouardi, Younes Samih, Sarfraz Ahmad, Amr Keleg, Omar El Herraoui, Kareem Elzeky, Abed Alhakim Freihat, Mohamed Anwar, Zhuohan Xie, Junhong Liang, Mohammad Rustom Al Nasar, Preslav Nakov, Fajri Koto

ReRec: Reasoning-Augmented LLM-based Recommendation Assistant via Reinforcement Fine-tuning (2026.acl-long)

Challenge:	Most benchmarks focus on short text snippets in Modern Standard Arabic (MSA), overlooking cultural nuances that naturally arise in dialogues.
Approach:	They propose a culturally grounded conversational dataset covering 13 Arabic-speaking countries, in both Modern Standard Arabic (MSA) and each country’s respective dialect, spanning 12 daily-life topics and 54 fine-grained subtopics.
Outcome:	The proposed model performs worse on all three tasks than the MSA benchmark.

Copied to clipboard

Jiani Huang, Shijie Wang, Liang-Bo Ning, Wenqi Fan, Li Qing

MoA: Heterogeneous Mixture of Adapters for Parameter-Efficient Fine-Tuning of Large Language Models (2026.acl-long)

Challenge:	Existing reasoning-augmented systems that handle complex queries are lacking . we present a framework that enhances LLM-based recommendation assistants .
Approach:	They propose a reinforcement fine-tuning framework that enhances LLM-based recommendation . they use a dual-graph Enhanced Reward Shaping framework to integrate recommendation metrics .
Outcome:	The proposed framework outperforms state-of-the-art recommendations and preserves core abilities.

Copied to clipboard

Jie Cao, Tianwei Lin, Bo Yuan, Rolan Yan, Hongyang He, Wenqiao Zhang, Juncheng Li, Dongping Zhang, Siliang Tang, Yueting Zhuang

Beyond the Context Window: Scaling Agentic RL via End-to-end Optimized Context Compression (2026.acl-long)

Challenge:	Existing methods for parameter-efficient fine-tuning (PEFT) are limited by computational costs and performance degradation.
Approach:	They propose a method that integrates Low-Rank Adaptation and Mixture-of-Experts (MoE) they propose combining expert load imbalance and representation collapse to improve LLM performance .
Outcome:	The proposed method outperforms homogeneous MoE-LoRA architectures in performance and parameter efficiency.

Copied to clipboard

Miao Lu, Weiwei Sun, Weihua Du, Zhan Ling, Xuesong Yao, Kang Liu, Jiecao Chen

Summarization

Challenge:	Existing reinforcement learning pipelines suffer from degraded instruction following, excessive rollout costs, and strict context limits.
Approach:	They propose a reinforcement learning (RL) fine-tuning of large language model (LLM) agents for long-horizon multi-turn tool use where context length quickly becomes a bottleneck.
Outcome:	The proposed framework improves the success rate while maintaining the same or even lower working context length compared to baselines.

Probing the Safety Robustness of LLMs in Latent Space (2026.acl-long)

Copied to clipboard

Tianle Gu, Kexin Huang, Zongqi Wang, Yixu Wang, Jie Li, Xin Wang, Yang Yao, Yujiu Yang, Yan Teng, Yingchun Wang

LoopTool: Closing the Data–Training Loop for Robust LLM Tool Calls (2026.acl-long)

Challenge:	Despite substantial progress in safety alignment techniques, aligned large language models can still produce unsafe responses under minor internal perturbations.
Approach:	They introduce Activation Steering Attack (ASA) and leverage the Negative Log-Likelihood (NLL) as a diagnostic signal to probe the local sensitivity of safety behaviors in latent space.
Outcome:	The proposed method is model-agnostic and supervision-free, enabling a general and reproducible diagnostic metric for analyzing safety robustness.

Copied to clipboard

Kangning Zhang, Weiwen Liu, Wenxiang Jiao, Kounianhua Du, Yuan Lu, Weinan Zhang, Yong Yu

Placing Puzzle Pieces Where They Matter: A Question Augmentation Framework for Reinforcement Learning (2026.acl-long)

Challenge:	Large Language Models (LLMs) are powerful tools for multi-step tasks, but static data pipelines hinder tool learning and cause noisy labels to persist.
Approach:	They propose a fully automated, model-aware data evolution framework that tightly integrates data synthesis and model training.
Outcome:	Experiments show that LoopTool-8B significantly surpasses its 32B data generator and achieves new state-of-the-art results on the BFCL-v3 and ACEBench benchmarks for its scale.

Copied to clipboard

Yangyi Fang, Haolin Shi

USB: A COMPREHENSIVE AND UNIFIED SAFETY EVALUATION BENCHMARK FOR MULTIMODAL LARGE LANGUAGE MODELS (2026.acl-long)

Challenge:	Reinforcement learning (RL) training on easy problems can cause overfitting and pass@k degradation, while training on hard problems yields sparse reward signals.
Approach:	They propose a hint injection framework that strategically identifies and provides critical reasoning steps during training.
Outcome:	Experiments on six mathematical reasoning benchmarks show that the proposed framework achieves comparable average performance to 32B baselines while preserving pass@k diversity across all k values.

Copied to clipboard

Baolin Zheng, Guanlin Chen, Qingyang Teng, Hongqiong Zhong, Yingshui Tan, Zhendong Liu, Weixun Wang, Jiaheng Liu, Jian Yang, Huiyun Jing, Jincheng Wei, Wenbo Su, Xiaoyong Zhu, Bo Zheng, Kaifu Zhang

FlowSearch: Advancing Deep Research with Dynamic Structured Knowledge Flow (2026.acl-long)

Challenge:	Existing safety benchmarks fail to provide reliable assessments due to limited risk coverage, insufficient scale and the oversight of complex modality combinations.
Approach:	They propose a framework that covers 61 risk categories across four modality interactions to address this gap.
Outcome:	The proposed framework covers 61 risk categories across four distinct modality interactions.

Copied to clipboard

Yusong Hu, Runmin Ma, Yue Fan, Jinxin Shi, Zongsheng Cao, Yuhao Zhou, Jiakang Yuan, Shuaiyu Zhang, Shiyang Feng, Xiangchao Yan, Shufei Zhang, Wenlong Zhang, Lei Bai, Bo Zhang

LLMEval-Fair: A Large-Scale Longitudinal Study on Robust and Fair Evaluation of Large Language Models (2026.acl-long)

Challenge:	FlowSearch is a multi-agent framework that actively constructs and evolves a dynamic structured knowledge flow to drive subtask execution and reasoning.
Approach:	They propose a multi-agent framework that actively constructs and evolves a dynamic structured knowledge flow to drive subtask execution and reasoning.
Outcome:	The proposed framework achieves competitive performance on GAIA, HLE, GPQA and TRQA benchmarks and is available to download.

Copied to clipboard

Ming Zhang, Yujiong Shen, Jingyi Deng, Yuhui Wang, Huayu Sha, Kexin Tan, Qiyuan Peng, Yue Zhang, Junzhe Wang, Shichun Liu, Yueyuan Huang, Jingqi Tong, Changhao Jiang, Yilong Wu, Zhihao Zhang, Mingqi Wu, Mingxu Chai, Zhiheng Xi, Shihan Dou, Tao Gui, Qi Zhang, Xuanjing Huang

SAME: Signer-Aware Mixture-of-Experts for Test-Time Adaptation in Sign Language Translation (2026.acl-long)

Challenge:	Existing evaluation of Large Language Models on static benchmarks is vulnerable to data contamination and leaderboard overfitting.
Approach:	LLMEval-Fair framework provides a framework for dynamic evaluation of Large Language Models . evaluators use a proprietary bank of 220k graduate-level questions to analyze model data .
Outcome:	LLMEval-Fair provides robust and credible evaluation framework for Large Language Models . it provides a strong empirical validation for the dynamic evaluation paradigm .

Copied to clipboard

Lujia Yang, Weicai Yan, Yongbo He, Qifei Zhang, Tao Jin, Jinshan Zhang, Meng Xi, Jianwei Yin

Controllable LLM Reasoning via Sparse Autoencoder-Based Steering (2026.acl-long)

Challenge:	Existing methods for supervised fine-tuning are limited due to labeled data . existing methods require long adaptation times and batch statistics are unavailable in streaming settings .
Approach:	They propose a plug-and-play, signer-aware Mixture-of-Experts (MoE) TTA architecture for SLT . they use a combination of lightweight MoE modules and unsupervised regularizers to decouple domain shift .
Outcome:	The proposed test-time adaptation outperforms existing TTA methods in sign language translation . the proposed architecture can be used in real-world deployments without labeling .

Copied to clipboard

Yi Fang, Wenjie Wang, Mingfeng Xue, Boyi Deng, Fengli Xu, Dayiheng Liu, Fuli Feng

LEDOM: Reverse Language Model (2026.acl-long)

Challenge:	Existing methods struggle to control fine-grained reasoning strategies due to conceptual entanglement in LRMs’ hidden states.
Approach:	They propose to decompose strategy-entangled hidden states into a disentangled feature space by using Sparse Autoencoders to identify the few strategy-specific features from the vast pool of SAE features.
Outcome:	The proposed method outperforms existing methods by 15% in control effectiveness.

Copied to clipboard

Xunjian Yin, Sitao Cheng, Yuxi Xie, Xinyu Hu, Li Lin, Xinyi Wang, Liangming Pan, William Yang Wang, Xiaojun Wan

AEGIS: A Holistic Benchmark for Evaluating Forensic Analysis of AI-Generated Academic Images (2026.acl-long)

Challenge:	Autoregressive language models are trained exclusively left-to-right, yet they are limited in their ability to factorize text.
Approach:	They propose a purely reverse autoregressive language model that factorizes text as a product of left-to-right conditionals.
Outcome:	The proposed model can be used to score forward outputs using reverse posterior estimates.

Copied to clipboard

Bo Zhang, Tzu-Yen Ma, Zichen Tang, Junpeng Ding, Zirui Wang, Yizhuo Zhao, Peilin Gao, Zijie Xi, Zixin Ding, Haiyang Sun, Haocheng Gao, Yuan Liu, Liangjia Wang, Yiling Huang, Yujie Wang, Yuyue Zhang, Ronghui Xi, Yuanze Li, Jiacheng Liu, Zhongjun Yang, Haihong E

The GaoYao Benchmark: A Comprehensive Framework for Evaluating Multilingual and Multicultural Abilities of Large Language Models (2026.acl-long)

Challenge:	AEGIS examines whether current models can effectively audit AI-generated images in academic papers.
Approach:	They propose a holistic benchmark for forensic analysis of AI-Generated academic ImageS that reveals limitations in academic image forensics.
Outcome:	AEGIS compared with existing benchmarks on seven academic categories and features key advances in forensic analysis.

Copied to clipboard

Yilun Liu, Chunguang Zhao, Mengyao Piao, Lingqi Miao, Shimin Tao, Minggui HE, Chenxin Liu, Zhang Li, null Mahongxia, Jiaxin Guo, Chen Liu, Liqun Deng, Jiansheng Wei, Xiaojun Meng, Fanyi Du, Daimeng Wei, Yanghua Xiao

METRO: Towards Strategy Induction from Expert Dialogue Transcripts for Non-collaborative Dialogues (2026.acl-long)

Challenge:	Existing multilingual evaluation benchmarks neglect cultural nuances and lack language coverage in subjective tasks.
Approach:	They propose a framework that categorizes evaluation tasks into three cultural layers and nine cognitive sub-layers.
Outcome:	The proposed framework surpasses prior coverage by up to 111% on 20+ LLMs.

Copied to clipboard

Haofu Yang, Jiaji Liu, Chen Huang, Faguo Wu, Wenqiang Lei, See-Kiong Ng

TamEdit: Trajectory-Aware Meta-Learning for Specificity-Preserving Continual Knowledge Editing (2026.acl-long)

Challenge:	Developing non-collaborative dialogue agents traditionally requires manual codification of expert strategies.
Approach:	They propose a method that formalizes expert knowledge into a Strategy Forest from raw transcripts.
Outcome:	The proposed method outperforms existing methods by 9%-10% in two benchmarks.

Copied to clipboard

Shiqiang Tian, Cheng Ding, Qin Chen, Jie Zhou, Liang He

LexRel: Benchmarking Legal Relation Extraction for Chinese Civil Cases (2026.acl-long)

Challenge:	Existing methods for continual knowledge editing focus on single edits or preventing knowledge forgetting.
Approach:	They propose a meta-learning method that preserves specificity for continual knowledge editing by capturing relationships between different single edits within the trajectory.
Outcome:	Experiments show that TamEdit outperforms baselines in continual editing while preserving general capabilities.

Copied to clipboard

Yida Cai, Ranjuexiao Hu, Huiyuan Xie, Chenyang Li, Yun Liu, Yuxiao Ye, Zhenghao Liu, Weixing Shen, Zhiyuan Liu

Agentic Memory: Learning Unified Long-Term and Short-Term Memory Management for Large Language Model Agents (2026.acl-long)

Challenge:	Legal relations are an important analytical framework for dispute resolution in civil cases.
Approach:	They propose a comprehensive schema for legal relations in civil cases with hierarchical taxonomy and definitions of arguments.
Outcome:	The proposed schema shows that existing LLMs lack the ability to identify civil legal relations and performance improves on downstream tasks.

Copied to clipboard

Yi Yu, Liuyi Yao, Yuexiang Xie, Qingquan Tan, Jiaqi Feng, Yaliang Li, Libing Wu

Profiling-Free Mixed-Precision Quantization for MoE LLMs via Fuzzy Rule Interpolation (2026.acl-long)

Challenge:	Existing methods handle long-term memory (LTM) and short-term (STM) as separate components, relying on heuristics or auxiliary controllers, which limits adaptability and end-to-end optimization.
Approach:	They propose a framework that integrates LTM and STM management directly into the agent's policy and propose 'agentic memory' to train such unified behaviors.
Outcome:	The proposed framework outperforms strong memory-augmented baselines on five long-horizon benchmarks and achieves higher-quality long-term memory and more efficient context usage.

Copied to clipboard

Huachen Qi, Ruiyu Zhuo, Bowen Shi, Xiang Chang, Fei Chao, Changjing Shang, Qiang Shen

Tailoring Diagnostic Modeling to Individual Learners: Personalized Distractor Generation via MCTS-Guided Reasoning Reconstruction (2026.acl-long)

Challenge:	Large Language Models are scaling in size and capability, driving substantial computational and memory costs.
Approach:	They propose a mixed-precision quantization framework that uses fuzzy rule interpolation to predict quantization error from only sparse samples.
Outcome:	The proposed framework accelerates the profiling phase by up to 15.7 on DeepSeek-V2 while achieving comparable or slightly superior zero-shot accuracy.

Copied to clipboard

Tao Wu, Jingyuan Chen, Wang Lin, Jian Zhan, Mengze Li, Fangzhou Jin, Min Zhang, Kun Kuang, Fei Wu

MTR-Bench: A Comprehensive Benchmark for Multi-Turn Reasoning Evaluation (2026.acl-long)

Challenge:	Current distractor generation methods produce shared distractors for all students, ignoring individual variations in reasoning, which limits their diagnostic effectiveness.
Approach:	They propose a method which tailors distractors to each student’s specific cognitive flaws, inferred from their past question-answering (QA) history.
Outcome:	The proposed framework outperforms existing methods in generating plausible distractors and adapts to group-level settings.

Copied to clipboard

Xiaoyuan Li, Keqin Bao, Yubo Ma, Moxin Li, Wenjie Wang, Rui Men, Yichang Zhang, Fuli Feng, Dayiheng Liu

Community-Aware Assessment of Social Textual Engagement and Resonance: A Human-Centric Perspective on User-Generated Content Evaluation (2026.acl-long)

Challenge:	Recent advances in Large Language Models (LLMs) have shown promising results in complex reasoning tasks.
Approach:	They propose to use a multi-turn reasoning evaluation framework to cover multi-turn interactions with the environments of large language models.
Outcome:	The proposed framework covers diverse reasoning capabilities, fine-grained difficulty granularity, and necessitates multi-turn interactions with the environments.

Copied to clipboard

Tianjiao Li, Kai Zhao, Xiang Li, Yang Liu, Huyang Sun

Query-Aware Knowledge Retrieval via Hyperbolic Structuring (2026.acl-long)

Challenge:	Traditional Video Quality Assessment (VQA) focuses on aesthetic fidelity and technical distortions.
Approach:	They propose a new task that evaluates whether a UGC item has positive community resonance based on multimodal attributes rather than visual quality alone.
Outcome:	The proposed task outperforms state-of-the-art baselines on CASTER-Bench . it provides interpretable and empathetic reasoning paths that align with real community feedback.

Copied to clipboard

Chuang Zhou, Junnan Dong, Yilin Xiao, Shengyuan Chen, Su Dong, di Yin, Xing Sun, Zhaozhuo Xu, Xiao Huang

EduBench: A Comprehensive Benchmarking Dataset for Evaluating Large Language Models in Diverse Educational Scenarios (2026.acl-long)

Challenge:	Existing approaches focus primarily on retrieving isolated factual knowledge entities while neglecting the critical reasoning relationships.
Approach:	They propose a query-centric retrieval framework that explicitly integrates structured knowledge graphs to support complex reasoning tasks.
Outcome:	Extensive experiments on three benchmark datasets show that HyperRAG outperforms baselines.

Copied to clipboard

Bin Xu, Yu Bai, Huashan Sun, Yiguan Lin, Siming Liu, Xinyue Liang, Yaolin Li, Zhuangzhi Dong, Jingren Zhang, Yufan Deng, Xinyu Zou, Yang Gao, Heyan Huang

WebClipper: Efficient Evolution of Web Agents with Graph-based Trajectory Pruning (2026.acl-long)

Challenge:	Existing benchmarks that focus on knowledge-intensive tasks do not reflect diverse educational scenarios.
Approach:	They propose a benchmark that incorporates 9 major scenarios and 4,000 educational contexts.
Outcome:	The proposed model performs comparable to state-of-the-art large models on the test set.

Copied to clipboard

Junjie Wang, Zequn Xie, Dan Yang, Jie Feng, Yue Shen, Duolin Sun, Meixiu Long, Yihan Jiao, Zhehao Tan, Jian Wang, Peng Wei, Jinjie Gu

MMAC: A Multilingual, Multimodal Alignment Framework for Cultural Grounding Evaluation (2026.acl-long)

Challenge:	Open-source web agents rely on long tool-call trajectories with cyclic reasoning loops and exploration of unproductive branches.
Approach:	They propose a framework that compresses web agent trajectories via graph-based pruning.
Outcome:	The proposed framework reduces tool-call rounds by 20% while improving accuracy and efficiency while maintaining the same level of performance as existing models.

Copied to clipboard

Weihua Zheng, Zhengyuan Liu, Tanmoy Chakraborty, Weiwen Xu, Xiaoxue Gao, Bryan Chen Zhengyu Tan, Bowei Zou, Chang Liu, Yujia Hu, Xing Xie, Xiaoyuan Yi, Jing Yao, Chaojun Wang, Long Li, Rui Liu, Huiyao Liu, Koji Inoue, Ryuichi Sumida, Tatsuya Kawahara, Fan Xu, Lingyu Ye, Wei Tian, Dongjun Kim, Jimin Jung, Jaehyung Seo, Nadya Yuki Wangsajaya, Pham Minh Duc, Ojasva Saxena, Palash Nandi, Xiyan Tao, Wiwik Karlina, Tuan Luong, Keertana Arun Vasan, Roy Ka-Wei Lee, Nancy F. Chen

Step-GRPO: Internalizing Dynamic Early Exit for Efficient Reasoning (2026.acl-long)

Challenge:	Existing models lack cultural alignment across modalities and languages . a new framework to assess cultural awareness across linguistics and languages is needed .
Approach:	They propose a framework that integrates tri-modally aligned cultural benchmarks and a five-dimensional evaluation protocol to assess cross-country awareness disparities.
Outcome:	The proposed framework assesses cultural awareness disparities across modalities and languages . it is the first dataset aligned at the input level across text, image, and speech .

Copied to clipboard

Benteng Chen, Weida Wang, Shufei Zhang, Mingbao Lin, Min Zhang

Gold-Medal-Level Olympiad Geometry Solving with Efficient Heuristic Auxiliary Constructions (2026.acl-long)

Challenge:	Large reasoning models that use long chain-of-thought excel at problem-solving but waste computational resources.
Approach:	They propose a framework that internalizes dynamic early-exit capabilities directly into the model.
Outcome:	The proposed framework reduces token consumption by 32.0% on a Qwen3-8B model compared to the vanilla model .

Copied to clipboard

Boyan Duan, Xiao Liang, Shuai Lu, Yaoxiang Wang, Yelong Shen, Kai-Wei Chang, Ying Nian Wu, Mao Yang, Weizhu Chen, Yeyun Gong

Rethinking Jailbreak Detection of Large Vision Language Models with Representational Contrastive Scoring (2026.acl-long)

Challenge:	Existing methods for geometry theorem proving in Euclidean geometry are challenging and require a neural network to perform.
Approach:	They propose a method for adding auxiliary points in geometry that runs on CPUs without relying on neural network-based inference.
Outcome:	The proposed method achieves silver-medal-level human performance on IMO-30 benchmark.

Copied to clipboard

Peichun Hua, Hao Li, Shanghao Shi, Zhiyuan Yu, Ning Zhang

The Digital Dunning-Kruger Effect: Decoupling Hallucinations via Geometric Hidden-state Observation for Semantic Truthfulness (2026.acl-long)

Challenge:	Large Vision-Language Models (LVLMs) are vulnerable to a growing array of multimodal jailbreak attacks, necessitating a generalizable defense that is efficient for practical deployment.
Approach:	They propose a framework that uses a lightweight projection to separate benign and malicious inputs in safety-critical layers.
Outcome:	The proposed framework enables a simple yet powerful contrastive score that differentiates true malicious intent from mere distribution shift.

Copied to clipboard

Yueheng Mao, Min Yu, Gengwang Li, Jianguo Jiang, Gang Li, Meng Zhang, Zhen Xu, Weiqing Huang, Ming Liu

d-TreeRPO: Towards More Reliable Policy Optimization for Diffusion Language Models (2026.acl-long)

Challenge:	Large Language Models (LLMs) often generate overconfident yet factually incorrect hallucinations.
Approach:	They propose a black-box-based framework that captures stubborn hallucinations by integrating internal geometric dynamics with output probability distributions.
Outcome:	The proposed framework outperforms white-box methods and reduces computational overhead by over 90%.

Copied to clipboard

Leyi Pan, Shuchang Tao, Yunpeng Zhai, Zheyu Fu, Liancheng Fang, Minghua He, Lingzhe Zhang, Zhaoyang Liu, Bolin Ding, Aiwei Liu, Lijie Wen

Traffic-R1: Reinforced LLMs Bring Human-Like Reasoning to Traffic Signal Control Systems (2026.acl-long)

Challenge:	Existing RL methods suffer from reliability bottlenecks due to reward sparsity and intractable computations . d-TreeRPO provides fine-grained and verifiable step-wise reward signals .
Approach:	They propose a reliable reinforcement learning framework for diffusion large language models that leverages tree-structured rollouts and bottom-up advantage computation based on verifiable outcome rewards.
Outcome:	The proposed framework outperforms baseline models and achieves significant improvements across reasoning benchmarks.

Copied to clipboard

Xingchen Zou, Yuhao Yang, Zheng Chen, Xixuan Hao, Yiqi Chen, Chao Huang, Yuxuan Liang

Nirvana: A Specialized Generalist Model With Task-Aware Memory Mechanism (2026.acl-long)

Challenge:	Rapid urbanization and surging vehicle ownership intensify congestion . rapid urbanization drives crash rates, slow emergency response, and burden transit-poor communities .
Approach:	They introduce a 3B-parameter foundation model with human-like reasoning for Traffic signal control (TSC) they use reinforcement learning and network communication to convert LLM into a traffic-control model that operates like a human traffic agent.
Outcome:	The proposed model outperforms baselines and training-intensive RL controllers on a simulated traffic environment and reduces queue lengths by more than 5%.

Copied to clipboard

Yuhua Jiang, Shuang Cheng, Yihao Liu, Ermo Hua, Che Jiang, Weigao Sun, Yu Cheng, Feifei Gao, Biqing Qi, Bowen Zhou

BoYaEval: Evaluating Multimodal Large Language Models on Understanding Ancient Chinese Musical Scores (2026.acl-long)

Challenge:	Existing models lack task-guided specialized memory mechanisms . specialized generalist models excel at general language tasks but struggle in specialized domains.
Approach:	They propose a specialized generalist model with specialized memory and updater that can optimize for specialized domains.
Outcome:	The proposed model matches or surpasses baselines on general benchmarks and achieves lowest perplexity across specialized domains.

Copied to clipboard

Jiajia Li, Weizhi Xue, Yao Yao, Qiwei Li, null Chenchong, Zuchao Li, Ping Wang, Hai Zhao

Rectifying the Emotional Flow: Aligning Priors and Dynamic Guidance for High-Arousal Text-to-Speech (2026.acl-long)

Challenge:	Multimodal Large Language Models excel in general tasks but struggle with specialized, structured cultural symbols.
Approach:	They evaluate 21 leading MLLMs and compare their performance to a benchmark for Ancient Chinese musical notation.
Outcome:	The benchmark evaluates 21 leading MLLMs on five types of ancient Chinese music notation systems.

Copied to clipboard

Fangming Feng, Dongjie Fu, Zequn Xie, Yu Zhang, Yangyang Wu, Zhou Zhao, Tao Jin

Perplexity-Aware Data Scaling Law: Perplexity Landscapes Predict Performance for Continual Pre-training (2026.acl-long)

Challenge:	Existing systems suffer from linguistic collapse when pursuing high intensity or fail to meet target emotional levels.
Approach:	They propose an inference framework that introduces a neutral prosody bias and a uniform Classifier-Free Guidance that distorts the acoustic manifold, leading to artifacts.
Outcome:	The proposed framework achieves superior linguistic accuracy and expressiveness without model retraining.

Copied to clipboard

Lei Liu, Hao Zhu, Xiaoyan Yang, Yue Shen, Zhixuan Chu, Jian Wang, Jinjie Gu, Kui Ren

Benchmarking Large Vision-Language Models on CFMME: A Comprehensive Chinese Financial Multimodal Evaluation Dataset (2026.acl-long)

Challenge:	Large language models (LLMs) have impressive capabilities across a wide range of domains, but their generalpurpose pre-training objectives often leave them illsuited for specialized applications such as healthcare.
Approach:	They propose a perplexity-aware data scaling law that establishes a predictive relationship between the perplexities of domain-specific data and the test loss.
Outcome:	Experiments on medical and general-domain benchmarks show that the proposed scaling law consistently identifies near-optimal training subsets with significantly reduced data consumption.

Copied to clipboard

Qian Chen, Xianyin Zhang, Yanzhi Liu, Lifan Guo, Feng Chen, Chi Zhang

AutoReproduce: Automatic AI Experiment Reproduction with Paper Lineage (2026.acl-long)

Challenge:	Large Vision-Language Models (LVLMs) have expanded capabilities beyond text understanding . a novel Chinese financial multimodal evaluation benchmark is used to evaluate LVLM capabilities .
Approach:	They propose a Chinese financial multimodal evaluation benchmark to evaluate LVLMs' capabilities . the model has an overall accuracy of 66.11% and an average score of 77.18 .
Outcome:	The proposed model achieves an overall accuracy of 66.11% on the question answering task and an average score of 77.18 on detection, recognition, and information extraction tasks.

Copied to clipboard

Xuanle Zhao, Zilin Sang, Yuxuan Li, Qi Shi, Weilun Zhao, Shuo Wang, Duzhen Zhang, Xu Han, Zhiyuan Liu, Maosong Sun

Challenge:	Efficient reproduction of research papers requires deep domain expertise.
Approach:	They propose a framework that systematically mines implicit knowledge from the cited literature to reproduce experimental code in a complete, end-to-end manner.
Outcome:	The proposed framework surpasses baselines across all metrics and reproduces experimental code in a complete, end-to-end manner.

How Can Synthetic Data Improve Multilingual Language Model Pretraining? A Data Quality Perspective (2026.acl-long)

Copied to clipboard

Tongyao Zhu, Qian Liu, Chang Ma, Jinghan Zhang, Longxu Dou, Junxian He, Shiqi Chen

Can Large Language Models Infer Causal Relationships from Real-World Text? (2026.acl-long)

Challenge:	Low-resource languages are a long-tail problem for multilingual LLMs due to limited high-quality training data.
Approach:	They propose a method that translates high-quality, knowledge-rich English data into low-resource languages . they propose SynRank, which leverages synthetic data as positive samples to train a classifier .
Outcome:	The proposed method matches handcrafted rule-based filtering by human experts and significantly improves knowledge-intensive tasks with less data.

Copied to clipboard

Ryan Saklad, Aman Chadha, Oleg V. Pavlov, Raha Moraffah

Distributional Clarity: The Hidden Driver of RL-Friendliness in Large Language Models (2026.acl-long)

Challenge:	Existing work evaluating large language models relies on synthetic or simplified texts with explicit causal relationships.
Approach:	They develop a benchmark to evaluate LLMs' ability to infer causal relationships from texts . they use a dataset of texts with different levels of explicitness and complexity .
Outcome:	The proposed benchmark is the first-ever real-world dataset for this task.

Copied to clipboard

Shaoning Sun, Mingzhu Cai, Huang He, Bingjin Chen, Siqi Bao, Yujiu Yang, Hua Wu, Haifeng Wang

ReasonRank: Empowering Passage Ranking with Strong Reasoning Ability (2026.acl-long)

Challenge:	RL-friendly models exhibit intra-class compactness and inter-class separation in probability assignments . under identical training, Qwen models achieve substantial gains, while others like Llama yield limited improvements.
Approach:	They propose a method to quantify distributional clarity in probability space . they show distributional clearness is a trainable property underlying RL-Friendliness .
Outcome:	The proposed model families achieve substantial gains under identical training, while others like Llama yield limited improvements.

Copied to clipboard

Wenhan Liu, Xinyu Ma, Weiwei Sun, Yutao Zhu, Yuchen Li, Dawei Yin, Zhicheng Dou

ATIR: Towards Audio-Text Interleaved Contextual Retrieval (2026.acl-long)

Challenge:	Existing rerankers perform poorly in complex ranking scenarios due to the scarcity of reasoning-intensive training data.
Approach:	They propose an automated reasoning-intensive training framework which generates high-quality training labels from training queries and passages.
Outcome:	The proposed model outperforms baselines significantly and achieves much lower latency than the pointwise reranker.

Copied to clipboard

Tong Zhao, Chenghao Zhang, Yutao Zhu, Zhicheng Dou

Task-Aware LLM Routing with Multi-Level Task-Profile-Guided Data Synthesis for Cold-Start Scenarios (2026.acl-long)

Challenge:	Recent multimodal information retrieval research has focused on images, largely overlooking audio.
Approach:	They propose an audio-text interleaved contextual retrieval task where queries can alternate between audio and text modalities.
Outcome:	The proposed model significantly improves over baselines.

Copied to clipboard

Hui Liu, Bin Zou, Kecheng Chen, Jie Liu, Wenya Wang, Haoliang Li

Challenge:	Existing routers generalize poorly in cold-start scenarios where in-domain training data is unavailable.
Approach:	They propose a task-type–aware router approach that models query-conditioned cost and performance via latent task-like variables with prior regularization derived from the synthesized task taxonomy.
Outcome:	The proposed framework improves performance and cost under cold-start and in-domain settings and enables efficient routing.

SelFusion: Self-distillation for Diffusion Language Models (2026.acl-long)

Copied to clipboard

Hyeongsoo Lim, Jinyoung Kim, Eun Seo Seo, Min Ho Jang, Ji Won Yoon

Benchmarking Web Agent Safety under E-commerce Deceptive Interfaces (2026.acl-long)

Challenge:	Existing knowledge distillation methods for autoregressive large language models (LLMs) are not effective for reducing generation quality, but they can be useful for real-time applications.
Approach:	They propose a self-distillation framework that allows for effective KD without external teacher . they propose to use two modes of knowledge distillation to determine distillation direction .
Outcome:	The proposed framework outperforms existing methods with external teachers on instruction-following tasks.

Copied to clipboard

Zijing Shi, Meng Fang, Ling Chen

Challenge:	Existing web agents are highly susceptible to multiple classes of deceptive interfaces, but they are not designed to mitigate these failures.
Approach:	They propose a lightweight plugin framework that allows controlled injection of deceptive interface patterns into existing web environments.
Outcome:	The proposed framework enables controlled injection of deceptive interface patterns into web environments.

Counteracting the Matthew Effect in Self-Improvement of LVLMs through Head-Tail Re-balancing (2026.acl-long)

Copied to clipboard

Xin Guo, Zhiheng Xi, Yiwen Ding, Yitao Zhai, Xiaowei Shi, Xunliang Cai, Tao Gui, Qi Zhang, Xuanjing Huang

MAGIC: Deep Geometric Evolution with Structural Consensus for Temporal Knowledge Graph Reasoning (2026.acl-long)

Challenge:	Large vision language models have impressive reasoning capabilities across complex multimodal tasks.
Approach:	They propose to use distribution-reshaping and trajectory-rebalancing to improve visual reasoning capabilities.
Outcome:	Experiments on Qwen2-VL-7B-Instruct and InternVL2.5-4B models show that their methods outperform baselines by 3.86 points.

Copied to clipboard

Chengao Liu, Yuan Li, Yingze Wang, Jianbin Jiao

MACS: Modality-Aware Capacity Scaling for Efficient Multimodal MoE Inference (2026.acl-long)

Challenge:	Existing multi-geometry approaches face two key bottlenecks: Riemannian depth barrier and gate collapse.
Approach:	They propose a framework for Temporal Knowledge Graph reasoning that integrates a Tangent-Residual Engine into multi-geometric spaces to regulate gradient flow and prevent collapse.
Outcome:	The proposed framework improves state-of-the-art in TKG reasoning by up to 2.9 points.

Copied to clipboard

Bo Li, Chuan Wu, Shaolin Zhu

Efficient Prior-Guided Reasoning for Robust Retrieval-Augmented Generation under Conflicts (2026.acl-long)

Challenge:	Existing methods for expert parallelism inference suffer from a significant efficiency bottleneck . existing methods fail to address information heterogeneity and modality dynamics .
Approach:	They propose a training-free inference framework that scales experts without training . they propose an Entropy-Weighted Load mechanism to quantify the semantic value of visual tokens .
Outcome:	Experiments show that MACS outperforms existing methods on multimodal benchmarks.

Copied to clipboard

Xiaowei Yuan, Ziyang Huang, Zhao Yang, Yequan Wang, Jun Zhao, Kang Liu

Markovian Linguistic-Temporal Bridge: Unlocking the Potential of LLMs for Time Series Forecasting (2026.acl-long)

Challenge:	Retrieval-Augmented Generation (RAG) has become a standard paradigm for grounding Large Language Models (LLMs) however, performance degrades substantially when faced with noisy, outdated, or conflicting retrieved information.
Approach:	They propose a framework that explicitly elicits the model’s parametric knowledge as prior information to guide reasoning on retrieved documents.
Outcome:	The proposed framework achieves robust performance across varying degrees of external inconsistency and noise.

Copied to clipboard

Siming Sun, Kai Zhang, Xuejun Jiang, Wenchao Meng, Qinmin Yang

Challenge:	Pretrained Large Language Models (LLMs) are based on token-level linguistic-temporal alignment, leading to stacking of logically disjointed tokens as input.
Approach:	They propose a framework that distills latent evolutionary patterns of language into a Markovian state transition graph, which is transferred as a structural prior to the time series domain.
Outcome:	The proposed framework achieves global structural isomorphism between the linguistic and temporal domains.

Template-assisted Contrastive Learning of Task-oriented Dialogue Sentence Embeddings (2026.acl-long)

Copied to clipboard

Minsik Oh, Jiwei Li, Guoyin Wang

MIRTH: Mutual-Information Reasoning with Temporal Hubs for Vision-Language-Action Agents (2026.acl-long)

Challenge:	Annotating and gathering utterance relationships in dialogues is difficult, while token-level annotations, entities, slots and templates, are much easier to obtain.
Approach:	They propose a template-aware augmentation method that utilizes template information to learn utterance embeddings via self-supervised contrastive learning framework.
Outcome:	The proposed method improves on five benchmark dialogue datasets and shows that it is more efficient than previous SOTA methods.

Copied to clipboard

Hao Sun, Yu Song, Teng Shiyu, Ziwei Niu, Yen-wei Chen

IRIS: Interleaved Reinforcement with Incremental Staged Curriculum for Cross-Lingual Mathematical Reasoning (2026.acl-long)

Challenge:	a recent study shows that VLA models suffer from temporal myopia that discards historical dynamics and reasoning gaps between high-level instructions and low-level motor commands.
Approach:	They propose a framework to address temporal myopia and autoregressive scalar decoding in VLAs . they propose two memory hubs that compress long-term scene evolution and short-term motion trends .
Outcome:	The proposed framework achieves state-of-the-art performance and exhibiting emergent error recovery capabilities.

Copied to clipboard

Navya Gupta, Rishitej Reddy Vyalla, Avinash Anand, Chhavi Kirtani, Erik Cambria, Zhengchen Zhang, Zhengkui Wang, Timothy Liu, Aik Beng Ng, Simon See, Rajiv Ratn Shah

FLARE: Fine-Grained Length-Aware Routing for Resource-Efficient Heterogeneous LLM Serving (2026.acl-long)

Challenge:	Curriculum learning fails to generate consistent step-by-step reasoning in multilingual and low-resource settings.
Approach:	They propose a framework that combines supervised fine-tuning with reverse curriculum reinforcement learning to generate consistent step-by-step reasoning.
Outcome:	The proposed framework outperforms single-axis benchmarks and multilingual test sets on math reasoning tasks and in high-resource languages.

Copied to clipboard

Yujia Fu, Heming Zhong, Dan Huang, Yutong Lu

MAESTRO: Meta-learning Adaptive Estimation of Scalarization Trade-offs for Reward Optimization (2026.acl-long)

Challenge:	Existing routers lack fine-grained resource awareness across deployment settings, which degrades efficiency metrics in real-world serving.
Approach:	They propose a length-centric, resource-aware multi-LLM routing framework that uses length-based models to estimate per-query latency and cost.
Outcome:	Experiments show that FLARE reduces latency and cost by up to 68% and 75% while maintaining competitive accuracy.

Copied to clipboard

Yang Zhao, Hepeng Wang, Xiao Ding, Yangou Ouyang, Bibo Cai, Kai Xiong, Jinglong Gao, Zhouhao Sun, Li Du, Bing Qin, Ting Liu

Bit-by-Bit: Progressive QAT Strategy with Outlier Channel Splitting for Stable Low-Bit LLMs (2026.acl-long)

Challenge:	Group-Relative Policy Optimization (GRPO) has emerged as an efficient paradigm for aligning Large Language Models (LLMs), but its efficacy is confined to domains with verifiable ground truths.
Approach:	They propose a meta-cognitive orchestration layer that treats reward scalarization as a dynamic latent policy, leveraging the model’s terminal hidden states as 'a semantic bottleneck' . Across seven benchmarks, MAESTRO consistently outperforms single-reward and static multi-objective baselines while preserving the efficiency advantages of GRPO.
Outcome:	The proposed model outperforms single-reward and static multi-objective baselines while preserving efficiency advantages.

Copied to clipboard

Binxing Xu, Hao Gu, Lujun Li, Hao Wang, Bei Liu, Jiacheng Liu, Qiyuan Zhu, Xintong Yang, Chao Li, Sirui Han, Yike Guo

OS-Symphony: A Holistic Framework for Robust and Generalist Computer-Using Agents (2026.acl-long)

Challenge:	Existing approaches to training LLMs at ultra-low precisions suffer from convergence instability and substantial training costs.
Approach:	They propose a progressive QAT framework with outlier channel splitting to address these issues . they use nested structure of integer quantization grids to enable a "train once, deploy any precision" paradigm .
Outcome:	The proposed framework outperforms baselines on both Llama2/3 and W2A16, with an 11 speedup over BF16.

Copied to clipboard

Bowen Yang, Kaiming Jin, Zhenyu Wu, Zhaoyang Liu, Qiushi Sun, Zehao Li, JingJing Xie, Zhoumianze Liu, Fangzhi Xu, Kanzhi Cheng, Yian Wang, Qingyun Li, Yu Qiao, Zun Wang, Zichen Ding

Learning from Cognition: Enhancing RL Efficiency for LLM Reasoning via Hierarchical Metacognitive Decomposition and Refinement (2026.acl-long)

Challenge:	Vision-Language Models (VLMs) lack visual-aware tutorial retrieval and historical visual context curation and pruning.
Approach:	They propose a framework that integrates an orchestrator and a Reflection-Memory Agent for robust automation.
Outcome:	Experimental results show that OS-Symphony delivers substantial performance gains across model scales.

Copied to clipboard

Zexu Sun, Yongcheng Zeng, Erxue Min, Heyang Gao, Bokai Ji, Dugang Liu, Xing Tang, Xiuqiang He, Xu Chen

BrowseComp-Plus: A Fair and Disentangled Evaluation Benchmark for Deep Search Agents (2026.acl-long)

Challenge:	Recent advances in Large Language Models have demonstrated notable inferential capacities via reinforcement learning (RL) however, “zero-RL” approaches relying on fixed prompt templates introduce substantial sampling inefficiencies for weak LLMs.
Approach:	They propose a hierarchical metacognitive RL framework that decomposes zero-accuracy problems into subproblems and prompts the policy to refine answers by referencing previous wrong solutions.
Outcome:	The proposed framework improves sample utilization and sample efficiency and accelerates convergence compared to baselines.

Copied to clipboard

Zijian Chen, Xueguang Ma, Shengyao Zhuang, Ping Nie, Kai Zou, Sahel Sharifymoghaddam, Andrew Liu, Joshua Green, Kshama Patel, Ruoxi Meng, Mingyi Su, Yanxi Li, Haoran Hong, Xinyu Shi, Xuye Liu, Hosna Oyarhoseini, Nandan Thakur, Crystina Zhang, Luyu Gao, Wenhu Chen, Jimmy Lin

LitVISTA: A Benchmark for Narrative Orchestration in Literary Text (2026.acl-long)

Challenge:	Existing benchmarks for deep search agents rely on blackbox web search APIs . dynamic and opaque web APIs hinder reproducibility and fair comparisons - authors .
Approach:	They propose a benchmark that employs a fixed corpus for controlled retrieval for deep search agents.
Outcome:	The new benchmark shows that agents that combine large language models with retrieval tools excel at complex, reasoning-intensive queries.

Copied to clipboard

Mingzhe Lu, Yiwen Wang, Yanbing Liu, Qi You, Chong Liu, Ruize Qin, Haoyu Dong, Wenyu Zhang, JiaRui Zhang, Yue Hu, Yunpeng Li

Challenge:	Existing large language models focus on causal coherence, neglecting the complex story arcs and orchestration inherent in human narratives.
Approach:	They propose a high-dimensional framework for narrative orchestration that unifies human and model perspectives while jointly characterizing narrative function and structure in a common space.
Outcome:	The proposed framework unifies human and model perspectives while jointly characterizing narrative function and structure in a common space.

RST-Guarder: Enhancing Long-Context Robustness for Safeguards via RST Parsing and Probabilistic Inference (2026.acl-long)

Copied to clipboard

Xu Zhang, Xiaojun Wan

EpiCaR: Knowing What You Don’t Know Matters for Better Reasoning in LLMs (2026.acl-long)

Challenge:	Existing guardrail models for harmful-content detection degrade on long-form inputs . Existing models are vulnerable to policy-violating responses, causing false positives based on benign content .
Approach:	They propose an inference-time method that improves harmful-content detection for long-form inputs without additional data curation or model training.
Outcome:	The proposed method improves harmful-content detection for long-form inputs without additional data curation or model training.

Copied to clipboard

Jewon Yeom, Jaewon Sok, Seonghyeon Park, Jeongjae Park, Taesup Kim

MoEC: A Memory-Routed Mixture-of-Experts Controller for Adaptive Minecraft Control (2026.acl-long)

Challenge:	Existing approaches to improving reasoning abilities of large language models incur a significant calibration cost.
Approach:	They propose an epistemic learning problem that integrates reasoning and calibration into an iterative supervised training framework.
Outcome:	The proposed method achieves Pareto-superiority over standard baselines in accuracy and calibration.

Copied to clipboard

Hui Wu, Chao Xu, Jianghui Wang, Ziqiong Liu, Dong Li, Yiwei Dai, Emad Barsoum

MECH: A Cost-Effective Multi-Task Cascade Framework for Classroom Opinion Evolution Recognition (2026.acl-long)

Challenge:	Existing systems rely on a monolithic policy to execute subgoals across varying contexts, causing inconsistent outcomes and scaling only partially mitigates.
Approach:	They propose a memory-routed mixtureof-experts controller for Adaptive Minecraft Control that routes via a subgoal-indexed expert memory and regulates capacity through failure-triggered expert growth and redundancy-aware consolidation.
Outcome:	The proposed controller shows significant gains in adaptability, robustness, and execution consistency over strong baselines.

Copied to clipboard

Yancui Li, Xiaoyu Zhou, Guoyi Miao, Fang Kong

Debiasing Reward Models via Causally Motivated Inference-Time Intervention (2026.acl-long)

Challenge:	Existing studies focus on dialogue act annotation, overlooking the deeper dimension of opinion evolution.
Approach:	They propose a framework for Classroom Opinion Evolution Recognition that translates "Action-Opinion" dualism into a risk-aware routing mechanism.
Outcome:	The proposed framework achieves state-of-the-art accuracy of 78.55% while reducing API costs by 44.4%.

Copied to clipboard

Kazutoshi Shinoda, Kosuke Nishida, Kyosuke Nishida

Steganography Beyond Pixels: Reimagining Image Steganography as Cross-Modal Linguistic Communication (2026.acl-long)

Challenge:	Existing approaches for mitigating spurious features in RMs focus on response length . Existing methods focus on RM activation, resulting in performance trade-offs .
Approach:	They propose a method that uses neurons to suppress spurious features in RMs at inference time.
Outcome:	The proposed method reduces sensitivity to spurious features without inducing performance trade-offs on RM benchmarks.

Copied to clipboard

Ren Lijing, Denghui Zhang

Hard2Verify: A Step-Level Verification Benchmark for Open-Ended Frontier Math (2026.acl-long)

Challenge:	steganography relies on detectable pixel-level perturbations to conceal sensitive data . conventional cryptography fails to conceal the act of transmission, resulting in conspicuous anomalies . stegography is a discrete approach to concealing data within innocuous communication channels .
Approach:	They propose a steganography framework that reorients stegographic containers from visual to linguistic domains by quantizing high-resolution secret images into binary payloads.
Outcome:	The proposed framework bridges the gap between visual and linguistic domains.

Copied to clipboard

Shrey Pandit, Austin Xu, Xuan-Phi Nguyen, Yifei Ming, Caiming Xiong, Shafiq Joty

Agent-based Substructure Counting under Local Differential Privacy (2026.acl-long)

Challenge:	Large language model (LLM)-based reasoning systems have recently achieved gold medal-level performance in the IMO 2025 competition .
Approach:	They propose a human-annotated step-level verification benchmark that measures step- level verifiers at the frontier.
Outcome:	The proposed benchmark outperforms closed-source models in step-level verification and the impact of scaling verifier compute.

Copied to clipboard

Yuting Zhang, Kai Wang, Wei Ni, Ying Zhang, Wenjie Zhang

NeuReasoner: Towards Explainable, Controllable, and Unified Reasoning via Mixture-of-Neurons (2026.acl-long)

Challenge:	Recent studies have demonstrated the ability of Large Language Models (LLMs) to process graph problems.
Approach:	They propose to decompose substructure counting into node-level tasks distributed among node agents and embed the knowledge of distributed algorithms and DP frameworks in the curator agent and privacy controller.
Outcome:	Extensive experiments on 6 real-world datasets validate the effectiveness of the proposed framework for substructure counting tasks under edge local differential privacy (LDP).

Copied to clipboard

Haonan Dong, Kehan Jiang, Haoran Ye, Wenhao Zhu, Zhaolu Kang, Guojie Song

Red Teaming Large Reasoning Models (2026.acl-long)

Challenge:	Existing Large Reasoning Models (LRMs) lack explainability and controllability . Existing models target isolated levels without unification, while relying on RL .
Approach:	They propose an explainable, controllable, and unified reasoning framework driven by MoN.
Outcome:	The proposed framework achieves performance gains of 27.0% while reducing token consumption by 19.6% 63.3%.

Copied to clipboard

Jiawei Chen, Yang Yang, Chao Yu, Yu Tian, Zhi Cao, Xue Yang, Linghao Li, Hang Su, Zhaoxia Yin

InsLogicBench: An Argumentation Logic Grounded Benchmark for Complex Insurance Claims Adjudication (2026.acl-long)

Challenge:	Large Reasoning Models (LRMs) have emerged as a powerful advancement in multi-step reasoning tasks, but they introduce safety and reliability risks, such as CoT-hijacking and prompt-induced inefficiencies.
Approach:	They propose a unified benchmark to assess the trustworthiness of Large Reasoning Models.
Outcome:	The proposed benchmark evaluates truthfulness, safety and efficiency on 26 models.

Copied to clipboard

Jin Liu, Yunpeng Liu, Keyi Wang, Jie Shi, Xiao Xu, Wenkang Huang, Xingzhong Xu, Xin Liang, Yanghua Xiao

LinkQA: Synthesizing Diverse QA from Multiple Seeds Strongly Linked by Knowledge Points (2026.acl-long)

Challenge:	Existing benchmarks for insurance claims adjudication are limited to information retrieval or simple multiple-choice setups.
Approach:	They propose a benchmark that provides complete reasoning traces linking factual inputs, relevant policy clauses, and final verdicts.
Outcome:	The proposed benchmark shows that models often produce correct decisions but fail to provide precise justifications, highlighting a critical discrepancy between decision accuracy and logical reasoning capabilities.

Copied to clipboard

Xuemiao Zhang, Can Ren, Chengying Tu, Rongxiang Weng, Hongfei Yan, Jingang Wang, Xunliang Cai

Comparative Analysis of the Intrinsic Metrics for Tokenizers and their effect on Downstream Tasks for Hindi and Marathi (2026.acl-long)

Challenge:	Existing training data is limited in high-quality training data, limiting the ability to produce high-performance LLMs.
Approach:	They propose a KP-graph-based synthesis framework that extracts KPs from QA seed data and constructs a graph of KP data from multiple seeds strongly linked by KP.
Outcome:	The proposed framework enables flexible control over discipline and difficulty distributions while balancing KP coverage and popularity.

Copied to clipboard

Shagun Dwivedi, Kaushik Gopalan

Omni-Embed-Audio: Leveraging Multimodal LLMs for Robust Audio-Text Retrieval (2026.acl-long)

Challenge:	Various studies have shown that the performance of language models is poor in non-English or non-European languages.
Approach:	They propose a grapheme cluster tokenizer which shows better performance than other popular tokenizers.
Outcome:	The proposed tokenizers show better or competitiveness on question-answering tasks . the proposed tokenization model is highly correlated to the performance of other tokenizer models .

Copied to clipboard

HaeJun Yoo, Yongseop Shin, Insung Lee, Myoung-Wan Koo, Du-Seong Chang

SPIO: Ensemble and Selective Strategies via LLM-Based Multi-Agent Planning in Automated Data Science (2026.acl-long)

Challenge:	Experiments with AudioCaps, Clotho, and MECAT show that OEA achieves comparable text-to-text retrieval performance to state-of-the-art M2D-CLAP.
Approach:	They propose a retrieval-oriented encoder leveraging multimodal LLMs with native audio understanding that allows users to express their queries in five different ways.
Outcome:	Experiments on AudioCaps, Clotho, and MECAT show that OEA achieves comparable text-to-audio retrieval performance to state-of-the-art M2D-CLAP while demonstrating clear advantages in two critical areas.

Copied to clipboard

Wonduk Seo, Juhyeon Lee, Yanjun Shao, Qingshan Zhou, Seunghyun Lee, Yi Bu

Comparing human and language models sentence processing difficulties on complex structures (2026.acl-long)

Challenge:	Large Language Models (LLMs) have enabled dynamic reasoning in automated data analytics, but rigid, single-path workflows restrict strategic exploration and often lead to suboptimal outcomes.
Approach:	a new framework replaces rigid workflows with adaptive, multi-path planning . the framework offers two operating modes: SPIO-S and SPIO -E .
Outcome:	a new framework outperforms state-of-the-art pipelines on Kaggle and OpenML benchmarks.

Copied to clipboard

Samuel Joseph Amouyal, Aya Meltzer-Asscher, Jonathan Berant

Phun-Bench: Evaluating LLMs on Phonological Understanding in Chinese (2026.acl-long)

Challenge:	Large language models (LLMs) that converse with humans are a reality, but do LLMs experience human-like processing difficulties?
Approach:	They systematically compare human and LLM sentence comprehension across seven challenging linguistic structures.
Outcome:	The proposed model achieves near perfect accuracy on non-GP structures, but struggles on GP structures.

Copied to clipboard

Xing Yue, Yongliang Shen, Weiming Lu

Causal2Vec: Improving Decoder-only LLMs as Embedding Models through a Contextual Token (2026.acl-long)

Challenge:	Existing benchmarks on LLMs’ phonological abilities are either solvable through rote memorization or intertwined with other abilities, making them inadequate to measure LLM’s genuine ability in phonological understanding.
Approach:	They propose to use a Chinese benchmark to evaluate LLMs' phonological understanding to test their ability to recall correct pronunciations.
Outcome:	The proposed benchmarks show that LLMs excel at recalling correct pronunciations, but struggle to leverage phonological knowledge in the flexible and intuitive way that human speakers do.

Copied to clipboard

Ailiang Lin, Zhuoyun Li, Yusong Wang, Kotaro Funakoshi, Manabu Okumura

Enhancing Multimodal Large Language Models for Ancient Chinese Character Evolution Analysis via Glyph-Driven Fine-Tuning (2026.acl-long)

Challenge:	Existing methods modify attention mechanism to be bidirectional, undermining LLMs’ ability to extract semantic information acquired during pre-training.
Approach:	They propose a general-purpose embedding model that pre-encodes input text into a single Contextual token and then prepends it to the LLM's input sequence.
Outcome:	The proposed model improves performance of decoder-only large language models without altering their architectures or introducing significant computational overhead.

Copied to clipboard

Rui Song, Lida Shi, Ruihua Qi, Yingji Li, Hao Xu

See2Refine: Vision-Language Feedback Improves LLM-Based eHMI Action Designers (2026.acl-long)

Challenge:	Existing studies have explored the evolutionary analysis of ancient scripts, with particular attention to the transformation of character forms from oracle bone inscriptions to regular script.
Approach:	They propose a benchmark framework that leverages MLLMs to analyze the evolution of ancient Chinese scripts.
Outcome:	The proposed framework improves performance on core tasks and character recognition and evolutionary reasoning tasks while limiting performance on other tasks.

Copied to clipboard

Ding Xia, Xinyue Gui, Mark Colley, Fan Gao, Zhongyi Zhou, Dongyuan Li, Renhe Jiang, Takeo Igarashi

A Survey of Reinforcement Learning for Large Language Models under Data Scarcity: Challenges and Solutions (2026.acl-long)

Challenge:	External Human-Machine Interfaces (eHMIs) are emerging as promising solutions to address this communication gap.
Approach:	They propose a framework that uses vision-language models (VLMs) for perceptual evaluation as automated visual feedback to improve an LLM-based eHMI action designer.
Outcome:	The proposed framework outperforms prompt-only LLM designers and manually specified baselines in three eHMI modalities and multiple LLM model sizes.

Copied to clipboard

Zhiyin Yu, Yuchen Mou, Juncheng Yan, Junyu Luo, Chunchun Chen, Xing Wei, Yunhui Liu, Hongru Sun, Yuxing Zhang, Jun Xu, Yatao Bian, Ming Zhang, Wei Ye, Tieke He, Jie Yang, Guanjie Zheng, Zhonghai Wu, Bo Zhang, Lei Bai, Xiao Luo

SADA: Bridging In-Context Learning and Fine-Tuning via State-Aligned Distillation Adapters (2026.acl-long)

Challenge:	Existing research on reinforcement learning for LLMs under data scarcity has not been unified.
Approach:	They propose a top-up hierarchical framework built around three complementary perspectives: data-centric, training-centric and framework-centric.
Outcome:	The proposed framework provides a clear conceptual foundation for understanding the design space of data-efficient RL for large language models and to guide researchers working in this emerging area.

Copied to clipboard

Wenhao Gao, Tianlong Wang, Wei Jia, Linhao Zhang, Aiwei Liu, Miao Fan, Zhou Xiao

Anti-Length Shift: Dynamic Outlier Truncation for Training Efficient Reasoning Models (2026.acl-long)

Challenge:	Prompt-based in-context learning and parameter fine-tuning are dominant paradigms for incorporating external information into large language models, but they incur high inference costs or require expensive retraining.
Approach:	They propose to convert prompts into temporary adapter weights to bridge this gap by converting prompts to temporary adapters.
Outcome:	The proposed model outperforms baselines on long-context language modeling and downstream NLU and summarization benchmarks while significantly reducing memory footprint and latency.

Copied to clipboard

Wei Wu, Liyi Chen, Congxi Xiao, Tianfu Wang, Qimeng Wang, Chengqiang Lu, Yan Gao, null Yiwu, Yao Hu, Hui Xiong

The Proxy Presumption: From Semantic Embeddings to Valid Social Measures (2026.acl-long)

Challenge:	Existing efficient reasoning methods rely on explicit length penalties for excessive verbosity on simple queries.
Approach:	They propose a training-time intervention that selectively suppresses redundant tokens . they find length shift occurs when models generate unnecessary reasoning on trivial inputs - a phenomenon that is often unexplored .
Outcome:	The proposed method reduces inference token usage by 78% while increasing accuracy compared to the initial policy and surpasses state-of-the-art efficient reasoning methods.

Copied to clipboard

Baishi Li, Ta Yu, Kelvin J.l. Koa, Ke-Wei Huang

Nested Browser-Use Learning for Agentic Information Seeking (2026.acl-long)

Challenge:	Natural Language Processing is evolving into a primary instrument for Computational Social Science . but, this transition faces a validity challenge: the “Proxy Presumption” .
Approach:	They propose a method to bridge the gap between semantic embeddings and valid social measures by using the Construct Validity Protocol.
Outcome:	The proposed method bridges the gap between semantic embeddings and valid social measures . it reduces confounding in embeddable space and provides a standardized Validity Suite .

Copied to clipboard

Baixuan Li, Jialong Wu, Wenbiao Yin, Kuan Li, Zhongwang Zhang, Huifeng Yin, Zhengwei Tao, Liwen Zhang, Pengjun Xie, Jingren Zhou, Yong Jiang, Wentao Zhang, Zhiqiang Gao

Challenge:	Existing information-seeking (IS) agents rely on the web for their information acquisition.
Approach:	They propose a browser-action framework that decouples interaction control from page exploration through a nested structure.
Outcome:	Empirical results show that NestBrowse offers clear benefits in practice.

CamoQuery: Language-Guided Reasoning Camouflaged Object Segmentation (2026.acl-long)

Copied to clipboard

Tianxin Han, Qing Dong, Xingwei Wang, Jie Jia, Gang Wu, Bowen Yang, Fu Zhang

Large Language Models Are Bad Dice Players: LLMs Struggle to Generate Random Numbers from Statistical Distributions (2026.acl-long)

Challenge:	Existing methods for camouflaged object segmentation are limited to vision-only mask prediction under fixed task assumptions.
Approach:	They propose a language-guided reasoning camouflaged object segmentation task that generates an intent-consistent segmentation mask from an image and an implicit query text instruction.
Outcome:	The proposed task can generate an intent-consistent segmentation mask from a camouflaged image and an implicit query text instruction.

Copied to clipboard

Minda Zhao, Yilun Du, Mengyu Wang

PROMPRINT: Prompt Fingerprinting via First-Token Response for LLM App Cloning Detection (2026.acl-long)

Challenge:	Existing large language models lack a functional internal sampler to faithfully sample from specified probability distributions . lack of robust sampling mechanisms across diverse application scenarios is a critical functional requirement .
Approach:	They propose to use a dual-protocol design to disentangle failure modes . batch generation achieves only modest statistical validity, while independent requests collapse almost entirely .
Outcome:	The proposed model fails to enforce uniform answer-position constraints and violates demographic targets in attribute-constrained text-to-image prompt synthesis.

Copied to clipboard

Jungmin Lee, Peizhuo Lv, Yeonjoon Lee

MAB-DQA: Addressing Query Aspect Importance in Document Question Answering with Multi-Armed Bandits (2026.acl-long)

Challenge:	Large Language Model apps are increasingly regarded as intellectual property . cloned apps pose risks of copyright infringement and malicious misuse .
Approach:	They propose a fingerprinting approach that optimizes queries that induce the LLM to generate a specific first token associated with the given system prompt.
Outcome:	The proposed fingerprinting approach is robust to partial system prompt modifications and effective under injection of adversarial instructions.

Copied to clipboard

Yixin Xiang, Yunshan Ma, Xiaoyu Du, Yibing Chen, Yanxin Zhang, Jinhui Tang

DVCQR: Dual-View Conversational Query Rewriting with Stage-wise Reinforcement Learning (2026.acl-long)

Challenge:	Document Question Answering (DQA) requires interpreting visual layouts, which has prompted recent studies to adopt multimodal Retrieval-Augmented Generation (RAG) that processes page images for answer generation.
Approach:	They propose a multi-armed bandit-based DQA framework that explicitly models the varying importance of multiple implicit aspects in a query.
Outcome:	The proposed framework shows an improvement of 5%-18% over the state-of-the-art method on four benchmarks.

Copied to clipboard

Chenyi Li, Xinhui Tu, Zaixiang Wang

CoEvolve: Training LLM Agents via Agent-Data Mutual Evolution (2026.acl-long)

Challenge:	Recent approaches to improve retrieval effectiveness rely on a single rewrite . however, they often suffer from conflicting optimization signals .
Approach:	They propose a dual-view CQR framework that generates two complementary rewrites for each query.
Outcome:	Experiments show that DVCQR outperforms state-of-the-art methods on most metrics . the proposed framework generates two complementary rewrites for each query .

Copied to clipboard

Shidong Yang, Ziyu Ma, Tongwen Huang, Yiming Hu, Yong Wang, Xiangxiang Chu

Challenge:	Extensive experiments on AppWorld and BFCL demonstrate consistent and significant improvements over strong base models, yielding absolute gains of 19.43%, 15.58%, and 18.14%, respectively.
Approach:	They propose a framework that extracts feedback signals such as forgetting and uncertainty from rollout trajectories and utilizes them to guide LLM-based task synthesis.
Outcome:	Extensive experiments on AppWorld and BFCL show that the proposed framework improves over strong base models.

Make LLMs See Like Investigators, Not Just Think More: The Role of Structured Analysis in Investigative Reasoning (2026.acl-long)

Copied to clipboard

Jaewook Lee, Myeong-Cheol Kang, Jong-hun Shin

FormalScience: Scalable Human-in-the-Loop Autoformalisation of Science with Agentic Code Generation in Lean (2026.acl-long)

Challenge:	Criminal investigators and intelligence analysts have developed structured analytic techniques to evaluate competing hypotheses under incomplete information.
Approach:	They focus on the task of analyzing evidence from complex narratives and identifying the perpetrator among suspects using the MuSR murder mystery benchmark.
Outcome:	The PRISM framework outperforms general-purpose strategies across all models, with its effectiveness manifesting regardless of model scale.

Copied to clipboard

Jordan Meadows, Lan Zhang, Andre Freitas

MUR: Momentum Uncertainty guided Reasoning for Large Language Models (2026.acl-long)

Challenge:	Formalising informal mathematical reasoning into formally verifiable code is a significant challenge for large language models.
Approach:	They propose a domain-agnostic human-in-the-loop agentic pipeline to aid autoformalisation in scientific domains.
Outcome:	The proposed system produces syntactically correct and semantically aligned proofs for low cost.

Copied to clipboard

Hang Yan, Fangzhi Xu, Rongman Xu, Yifei Li, Jian Zhang, Haoran Luo, Xiaobao Wu, Anh Tuan Luu, Haiteng Zhao, Qika Lin, Jun Liu

BiMol-Diff: A Unified Diffusion Framework for Molecular Generation and Captioning (2026.acl-long)

Challenge:	Existing methods for optimizing reasoning quality are limited by overthinking.
Approach:	They propose a method that allocates thinking budgets to critical reasoning steps by tracking and aggregating step-wise uncertainty over time.
Outcome:	The proposed method reduces computation by over 45% on average while improving accuracy by 0.33–3.46%.

Copied to clipboard

Aditya Hemant Shahane, Anuj Kumar Sirohi, Devansh Arora, Nitin Kumar, Prathosh AP, Sandeep Kumar

PEAP: Proactive Embodied Action Sequence Planning with Joint Understanding of Vision and Audio Perception (2026.acl-long)

Challenge:	Autoregressive models struggle with long-range dependencies, while standard diffusion processes apply uniform corruption across positions.
Approach:	They propose a framework that bridges molecular structures and natural language . it assigns position-dependent corruption based on token recovery difficulty .
Outcome:	The proposed framework improves molecule reconstruction and captioning performance on two datasets.

Copied to clipboard

Tianwei Lan, Jiaqi Wu, Zeming Liu, Zhaoxin Fan, Haifeng Wang, Yuhang Guo

Automated Creativity Evaluation of Language Models Across Open-Ended Tasks (2026.acl-long)

Challenge:	Embodied action sequence planning focuses on the capability of embodied agents to implement action planning via environmental perception without explicit human instructions.
Approach:	They propose to use a multimodal dataset to evaluate the performance of multiple large language models to evaluate their models' environmental perception capabilities.
Outcome:	The proposed model shows that it lacks accurate environmental perception capabilities and that it can improve on the PEAP dataset.

Copied to clipboard

Tan Min Sen, Zachary Choy Kit Chun, Syed Ali Redha Alsagoff, Nadya Yuki Wangsajaya, Banerjee Mohor, Swaagat Bikash Saikia, Alvin Chan

Aligning Agents via Planning: A Benchmark for Trajectory-Level Reward Modeling (2026.acl-long)

Challenge:	Existing methods for evaluating creativity are tightly coupled to specific tasks and limiting scalability and generality.
Approach:	They propose a domain-agnostic framework for quantifying LLM creativity across open-ended tasks.
Outcome:	The proposed framework captures key facets of creativity including novelty, diversity, and task fulfilment with over 60% improved efficiency.

Copied to clipboard

Jiaxuan Wang, Yulan Hu, Wenjin Yang, Zheng Pan, Xin Li, Lan-Zhe Guo

Challenge:	Large Language Models (LLMs) evolve into agentic systems capable of autonomous tool invocation and complex reasoning.
Approach:	They propose a trajectory-level preference benchmark to evaluate judges' ability to distinguish preferred versus distractor agent trajectories in tool-integrated environments.
Outcome:	The proposed benchmark evaluates how well judges distinguish preferred versus distractor agent trajectories in complex tool-using scenarios.

From Synthesis to Clinical Assistance: A Strategy-Aware Agent Framework for Autism Intervention based on Real Clinical Dataset (2026.acl-long)

Copied to clipboard

Junhong Lai, Shuzhong Lai, Yanhao Yu, Wanlin Chen, Chenyu Yan, Haifeng Li, Lin Yao, Yueming Wang

How Training Data Shapes the Use of Parametric and In-Context Knowledge in Language Models (2026.acl-long)

Challenge:	Applied Behavior Analysis (ABA) is the gold standard for clinical intervention, but large language models struggle to adhere to its standardized procedures.
Approach:	They propose a strategy-aware framework to unify high-fidelity intervention dialogue synthesis and clinical decision support.
Outcome:	Experiments show that ASDAgent achieves nearly 80% strategic consistency with human experts.

Copied to clipboard

Minsung Kim, Dong-Kyum Kim, Jea Kwon, Nakyeong Yang, Kyomin Jung, Meeyoung Cha

PAR: Training-Free Positional Perturbation and Attention Recycling for Faithful OCR (2026.acl-long)

Challenge:	Large language models leverage parametric and in-context knowledge in training . however, when these sources conflict, models arbitrate based on their internal confidence .
Approach:	They conduct controlled experiments using synthetic corpora to identify data properties that shape knowledge utilization.
Outcome:	The results show that the robust use of both knowledge sources is an emergent property . the results provide guidance for designing training data that supports the reliability of parametric and in-context knowledge in language models.

Copied to clipboard

Yao Yao, Manwen Liao, Weitian Zhang, Zuchao Li, Hai Zhao

BTC-LLM: Efficient Sub-1-Bit LLM Quantization via Learnable Transformation and Binary Codebook (2026.acl-long)

Challenge:	In high-precision tasks, vision language models suffer from Linguistic Priors Hallucination .
Approach:	They propose a training-free, inference-time intervention framework to mitigate this by integrating visual encoders with Large Language Model decoders.
Outcome:	The proposed framework reduces hallucination rates by 12% in long-context scenarios while maintaining robust generalization on standard benchmarks.

Copied to clipboard

Hao Gu, Lujun Li, Hao Wang, Lei Wang, Zheyu Wang, Bei Liu, Jiacheng Liu, Qiyuan Zhu, Sirui Han, Yike Guo

Learning What to Ignore: Mitigating Negative Transfer in Medical Knowledge Fusion via Clinical Task-Adaptive Selection (2026.acl-long)

Challenge:	Recent sparsity-aware binarization approaches can achieve sub-1-bit compression, but they face performance degradation, mask-management overhead, and limited hardware compatibility.
Approach:	They propose a binary quantization framework that leverages binary pattern clustering and weight transformation to overcome performance degradation and mask-management overhead.
Outcome:	The proposed framework achieves state-of-the-art compression (1.11–0.7 bits) it maintains high performance with only a 3.1% accuracy drop in zero-shot benchmarks while delivering a 1.6 speedup over FP16.

Copied to clipboard

Xinyan Deng, Shoubin Dong, Xiaorou Zheng

MMTutorBench: The First Multimodal Benchmark for AI Math Tutoring (2026.acl-long)

Challenge:	Existing approaches to longitudinal EHR modeling struggle to balance structural authority of static ontologies with reasoning flexibility of large language models.
Approach:	They propose a framework that integrates external medical knowledge into longitudinal EHR modeling to mitigate clinical data sparsity.
Outcome:	The proposed framework outperforms state-of-the-art models on four clinical tasks.

Copied to clipboard

Tengchao Yang, Sichen Guo, Mengzhao Jia, Jiaming Su, Yuanyang Liu, Zhihan Zhang, Meng Jiang

Numerical Reasoning

Challenge:	Existing benchmarks for AI math tutoring largely overlook these skills.
Approach:	They evaluate 12 leading multimodal large language models and find clear performance gaps between them.
Outcome:	The proposed benchmarks show that they can solve 770 problems and provide diagnostics and guidance to students step by step.

GrACE: A Generative Approach to Better Confidence Elicitation and Efficient Test-Time Scaling in Large Language Models (2026.acl-long)

Copied to clipboard

Zhaohan Zhang, Ziquan Liu, Ioannis Patras

AutoGraph-R1: End-to-End Reinforcement Learning for Knowledge Graph Construction (2026.acl-long)

Challenge:	Existing methods for assessing the reliability of Large Language Models (LLMs) by confidence elicitation require expensive computational overhead or suffer from poor calibration, making them unreliable for real-world deployment.
Approach:	They propose a Generative Approach to Confidence Elicitation that enables reliable confidence elicitation for Large Language Models.
Outcome:	The proposed method achieves the best discriminative capacity and calibration on open-ended tasks without resorting to additional sampling or an auxiliary model.

Copied to clipboard

Hong Ting Tsang, Jiaxin Bai, Haoyu Huang, Qiao Xiao, Tianshi Zheng, Baixuan Xu, Shujie Liu, Yangqiu Song

Understanding Structured Financial Data with LLMs: A Case Study on Fraud Detection (2026.acl-long)

Challenge:	Currently, knowledge graphs are decoupled from their downstream application, resulting in suboptimal graph structures.
Approach:	They propose a framework to directly optimize KG construction for task performance using Reinforcement Learning (RL).
Outcome:	The proposed framework improves performance across multiple QA benchmarks and consistently achieves significant performance gains over task-agnostic baseline graphs.

Copied to clipboard

Xuwei Tan, Yao Ma, Xueru Zhang

Gap-K%: Measuring Top-1 Prediction Gap for Detecting Pretraining Data (2026.acl-long)

Challenge:	Large Language Models (LLMs) are expensive to develop and maintain and require extensive feature engineering to perform.
Approach:	They propose a two-stage approach that serializes a compact subset of numeric/categorical attributes into natural language and performs retrieval-augmented in-context learning over label-aware, instance-level exemplars.
Outcome:	The proposed approach significantly improves F1/MCC over direct prompting and is competitive with strong tabular baselines in several settings.

Copied to clipboard

Minseo Kwak, Jaehyung Kim

Experience Retrieval-Augmentation with Electronic Health Records Enables Accurate Discharge QA (2026.acl-long)

Challenge:	Existing state-of-the-art methods for pretraining data are largely undisclosed, resulting in ethical and copyright concerns.
Approach:	They propose a method that leverages the log probability gap between the top-1 predicted token and the target token, incorporating a sliding window strategy to capture local correlations and mitigate token-level fluctuations.
Outcome:	The proposed method outperforms baselines on WikiMIA and MIMIR benchmarks and achieves state-of-the-art performance.

Copied to clipboard

Justice Ou, Tinglin Huang, Yilun Zhao, Ziyang Yu, Peiqing Lu, Yifei Shen, Rex Ying

Are they lovers or friends? Evaluating LLMs’ Social Reasoning in English and Korean Dialogues (2026.acl-long)

Challenge:	Existing methods to improve the reliability of Large Language Models (LLMs) in clinical applications require factual knowledge from open-ended datasets and clinical case-based knowledge to provide context grounded in real-world patient experiences.
Approach:	They propose a retrieval-augmented generation framework based on the electronic health record to offer contextual information from other patients’ discharge reports.
Outcome:	The proposed framework outperforms a text-based ranker in a clinical QA dataset with 1,280 discharge-related questions .

Copied to clipboard

Eunsu Kim, Junyeong Park, Juhyun Oh, Kiwoong Park, Seyoung Song, A. Seza Doğruöz, Alice Oh, Najoung Kim

SAIR-Comb : A Structure-Aware Iterative Refinement Framework for Combinatorics Autoformalization (2026.acl-long)

Challenge:	Existing studies on LLMs' ability to infer social relationships have limited results for Korean and English.
Approach:	They propose a social reasoning task based on a 1.1k-dialogue dataset in English and Korean sourced from movie scripts to evaluate LLMs' ability to infer the social relationships between speakers.
Outcome:	The proposed task evaluates the ability of LLMs to infer the social relationships between speakers in 1.1k-dialogue datasets in English and Korean.

Copied to clipboard

Weijie Jiang, Gaolei He, Beibei Xiong, Jianlin Wang, Zhengfeng Yang

Stop Hardening Everything: A Training-Free Neuron-Level Defense for Neural Ranking Models (2026.acl-long)

Challenge:	Large language models (LLMs) have catalyzed advances in mathematical reasoning, propelling the development of automated theorem proving (ATP).
Approach:	They propose a Structure-Aware Iterative Refinement framework for Combinatorics powered by Lean 4 and LLMs.
Outcome:	The proposed framework achieves strong performance on the specialized CombiBench while remaining highly competitive on general-domain benchmarks.

Copied to clipboard

Yu-An Liu, Ruqing Zhang, Hongru Song, Jiafeng Guo, Yixing Fan, Xueqi Cheng

TED-TTS: Training-Free Intra-Utterance Emotion and Duration Control for Text-to-Speech Synthesis (2026.acl-long)

Challenge:	Existing defenses for neural ranking models are data-centric and require retraining and adversarial data generation.
Approach:	They propose a model-centric defense that addresses vulnerability at its architectural source without costly retraining or adversarial data generation.
Outcome:	The proposed approach outperforms state-of-the-art models on MS MARCO and TREC 19 while maintaining strong performance on clean data.

Copied to clipboard

Qifan Liang, Yuansen Liu, Ruixin Wei, Nan Lu, Junchuan Zhao, Ye Wang

VIB-Probe: Detecting and Mitigating Hallucinations in Vision-Language Models via Variational Information Bottleneck (2026.acl-long)

Challenge:	Existing controllable Text-to-Speech methods limited to inter-utterance-level control . utterance expressiveness remains a challenge in building human-like TTS synthesis systems .
Approach:	They propose a training-free controllable framework for pretrained zero-shot TTS to enable intra-utterance emotion and duration expression.
Outcome:	The proposed framework achieves state-of-the-art intra-utterance consistency while maintaining baseline-level speech quality.

Copied to clipboard

Feiran Zhang, Yixin Wu, Zhenghua Wang, Xiaohua Wang, Changze Lv, Xuanjing Huang, Xiaoqing Zheng

Detoxification for LLM: From Dataset Itself (2026.acl-long)

Challenge:	Existing hallucination detection methods rely on external verification tools . however, entanglement of visual-linguistic syntax and noise makes it difficult to detect hallucis .
Approach:	They propose a hallucination detection framework that leverages the Variational Information Bottleneck theory to detect hallucinic heads and to infer hallucication mitigation strategies.
Outcome:	The proposed framework outperforms baselines in hallucinations and noise detection environments.

Copied to clipboard

Wei Shao, Yihang Wang, Gao yu Zhu, Ziqiang Cheng, Lei Yu, Jiafeng Guo, Xueqi Cheng

When Is Thinking Enough? Early Exit via Sufficiency Assessment for Efficient Reasoning (2026.acl-long)

Challenge:	Existing methods for large language models focus on post-training stage or inference time, while few tackle the source of toxicity, namely, the dataset itself.
Approach:	They propose to localize and rewrite toxic spans in raw corpora with SoCD, which guides an LLM to localized and preserving semantics while preserving toxicity.
Outcome:	The proposed method reduces TP from 0.42 to 0.18 and Expected Maximum Toxicity (EMT) from 0.43 to 0.20 on three LLMs.

Copied to clipboard

Yang Xiang, Yixin Ji, Ruotao Xu, Dan Qiao, Zheming Yang, Juntao Li, Min Zhang

Instant Personalized Large Language Model Adaptation via Hypernetwork (2026.acl-long)

Challenge:	Existing approaches to early exit reasoning often rely on handcrafted or empirical indicators that are unreliable and impractical.
Approach:	They propose a framework that allows LRMs to assess the sufficiency of its chain-of-thought and determine the optimal point for early exit.
Outcome:	The proposed framework reduces reasoning length by 28.9%–34.9% with minimal performance loss, effectively mitigating overthinking.

Copied to clipboard

Zhaoxuan Tan, Zixuan Zhang, Haoyang Wen, Zheng Li, Rongzhi Zhang, Pei Chen, Fengran Mo, Zheyuan Liu, Qingkai Zeng, Qingyu Yin, Meng Jiang

From Off-Policy to On-Policy: Enhancing GUI Agents via Bi-level Expert-to-Policy Assimilation (2026.acl-long)

Challenge:	Existing parameter-efficient fine-tuning methods require training a separate adapter for each user, making them computationally expensive and impractical for real-time updates.
Approach:	They propose a scalable framework that maps a user's profile directly to a full set of adapter parameters.
Outcome:	The proposed framework outperforms prompt-based personalization and OPPU while using substantially fewer computational resources at deployment.

Copied to clipboard

Zezhou Wang, Ziyun Zhang, Xiaoyi Zhang, Zhuzhong Qian, Yan Lu

Challenge:	Vision-language models are increasingly deployed as computer-use agents that operate desktops and browsers.
Approach:	They propose a method that turns static expert traces into policy-aligned guidance . they propose RLVR with a per-task, dynamically updated cache to decompose planning and execution .
Outcome:	The proposed model improves UITARS1.5-7B success from 22.87% to 32.13% on OSWorld-Verified and raises a held-out split from 5.74% to 10.30% on MMBench-GUI and Online-Mind2Web.

Safeguarding LLM Fine-tuning via Push-Pull Distributional Alignment (2026.acl-long)

Copied to clipboard

Haozhong Wang, Zhuo Li, Yibo Yang, He Zhao, Hongyuan Zha, Dandan Guo

I²B-LPO: Latent Policy Optimization via Iterative Information Bottleneck (2026.acl-long)

Challenge:	Existing safety defenses for large language models fail to explicitly repel harmful patterns . Optimal transport (SOT) allows for safe fine-tuning without sacrificing safety .
Approach:	They propose a framework that reframes safe fine-tuning from instance-level filtering challenge to distribution-level alignment task grounded in Optimal Transport.
Outcome:	a new framework improves safety of large language models while maintaining competitive performance . the proposed framework reduces the risk of errors and improves model performance compared to baselines .

Copied to clipboard

Huilin Deng, Hongchen Luo, Yue Zhu, Long Li, Zhuoyue Chen, Xinghao Zhao, Ming LI, Chuyang Zhao, Jihai Zhang, MengChang Wang, Yang Cao, Yu Kang

What Do Prosody and Text Convey? Characterizing How Meaningful Information is Distributed Across Multiple Channels (2026.acl-long)

Challenge:	Existing methods for large language model reasoning suffer from exploration collapse due to the semantic homogeneity of random rollouts.
Approach:	They propose to use latent policy optimization via iterative information bottleneck to optimize reasoning trajectories by diversifying reasoning .
Outcome:	Empirical results show that the proposed method achieves state-of-the-art performance with margins of up to 5.3% in accuracy and 7.4% in diversity metrics.

Copied to clipboard

Aditya Yadavalli, Tiago Pimentel, Tamar I Regev, Ethan Gotlieb Wilcox, Alex Warstadt

Right at My Level: A Unified Multilingual Framework for Proficiency-Aware Text Simplification (2026.acl-long)

Challenge:	Prosody—the melody of speech—conveys critical information often not captured by the words or text of a message.
Approach:	They propose an information-theoretic approach to quantify how much is conveyed by prosody that is not recoverable from text alone.
Outcome:	The proposed framework can quantify how much is conveyed by prosody that is not recoverable from text alone and crucially, what prosody conveys.

Copied to clipboard

Jinhong Jeong, Junghun Park, Youngjae Yu

See the Forest for the Trees: Loosely Speculative Decoding via Visual-Semantic Guidance for Efficient Inference of Video LLMs (2026.acl-long)

Challenge:	Existing large language model-based readability control methods rely on pre-labeled sentence corpora and primarily target English.
Approach:	They propose a framework for adaptive multilingual text simplification without parallel corpora supervision that integrates three reward modules: vocabulary coverage, semantic preservation, and coherence.
Outcome:	The proposed framework achieves higher lexical coverage at target proficiency levels while maintaining original meaning and fluency compared to stronger LLMs.

Copied to clipboard

Yicheng Ji, Jun Zhang, Jinpeng Chen, Cong Wang, Lidan Shou, Gang Chen, Huan Li

LaoBench: A Large-Scale Multidimensional Lao Benchmark for Large Language Models (2026.acl-long)

Challenge:	Existing methods for video understanding suffer from autoregressive generation of tokens.
Approach:	They propose a training-free loosely SD framework for Video-LLMs that uses visual-relevant tokens to accurately pinpoint the latter.
Outcome:	The proposed framework boosts the accepted length and speedup ratio by 136% and 35% compared to SOTA training-free SD methods for Video-LLMs.

Copied to clipboard

Jian Gao, Richeng Xuan, Zhaolu Kang, Dingshi Liao, Wenxin Huang, Zongmou Huang, Yangdi Xu, Bowen Qin, Zheqi He, Xi Yang, null Changjinli, Yonghua Lin

Gradient-Guided Multi-Judge Prompt Optimization (2026.acl-long)

Challenge:	Existing SEA-focused benchmarks miss Lao-specific cultural grounding and linguistic properties.
Approach:	They propose a multi-dimensional benchmark for assessing large language models in Lao . they use open-source and held-out subsets to evaluate languages with a hybrid pipeline .
Outcome:	LaoBench is the first large-scale, high-quality, and multidimensional benchmark for assessing LLM language understanding and reasoning in Lao.

Copied to clipboard

ChenZhuo Zhao, Xinda Wang, Pu Zhao, Yue Huang, Junting Lu, Ziqian Liu, Qingwei Lin, Saravan Rajmohan, Dongmei Zhang

ASTRA: Adaptive Semantic Tree Reasoning Architecture for Complex Table Question Answering (2026.acl-long)

Challenge:	Existing approaches to prompt optimization trade off signal quality against computational cost.
Approach:	They propose a framework that uses a first-order gradient approximation to score segment importance in a continuous masking direction.
Outcome:	The proposed framework improves efficiency and robustness by using a first-order gradient approximation to score segment importance in a continuous masking direction.

Copied to clipboard

Xiaoke Guo, Songze Li, Zhiqiang Liu, Zhaoyan Gong, Yuanxiang Liu, Huajun Chen, Wen Zhang

DEBAR: Mitigating Contextual Bias in Cross-Document Relation Extraction via Dual-Stream Decoupling (2026.acl-long)

Challenge:	Existing serialization methods fail to capture explicit hierarchies and lack schema flexibility . Existing tree-based approaches suffer from limited semantic adaptability .
Approach:	They propose a method that leverages the global semantic awareness of LLMs to reconstruct tables into Logical Semantic Trees.
Outcome:	The proposed method achieves state-of-the-art (SOTA) performance on complex table benchmarks.

Copied to clipboard

Zhixuan Yang, Fu Zhang, Huangming Xu, Jingwei Cheng

IF-RewardBench: Benchmarking Judge Models for Instruction-Following Evaluation (2026.acl-long)

Challenge:	Existing methods focus on sentence-level or singledocument settings, resulting in one-sided relation transfer contextual bias and incomplete reasoning chains.
Approach:	They propose a framework to explicitly decouple and preserve bidirectional bridge evidence and a dynamic loss optimization objective to separate head and tail contexts.
Outcome:	The proposed framework decouples and preserves bidirectional bridge evidence while capturing global dependencies through iterative message passing.

Copied to clipboard

Bosi Wen, Yilin Niu, Cunxiang Wang, Xiaoying Ling, Ying Zhang, Pei Ke, Hongning Wang, Minlie Huang

Challenge:	Existing benchmarks for instruction-following lack data coverage and oversimplified pairwise evaluation paradigms that misalign with model optimization scenarios.
Approach:	They propose a meta-evaluation benchmark for instruction-following that covers diverse instruction and constraint types and a preference graph for each instruction.
Outcome:	Extensive experiments on IF-RewardBench show that the proposed benchmark achieves a stronger positive correlation with downstream task performance compared to existing benchmarks.

LegalChainReasoner: Grounding Criminal Judicial Opinion Generation via Structured Legal Chains (2026.acl-long)

Copied to clipboard

Weizhe Shi, Qiqi Wang, Yihong Pan, Qian Liu, Kaiqi Zhao

InsideOut: Measuring and Mitigating Insider–Outsider Bias in Interview Script Generation (2026.acl-long)

Challenge:	Current legalAI tasks divide sentencing and legal reasoning into two separate tasks, resulting in inconsistency between the reasoning and predictions.
Approach:	They propose a new task that generates both legal reasoning and sentencing decisions using a framework that applies structured legal chains to guide the model through comprehensive case assessments.
Outcome:	The proposed model outperforms baseline models on real-world, open-source Chinese legal case datasets.

Copied to clipboard

Yixin Wan, Xingrun Chen, Kai-Wei Chang

Beyond Meta-Reasoning: Metacognitive Consolidation for Self-Improving LLM Reasoning (2026.acl-long)

Challenge:	Recent research has raised concerns about culture-related fairness issues in LLM-generated content.
Approach:	They propose to use 4,000 generation prompts and three evaluation metrics to quantify LLMs' insider-outsider bias .
Outcome:	The proposed method reduces bias in Llama model by 89.70% and mitigates bias on Qwen by 82.54% on cultural alignment gap metric.

Copied to clipboard

Ziqing Zhuang, Linhai Zhang, Jiasheng Si, Deyu Zhou, Yulan He

Mnemis: Dual-Route Retrieval on Hierarchical Graphs for Long-Term LLM Memory (2026.acl-long)

Challenge:	Existing approaches for improving LLM reasoning remain episodic and lack reusable meta-reasoning skills.
Approach:	They propose a framework that consolidates metacognitive experience from past reasoning episodes into reusable knowledge that improves future meta-reasoning.
Outcome:	The proposed framework consolidates metacognitive experience from past reasoning episodes into reusable knowledge that improves future meta-reasoning.

Copied to clipboard

Zihao Tang, Xin Yu, Ziyu Xiao, Zengxuan Wen, Zelin Li, Jiaxi Zhou, Hualei Wang, Haohua Wang, Haizhen Huang, Weiwei Deng, Feng Sun, Qi Zhang

Consolidation or Adaptation? PRISM: Disentangling SFT and RL Data via Gradient Concentration (2026.acl-long)

Challenge:	Existing methods for retrieving historical messages are based on similarity-based mechanisms.
Approach:	They propose a system that integrates System-1 similarity search with a complementary System-2 mechanism, termed Global Selection.
Outcome:	The proposed framework achieves state-of-the-art on long-term memory benchmarks and 93.9 on LoCoMo and 91.6 on LongMemEval-S.

Copied to clipboard

Yang Zhao, Yangou Ouyang, Xiao Ding, Hepeng Wang, Bibo Cai, Kai Xiong, Jinglong Gao, Zhouhao Sun, Li Du, Bing Qin, Ting Liu

ADVICE: Answer-Dependent Verbalized Confidence Estimation (2026.acl-long)

Challenge:	Existing data arbitration strategies for large language model training rely on surface-level heuristics that fail to diagnose intrinsic learning needs.
Approach:	They propose a framework that arbitrates data based on its degree of cognitive conflict with the model's existing knowledge.
Outcome:	Extensive experiments on WebShop and ALFWorld show that PRISM outperforms state-of-the-art hybrid methods while reducing computational costs by up to 3.22 .

Copied to clipboard

KiJung Seo, Sehun Lim, Taeuk Kim

Challenge:	Recent advances in large language models (LLMs) have enabled them to communicate their confidence in natural language, improving transparency and reliability.
Approach:	They propose a framework that promotes answer-grounded confidence estimation and analyze the dynamics of verbalized confidence estimation.
Outcome:	The proposed framework significantly improves confidence calibration while exhibiting strong generalization to unseen settings without degrading task performance.

Powering Verifiable Learning via Automated Evolutionary Data Synthesis (2026.acl-long)

Copied to clipboard

He Du, Bowen Li, Aijun Yang, Siyang He, Qipeng Guo, Kai Chen, Dacheng Tao

SPARK: Strategic Policy-Aware Exploration via Dynamic Branching for Long-Horizon Agentic Learning (2026.acl-long)

Challenge:	Existing approaches to building generalizable verifiable data are task-specific and lack a principled, universal evaluator of verifikatability.
Approach:	They propose a task-agnostic, strategy-guided, executably-checkable data synthesis framework that synthesizes problems, diverse candidate solutions and verification artifacts from a single source.
Outcome:	The proposed framework synthesizes problems, candidates, and verification artifacts from human-annotated and strategy-induced checks and iteratively discovers strategies.

Copied to clipboard

Jinyang Wu, Shuo Yang, Yuhao Shen, Shuai Zhang, Zhengqi Wen, Jianhua Tao

Challenge:	Existing methods for training large language models waste computation budget on trivial steps while failing to guarantee sample quality.
Approach:	They propose a framework that selectively branches at critical decision states for resource-efficient exploration.
Outcome:	The proposed framework activates adaptive branching exploration at critical decision states to probe promising trajectories, thereby achieving precise resource allocation that prioritizes sampling quality over blind coverage.

Scaling Laws for Code: A More Data-Hungry Regime (2026.acl-long)

Copied to clipboard

Xianzhen Luo, Wenzhen Zheng, Qingfu Zhu, Rongyi Zhang, Houyi Li, Siming Huang, YuanTao Fan, Wanxiang Che

Challenge:	Code Large Language Models (LLMs) are revolutionizing software engineering, but scaling laws are primarily analyzed on Natural Language (NL).
Approach:	They fit Chinchilla law and Farsser law to test scaling laws for code . they find code is more data-hungry and requires higher data-to-parameter ratio .
Outcome:	The proposed scaling laws show that the more expressive Farsser law offers greater accuracy and scales with model size.

When TableQA Meets Noise: A Dual Denoising Framework for Complex Questions and Large-scale Tables (2026.acl-long)

Copied to clipboard

Shenghao Ye, Yu Guo, Dong Jin, Yuxiang Wang, Yikai Shen, Yunpeng Hou, Shuangwu Chen, null Jianyang, Xiaofeng Jiang

VRPO: Rethinking Value Modeling for Robust RL under Noisy Supervision in LLM Post-Training (2026.acl-long)

Challenge:	Extensive research shows that noisy data significantly degrades the performance of table reasoning in real-world applications.
Approach:	They propose a dual denoising framework for complex questions and large-scale tables that uses Tree-guided table pruning to remove irrelevant data step by step.
Outcome:	The proposed framework achieves outstanding performance on TableQA tasks with complex questions and large-scale tables.

Copied to clipboard

Dingwei Zhu, Shihan Dou, Zhiheng Xi, Senjie Jin, Guoqiang Zhang, Jiazheng Zhang, Junjie Ye, Mingxu Chai, Enyu Zhou, Ming Zhang, Yuhui Wang, Caishuang Huang, Chenhao Huang, Yunke Zhang, Yuran Wang, Tao Gui, Qi Zhang, Xipeng Qiu, Xuanjing Huang

Taming Actor-Observer Asymmetry in Agents via Dialectical Alignment (2026.acl-long)

Challenge:	Reinforcement Learning (RL) in real-world environments often suffers from ambiguous or incomplete supervision.
Approach:	They propose a framework that enhances value modeling for robust RL in LLM post-training by integrating auxiliary losses guided by entropy and perplexity from a frozen language model and variational information bottleneck.
Outcome:	The proposed framework outperforms baselines on multi-turn dialogue, math reasoning, and science QA with rule-based and model-based rewards.

Copied to clipboard

Bobo Li, Wu Rui, Zibo Ji, Meishan Zhang, Hao Fei, Min Zhang, Mong-Li Lee, Wynne Hsu

Knowledge Beyond Language: Bridging the Gap in Multilingual Machine Unlearning Evaluation (2026.acl-long)

Challenge:	Large Language Models (LLMs) have enabled the development of powerful autonomous systems.
Approach:	They propose a model trained through dialectical alignment to enforce perspective-invariant reasoning.
Outcome:	The proposed model mitigates attribution inconsistency and significantly improves fault resolution rates in ambiguous scenarios.

Copied to clipboard

Kyomin Hwang, Hyeonjin Kim, Sangyeon Cho, Nojun Kwak

AT²PO: Agentic Turn-based Policy Optimization via Tree Search (2026.acl-long)

Challenge:	Existing methods to evaluate knowledge independently for each language fail to capture cross-linguistic distribution of information.
Approach:	They propose to use two metrics to evaluate the information spread across languages to measure unlearning quality and consistency.
Outcome:	The proposed methods provide insights into unique phenomena exclusive to MMU and offer a new perspective on MMU evaluation.

Copied to clipboard

Zefang Zong, Dingwei Chen, Yang Li, Qi Yi, Bo Zhou, Chengming Li, BO Qian, Peng Chen, Jie Jiang

SecureVibeBench: Benchmarking Secure Vibe Coding of AI Agents via Reconstructing Vulnerability-Introducing Scenarios (2026.acl-long)

Challenge:	Recent advances in large language models (LLMs) have catalyzed the development of autonomous agents capable of executing complex, multi-turn tasks.
Approach:	They propose a framework for agentic reinforcement learning that integrates turn-level tree search with tree search to address key challenges.
Outcome:	The proposed framework addresses key challenges: limited exploration diversity, sparse credit assignment, and misaligned policy optimization.

Copied to clipboard

Junkai Chen, Huihui Huang, Yunbo Lyu, Junwen An, Jieke Shi, Chengran Yang, Ting Zhang, Haoye Tian, Yikun Li, Zhenhao Li, Xin Zhou, Xing Hu, David Lo

Challenge:	Existing benchmarks fail to capture scenarios in which vulnerabilities are introduced by humans . we evaluate 5 popular code agents supported by 5 LLMs on SecureVibeBench .
Approach:	They propose a benchmarking tool that compares 105 C/C++ secure coding tasks . they use real-world open-source vulnerabilities and a comprehensive evaluation tool .
Outcome:	The proposed benchmarks show that code agents struggle to produce correct and secure code . the best performing agent produces merely 23.8% correct and secured solutions .

Text-Guided Multi-Scale Frequency Representation Adaptation (2026.acl-long)

Copied to clipboard

Weicai Yan, Xinhua Ma, Wang Lin, Tao Jin

ADAPT: Benchmarking Commonsense Planning under Unspecified Affordance Constraints (2026.acl-long)

Challenge:	Existing methods for fine-tuning visual signals are limited by their size and complexity.
Approach:	They propose a multi-scale frequency-based fine-tuning method that integrates textual information and performs multi-level fine- tuning of visual signals in the frequency domain.
Outcome:	Extensive experiments on multimodal models, including CLIP and LLaVA, demonstrate that the proposed method significantly improves performance and efficiency with minimal cost and fast convergence within one epoch.

Copied to clipboard

Pei-An Chen, Yongching Liang, Jia-Fong Yeh, Hung-Ting Su, Yi-Ting Chen, Min Sun, Winston H. Hsu

GeoRA: Geometry-Aware Low-Rank Adaptation for RLVR (2026.acl-long)

Challenge:	Existing methods for embodied agents focus on directly executing instructions without considering whether objects can be manipulated.
Approach:	They propose a benchmark that evaluates embodied agents in dynamic environments . they use plug-and-play module that augments existing planners with explicit affordance reasoning .
Outcome:	The proposed benchmark evaluates embodied agents in dynamic environments with unpredictable affordances . ADAPT significantly improves robustness and task success across seen and unseen environments .

Copied to clipboard

Jiaying Zhang, Lei Shi, Jiguo Li, Jun Xu, Jiuchong Gao, Jinghua Hao, Renqing He

KoCo: Conditioning Language Model Pre-training on Knowledge Coordinates (2026.acl-long)

Challenge:	Existing parameter-efficient methods for RLVR face limitations . low-rank adaptation methods do not account for the distinct optimization dynamics .
Approach:	They propose a low-rank adaptation method tailored for RLVR that exploits the anisotropic structure of RL update subspace and extracts its principal directions via Singular Value Decomposition (SVD).
Outcome:	Experiments on large reasoning models show that GeoRA outperforms strong low-rank baselines across RLVR settings while showing stronger generalization and less forgetting on out-of-domain tasks.

Copied to clipboard

Yudong Li, Jiawei Cai, Linlin Shen

Truth or Sophistry? LoFa: A Benchmark for LLM Robustness Against Logical Fallacies (2026.acl-long)

Challenge:	Standard Large Language Model (LLM) pretraining treats corpora as flattened token sequences . a new method that maps every document into a three-dimensional semantic coordinate can bridge this gap .
Approach:	They propose a method that maps every document into a three-dimensional semantic coordinate . they say it equips the model with explicit contextual awareness to learn the documents .
Outcome:	Experiments show that knowledge coordinates help model distinguish stable facts from noise . authors say that the method significantly improves performance across 10 downstream tasks .

Copied to clipboard

Xudong Shen, li Yuan, Ye Chen, Xin Wu, Yi Cai, Zhiyong Wu

Merging Triggers, Breaking Backdoors: Defensive Poisoning for Instruction-Tuned Language Models (2026.acl-long)

Challenge:	Prior work has focused on the ability of Large Language Models to identify or classify fallacies, but their robustness against these fallacias in persuasive contexts remains largely unexplored.
Approach:	They propose a new metric to assess LLM robustness against fallacies by pairing factual questions with fallacious arguments and developing a multi-round debate framework to assess model resilience.
Outcome:	The proposed metric disentangles robustness from a model’s knowledge limitations and demonstrates unique vulnerability profiles across models.

Copied to clipboard

San Kim, Gary Lee

SHAPE: Stage-aware Hierarchical Advantage via Potential Estimation for LLM Reasoning (2026.acl-long)

Challenge:	Large Language Models (LLMs) are vulnerable to backdoor attacks, where adversaries poison a small subset of data to implant hidden behaviors.
Approach:	They propose a training pipeline that immunizes instruction-tuned LLMs against backdoor attacks.
Outcome:	The proposed defenses lower attack success rates while preserving instruction-following ability.

Copied to clipboard

Zhengyang Ai, Zikang Shan, Xiaodong Ai, Jingxian Tang, Hangkai Hu, Pinyan Lu

Looking Beyond the One: Operationalizing and Eliciting Visual Ambiguity in VLLMs (2026.acl-long)

Challenge:	Existing methods for process supervision fail to distinguish meaningful progress from mere verbosity . existing methods lack a coherent approach to process supervision .
Approach:	They propose a framework that formalizes reasoning as a trajectory through a state space of empirical solvability.
Outcome:	The proposed framework achieves an average accuracy gain of 3% with 30% reduced token consumption.

Copied to clipboard

Yuchong Chen, Bowei Zou, Yuhan Chen, Yifan Fan, Xinyu Li, Shujun Cao, Yu Hong

JanusMM: A Benchmark for Self-Deprecation Understanding in Real-World Multimodal Conversations (2026.acl-long)

Challenge:	Visual question answering systems typically collapse ambiguity, committing to a single interpretation during decoding and evaluation.
Approach:	They operationalize ambiguity as the existence of multiple answer-supporting regions in an image . they show that ambiguities are already encoded in their internal representations .
Outcome:	The proposed approach makes ambiguity observable without exhaustive annotations . ambiguities are already encoded in models, but not reliably expressed in outputs despite hidden states .

Copied to clipboard

Xinyi Xu, Bingguang Hao, Yongyi Xiong, Zimo Chen, Xinchen Liu, Hongxin Guo, Xuelong Wang, Silin Zhou, Shihan Dou

Attention as Selector: Unlocking VLM Attention for Long Document Page Retrieval (2026.acl-long)

Challenge:	Multimodal large language models (MLLMs) are a common communicative strategy in human society, often using image-text interplay to express emotions and intentions.
Approach:	They propose to evaluate multimodal large language models (MLLMs)' understanding of self-deprecation in real-world conversations using 2,016 bilingual memes.
Outcome:	The proposed framework evaluates MLLMs' understanding of self-deprecation in real-world conversations.

Copied to clipboard

Minfeng Zhu, Linxin Bao, Wei Chen, Linchao Zhu

LAMCL: A Length-aware Momentum Contrastive Learning Framework for Multiscale Machine-Revised Text Detection (2026.acl-long)

Challenge:	Existing page-level retrieval methods lack query–page interaction before similarity scoring . Existing methods require large-scale datasets to align visual and textual embeddings .
Approach:	They propose a retrieval framework that utilizes attention mechanisms inside VLMs for page selection.
Outcome:	The proposed retrieval framework outperforms embedding-based retrieval methods on four long-document benchmarks.

Copied to clipboard

Bing Zhou, Zhe Huang, Shilei Tan, Kai Zhao, Zhou Yongcheng

Ascending the Infinite Ladder: Benchmarking Spatial Deformation Reasoning in Vision-Language Models (2026.acl-long)

Challenge:	Recent detection methods struggle to capture fine-grained semantic differences, especially for short texts.
Approach:	They propose a framework for machine-revised text detection that integrates two modules to enhance discriminative semantic features.
Outcome:	The proposed method outperforms existing detectors in identifying machine-revised text across diverse practical scenarios, tasks, and LLMs.

Copied to clipboard

Jiahuan Zhang, Shunwen Bai, Tianheng Wang, KaiWen Guo, Zijia Song, Hanqing WU, Guozheng Rao, Kai Han, Kaicheng Yu

Safe-FedLLM: Delving into the Safety of Federated Large Language Models (2026.acl-long)

Challenge:	Existing benchmarks explore aspects of threedimensional spatial reasoning and visual-language reasoning in dynamic environments, but they are unable to perform well on 3D spatial deformation reasoning.
Approach:	They propose to use a ladder competition format to assess the model's spatial deformation reasoning abilities to determine its performance.
Outcome:	The proposed framework assesses the performance of Vision-Language Models in spatial deformation reasoning tasks.

Copied to clipboard

Mingxiang Tao, Yu Tian, Wenxuan Tu, Yue Yang, Xue Yang, Xiangyan Tang

TEMA: Anchor the Image, Follow the Text for Multi-Modification Composed Image Retrieval (2026.acl-long)

Challenge:	Existing work on federated learning for large language models (FL) addresses privacy and data-silo issues in the training of large language model training.
Approach:	They propose a probe-based defense framework for FedLLM that constructs defenses across three levels: Step-Level, Client-Level and Shadow-Level.
Outcome:	The proposed framework improves FedLLM's robustness against malicious clients while maintaining competitive performance on benign data.

Copied to clipboard

Zixu Li, Yupeng Hu, Zhiheng Fu, Zhiwei Chen, Yongqi Li, Liqiang Nie

ImCoref-CeS: An Improved Lightweight Pipeline for Coreference Resolution with LLM-based Checker-Splitter Refinement (2026.acl-long)

Challenge:	Composed Image Retrieval (CIR) is an image retrieval paradigm that enables users to retrieve a target image using a multimodal query that consists of a reference image and modification text.
Approach:	They propose a text-oriented entity mapping architecture that allows users to use a reference image and modification text to retrieve a target image.
Outcome:	The proposed framework is superior in both original and multi-modification scenarios while maintaining an optimal balance between retrieval accuracy and computational efficiency.

Copied to clipboard

Kangyang Luo, Yuzhuo Bai, Shuzheng Si, Cheng Gao, Zhitong Wang, Yingli Shen, Wenhao Li, Zhu Liu, Yufeng Han, Jiayi Wu, Cunliang Kong, Maosong Sun

When Efficiency Meets Safety: A Benchmark Security Analysis of KV Cache Compression in Large Language Models (2026.acl-long)

Challenge:	Existing supervised neural methods for coreference resolution are underexplored . current methods rely on small language models, but their potential is underexploited .
Approach:	They propose a framework that integrates an enhanced supervised model with LLM-based reasoning.
Outcome:	The proposed method surpasses existing state-of-the-art methods in coreference resolution.

Copied to clipboard

Xiaoxiao Ma, Kuofeng Gao, Zeyi Lu, Wenxi Jiang, Hao Fang, Hao Wu, Bin Chen, Shu-Tao Xia

WebAggregator: Enhancing Compositional Reasoning Capabilities of Deep Research Agent Foundation Models (2026.acl-long)

Challenge:	Key-Value (KV) caching is widely used in large language models to enable long-context inference efficiently, yet its security implications remain underexplored.
Approach:	They propose a history-aware, per-head feedback merging strategy that prevents safety degradation while maintaining efficiency.
Outcome:	The proposed strategy prevents safety degradation while maintaining efficiency.

Copied to clipboard

Rui Wang, Ce Zhang, Jun-Yu Ma, Jianshu Zhang, Hongru Wang, Yi Chen, Boyang Xue, Tianqing Fang, Zhisong Zhang, Hongming Zhang, Haitao Mi, Dong Yu, Kam-Fai Wong

SEARL: Joint Optimization of Policy and Tool Graph Memory for Self-Evolving Agents (2026.acl-long)

Challenge:	Existing agentic systems are retrieval-heavy but reasoning-light . current systems lack compositional reasoning, a key component of deep research .
Approach:	They propose a data synthesis pipeline WebAggregator to shift agentic paradigm . they use Proactive Explorer to collect interconnected knowledge and Compositional Logic Proposer to weave knowledge into complex questions .
Outcome:	The proposed pipeline surpasses GPT-4.1 and matches Claude-3.7-Sonnet on GAIA, WebWalkerQA, and XBench.

Copied to clipboard

Xinshun Feng, Xinhao Song, Lijun Li, Gongshen Liu, Jing Shao

Good Arguments Against the People Pleasers: How Reasoning Mitigates (Yet Masks) LLM Sycophancy (2026.acl-long)

Challenge:	Recent advances in Reinforcement Learning with Verifiable Rewards (RLVR) have demonstrated significant potential in single-turn reasoning tasks.
Approach:	They propose a tool-memory based self-evolving agentic framework that integrates planning with execution.
Outcome:	The proposed framework is able to extract explicit knowledge from historical data and leverage inter-trajectory correlations to densify reward signals.

Copied to clipboard

Zhaoxin Feng, Zheng Chen, Jianfei Ma, Yip Tin Po, Emmanuele Chersoni, Bo Li

Can Persona-Prompted LLMs Emulate Subgroup Values? An Empirical Analysis of Generalisability and Fairness in Cultural Alignment (2026.acl-long)

Challenge:	Recent studies have identified a critical drawback of aligning models with human judgments and outputs that are flawed or incorrect.
Approach:	They evaluate a range of LLMs to examine whether CoT reasoning mitigates sycophancy . they find that reasoning masks a tendency to scophage in some cases .
Outcome:	The proposed model models show that CoT reasoning reduces sycophancy but masks it in some cases.

Copied to clipboard

Bryan Chen Zhengyu Tan, Zhengyuan Liu, Xiaoyuan Yi, Jing Yao, Xing Xie, Nancy F. Chen, Roy Ka-Wei Lee

FoE: Forest of Errors Makes the First Solution the Best in Large Reasoning Models (2026.acl-long)

Challenge:	Current alignment paradigms treat "human values" as a monolithic entity, ignoring the fact that many societies are a mosaic of diverse subgroups with distinct and sometimes conflicting values, preferences, and norms.
Approach:	They examine whether Large Language Models can emulate distinct cultural values of subgroups . they use a global value survey to examine the value landscape of a multicultural society .
Outcome:	The proposed model improves on unseen, out-of-distribution subgroups by 17.4% . the model widens the disparity between subgroup groups when measured by distance-aware metrics.

Copied to clipboard

Kehan Jiang, Haonan Dong, Zhaolu Kang, Zhengzhou Zhu, Guojie Song

Lingua-Graph: A Unified Representation of Cross-Task Common Substructures for Analytic Language Processing (2026.acl-long)

Challenge:	Recent Large Reasoning Models (LRMs) have demonstrated remarkable success in complex reasoning tasks.
Approach:	They propose a self-guided efficient reasoning framework that reduces FoE by pruning subs.
Outcome:	The proposed model outperforms eight competitive baselines while reducing token consumption by 37.7% 70.4%.

Copied to clipboard

Mingming Sun, Runze Jiang, Zhu Zhangchenxi, Minlong Peng, Yunfeng Cai

Adaptive Spatial and Temporal Redundancy Optimization for Efficient Reasoning in Large Language Models (2026.acl-long)

Challenge:	Existing structural-analytic tasks are fragmented by inconsistent task requirements . we propose a solution for the representation layer, called Lingua-Graph .
Approach:	They propose a representation-then-decision paradigm for structural-analytic tasks . they propose Graph-based representations that capture entities, facts, and relations .
Outcome:	The proposed model improves interpretability and higher hostability of entities, facts, and relations . the proposed model is available on github.com/rudaoshi/Lingua .

Copied to clipboard

Tianle Chen, Pengyu Cheng, Qiyuan Zhu, Jiacheng Wang, Bei Liu, Hao Gu, Ruijie Shen, Xiaofeng Hou, Sirui Han, Jiacheng Liu

AAPO: Enhancing the Reasoning Capabilities of LLMs with Advantage Margin (2026.acl-long)

Challenge:	Existing research to improve CoT efficiency falls into three categories, each with distinct limitations.
Approach:	They propose a training-free framework that addresses both dimensions of CoT reasoning by applying a progressive precision reduction strategy coupled with an entropy-based confidence mechanism for adaptive termination.
Outcome:	Empirical results show that the proposed framework achieves 11.3 efficiency gain without compromising accuracy.

Copied to clipboard

Jian Xiong, Jingbo Zhou, Jingyong Ye, Qiang Huang, Dejing Dou

Outcome-Grounded Advantage Reshaping for Fine-Grained Credit Assignment in Mathematical Reasoning (2026.acl-long)

Challenge:	Reinforcement learning (RL) has emerged as an effective approach for enhancing the reasoning capabilities of large language models.
Approach:	They propose an algorithm that optimizes cross-entropy loss using advantages enhanced through a margin-based estimation scheme.
Outcome:	Experimental results show that AAPO improves group relative advantage estimation compared to other methods.

Copied to clipboard

Ziheng Li, Liu Kang, Feng Xiao, Luxi Xing, Qingyi Si, Zhuoran Li, Weikang Gong, Deqing Yang, Yanghua Xiao, Hongcheng Guo

Learning to Edit Knowledge via Instruction-based Chain-of-Thought Prompting (2026.acl-long)

Challenge:	Group Relative Policy Optimization (GRPO) uses a coarse-grained credit assignment mechanism that propagates group-level rewards uniformly to to every token in a sequence, neglecting the varying contribution of individual reasoning steps.
Approach:	They introduce Outcome-grounded Advantage Reshaping (OAR) which redistributes advantages based on how much each token influences the model’s final answer.
Outcome:	Empirical results show that OAR-G outperforms GRPO on a high-fidelity attribution signal and suppresses low-impact tokens while preserving the advantage mass.

Copied to clipboard

Jinhu Fu, Yan Bai, Longzhu He, Yihang Lou, Yanxiao Zhao, Li Sun, Sen Su

BadScientist: Can a Research Agent Write Convincing but Unsound Papers that Fool LLM Reviewers? (2026.acl-long)

Challenge:	Existing knowledge editing methods focus on structured fact triples, overlooking diverse unstructured forms of factual information.
Approach:	They propose a method that allows LLMs to edit knowledge via Chain of Thoughts reasoning.
Outcome:	The proposed method achieves strong generalization across six diverse knowledge editing scenarios with a single round of training on three open-source language models.

Copied to clipboard

Fengqing Jiang, Yichen Feng, Yuetai Li, Luyao Niu, Basel Alomair, Radha Poovendran

Challenge:	Existing evidence suggests that LLMs are not able to detect scientifically unsound work from malicious or poorly designed research agents.
Approach:	They develop a framework that evaluates whether fabrication-oriented paper generation agents can deceive multi-model LLM review systems.
Outcome:	The proposed framework shows that fabricated papers achieve acceptance rates up to 18% . the framework shows only marginal improvements, with detection accuracy barely exceeding random chance.

DFAMS: Dynamic-flow guided Federated Alignment based Multi-prototype Search (2026.acl-long)

Copied to clipboard

Zhibang Yang, Xinke Jiang, Rihong Qiu, Ruiqing Li, Yihang Zhang, Yue Fang, Yongxin Xu, Hongxin Ding, Xu Chu, Junfeng Zhao, Yasha Wang

PARIF: Pushing the Pareto Frontier of Instruction Following and Reasoning with Curriculum Reinforcement Learning (2026.acl-long)

Challenge:	Existing methods for ambiguous queries struggle to retrieve high-quality documents . DFAMS outperforms advanced FR methods by 14.37% in knowledge classification accuracy .
Approach:	They propose a framework that leverages dynamic information flow to identify latent query intents and construct semantically aligned knowledge partitions for accurate retrieval across heterogeneous sources.
Outcome:	The proposed framework outperforms existing methods in classification accuracy and retrieval recall tests.

Copied to clipboard

Rongchuan Mu, Zexin Wang, Qianyu Wang, MingHua Ma, Zekun Wang, Ming Liu, Bing Qin

GeoLaux: A Benchmark for Evaluating MLLMs’ Geometry Performance on Long-Step Problems Requiring Auxiliary Lines (2026.acl-long)

Challenge:	Existing alignment methods struggle to balance general reasoning with instruction-following (IF) this is hindered by dependency on teacher models, reward hacking, and reasoning-answer inconsistencies.
Approach:	They propose a two-stage curriculum learning framework based on Reinforcement Learning from Verifiable Rewards to enhance both IF and general reasoning capabilities.
Outcome:	The proposed framework outperforms leading models on six representative IF tasks while achieving a 21.25% relative average improvement over the original model.

Copied to clipboard

Yumeng Fu, Jiayin Zhu, Lingling Zhang, Wenjun Wu, Bo Zhao, Shaoxuan Ma, Yushun Zhang, Jun Liu

Unlocking Multilingual Reasoning Capability of LLMs and LVLMs through Representation Engineering (2026.acl-long)

Challenge:	Existing benchmarks for Geometry problem solving lack fine-grained evaluation for long-step problems necessitating auxiliary line construction.
Approach:	They present a fine-grained annotated dataset with long-step reasoning and auxiliary line construction that provides a detailed evaluation of 23 leading MLLMs.
Outcome:	The proposed model performs significantly worse on long-step problems than short-step ones, with 18 models showing a performance drop of over 50%.

Copied to clipboard

Qiming Li, Xiaocheng Feng, Yixuan Ma, Ruihan Chen, Zihe Tong, Zekai Ye, Xiachong Feng, Libo Qin, Haoyu Ren, Kun Chen, Yunfei Lu, Dandan Tu, Bing Qin

Into the Gray Zone: Domain Contexts Can Blur LLM Safety Boundaries (2026.acl-long)

Challenge:	Existing approaches to enhance multilingual reasoning capabilities rely on costly multilingual training or employ prompting with external translation tools.
Approach:	They propose a training-free inference-time method to enhance multilingual reasoning capabilities via Representation Engineering without additional training data or tools.
Outcome:	The proposed method outperforms existing methods on four reasoning benchmarks in English and Thai and Swahili.

Copied to clipboard

Ki Sen Hung, Xi Yang, Chang Liu, Haoran Li, Kejiang Chen, Changxuan Fan, Tsun On Kwok, Weiming Zhang, Xiaomeng Li, Yangqiu Song

MED-COREASONER: Reducing Language Disparities in Medical Reasoning via Language-Informed Co-Reasoning (2026.acl-long)

Challenge:	a goal of LLM alignment is to balance usefulness with harmlessness, but this conflictes when knowledge serves both legitimate and malicious purposes.
Approach:	They propose a framework that combines safety-research contexts with adversarial interactions to exploit a vulnerability in Jargon queries.
Outcome:	a framework outperforms existing methods in analyzing Jargon queries, a study shows . it achieves 93% of attacks across seven models, while remaining useful, the authors say .

Copied to clipboard

Fan Gao, Sherry T. Tong, Jiwoong Sohn, Jiahao Huang, Junfeng Jiang, Ding Xia, Piyalitt Ittichaiwong, Kanyakorn Veerakanjana, Hyunjae Kim, Qingyu Chen, Edison Marrese-Taylor, Kazuma Kobayashi, Akiko Aizawa, Irene Li

Psyche-R1: Towards Reliable Psychological LLMs through Unified Empathy, Expertise, and Reasoning (2026.acl-long)

Challenge:	Existing models that use English and local languages have a multilingual gap . a language-informed co-reasoning framework can be used to improve multilingual reasoning .
Approach:	They propose a language-informed co-reasoning framework that elicits parallel English and local-language reasoning and abstracts them into structured concepts.
Outcome:	Experiments show that Med-CoReasoner improves multilingual reasoning performance by 5% . the framework produces clinically sound and culturally grounded reasoning traces .

Copied to clipboard

Chongyuan Dai, Jinpeng Hu, Hongchang Shi, Zhuo Li, Dan Guo, Xun Yang, Meng Wang

Mitigating Action-Relation Hallucinations in LVLMs via Relation-aware Visual Enhancement (2026.acl-long)

Challenge:	Recent reasoning-augmented LLMs have demonstrated impressive capabilities across a wide range of domains owing to their exceptional text understanding capabilities.
Approach:	They propose a Chinese psychological LLM that integrates empathy, psychological expertise, and reasoning.
Outcome:	The proposed model produces over 75k high-quality psychological questions paired with detailed rationales, generated through and iterative prompt-rationale optimization procedure, along with 73k empathetic dialogues.

Copied to clipboard

Zhenxin Qin, Qiang Li, Qingzhuo Wang, Ruiyang Qin, Zhihua Wei, Wen Shen

Agentic Oversight via Dialectic Reasoning (2026.acl-long)

Challenge:	Existing research has focused on mitigating object hallucinations but often overlooks more complex relation hallucines, especially action relations involving interactions between objects.
Approach:	They propose a framework to locate action-relevant image regions and enhance the LVLM’s attention to those regions by using a Relation-aware Visual Enhancement method.
Outcome:	The proposed method achieves superior performance in mitigating action-relation hallucinations with negligible additional inference cost.

Copied to clipboard

Leonardo Ranaldi, Federico Ranaldi

FOREVER: Forgetting Curve-Inspired Memory Replay for Language Model Continual Learning (2026.acl-long)

Challenge:	Existing approaches to align Large Language Models (LLMs) rely heavily on human annotations, but a Debate between expert models is a promising oversight mechanism.
Approach:	They propose a Debate between expert models to enable scalable oversight . they use a reasoning function to extend the framework to multilingual and multimodal spaces .
Outcome:	The proposed framework outperforms single-expert baselines in six multilingual and multimodal scenarios and shows that argument-mediated supervision instils unsupervised reasoning signals in expert models.

Copied to clipboard

Yujie Feng, Hao Wang, Jian Li, Xu Chu, Zhaolu Kang, Yiran Liu, Yasha Wang, Philip S. Yu, Xiao-Ming Wu

CharTide: Data-Centric Chart-to-Code Generation via Tri-Perspective Tuning and Inquiry-Driven Evolution (2026.acl-long)

Challenge:	Continual learning (CL) for large language models (LLMs) aims to enable sequential knowledge acquisition without catastrophic forgetting.
Approach:	They propose a framework that aligns replay schedules with a model-centric notion of time.
Outcome:	Experiments on three benchmarks show that FOREVER consistently mitigates catastrophic forgetting.

Copied to clipboard

Xiangxi Zheng, Kuang He, Jiayi Hu, Ping Yu, Rui Yan, Yuan Yao, Peng Hou, Anxiang Zeng, Alex Jinpeng Wang

PRISP: Privacy-Safe Few-Shot Personalization via Lightweight Adaptation (2026.acl-long)

Challenge:	Existing approaches to chart-to-code generation are constrained by data-centric limitations . authors present a new framework that redesigns both training and alignment data .
Approach:	They propose a data-centric framework that redesigns both training and alignment data for chart-to-code generation.
Outcome:	The proposed framework outperforms open-source baselines and is competitive with GPT-5.

Copied to clipboard

Junho Park, Dohoon Kim, Taesup Moon

IF-CRITIC: Towards a Fine-Grained LLM Critic for Instruction-Following Evaluation (2026.acl-long)

Challenge:	Existing methods for large language model personalization are limited by data-rich settings and privacy risks.
Approach:	They propose a lightweight and privacy-safe personalization framework tailored to constraints in large language models.
Outcome:	Experiments on a few-shot variant of the LaMP benchmark show that PRISP achieves strong overall performance compared to prior approaches.

Copied to clipboard

Bosi Wen, Yilin Niu, Cunxiang Wang, Pei Ke, Xiaoying Ling, Ying Zhang, Aohan Zeng, Hongning Wang, Minlie Huang

Challenge:	Existing evaluation models for instruction-following have many shortcomings, such as substantial costs and unreliable assessments.
Approach:	They propose an LLM critic for fine-grained instruction-following evaluation using a checklist generator and a constraint-level preference optimization method.
Outcome:	The proposed model beats strong LLM-as-a-Judge baselines in evaluations under lower computational overhead compared to baselines.

Know Your Place: Diagnosing Implicit Social Adaptation Failures in Chinese Large Language Models (2026.acl-long)

Copied to clipboard

Yu Tian, Jie Xing, Ziming Li, Jiang Li, Zehua Duo, Tian Lan, Xu Liu, Guanglai Gao, Xiangdong Su

Agent Newsroom: Efficient Chronological Report Generation via Dynamic Multi-Agent Collaboration (2026.acl-long)

Challenge:	Existing studies suggest that failures of large language models in social contexts are not due to limited linguistic competence, but to inappropriate recognition.
Approach:	They propose a framework that decomposes social adaptation into three orthogonal dimensions and conduct controlled comparisons across multiple Chinese LLMs under implicit and explicit conditions.
Outcome:	The proposed framework decomposes social adaptation into three orthogonal dimensions and conducts controlled comparisons across multiple Chinese LLMs under implicit and explicit conditions.

Copied to clipboard

Zhenhua Wang, Chunlei Wang, Yue Geng, Bang Wang

Locomo-Plus: Beyond-Factual Cognitive Memory Evaluation Framework for LLM Agents (2026.acl-long)

Challenge:	Large language models (LLMs) improve event synthesis, but most are monolithic and often process overlapping evidence with bursty reporting patterns.
Approach:	They propose a multi-agent framework that casts TLS as a newsroom-like collaboration.
Outcome:	Experiments on three benchmarks show that MAS-TLS improves semantic coverage and temporal grounding while substantially reducing token usage and latency.

Copied to clipboard

Yifei Li, Weidong Guo, Lingling Zhang, Rongman Xu, Muye Huang, Hui Liu, Lijiao Xu, Yu Xu, Jun Liu

Challenge:	Existing benchmarks and evaluation protocols focus on surface-level factual recall.
Approach:	They propose a benchmark for assessing cognitive memory under cue–trigger semantic disconnect.
Outcome:	The proposed framework reveals failures not captured by existing benchmarks.

CUB: Benchmarking Context Utilisation Techniques for Language Models (2026.acl-long)

Copied to clipboard

Lovisa Hagström, Youna Kim, Haeun Yu, Sang-goo Lee, Richard Johansson, Hyunsoo Cho, Isabelle Augenstein

VLN-NF: Feasibility-Aware Vision-and-Language Navigation with False-Premise Instructions (2026.acl-long)

Challenge:	Existing language models (LMs) can be distracted by irrelevant contexts or ignore relevant information that contradicts outdated parametric memory.
Approach:	They develop a benchmark to help diagnose CMTs under diverse noisy context conditions within retrieval-augmented generation (RAG) they find that most existing CMT struggle to handle the full spectrum of context types encountered in real-world RAG scenarios.
Outcome:	The proposed benchmark compares seven state-of-the-art methods across three datasets and tasks, and shows that many lack the robustness needed to handle the full spectrum of context types encountered in real-world RAG scenarios.

Copied to clipboard

Hung-Ting Su, Ting-Jun Wang, Jia-Fong Yeh, Min Sun, Winston H. Hsu

NiuTrans.LMT: Toward Inclusive and Scalable Multilingual Machine Translation with LLMs (2026.acl-long)

Challenge:	Existing Vision-and-Language Navigation benchmarks assume instructions are feasible and the referenced target exists.
Approach:	They propose a benchmark with false-premise instructions where the target is absent . they propose supervised room-level navigation with LLM/VLM-driven in-room exploration .
Outcome:	The proposed benchmark produces false-premise goals that are plausible but factually incorrect . ROAM achieves the best REV-SPL among compared methods, while baselines often under-explore and terminate prematurely under unreliable instructions.

Copied to clipboard

Yingfeng Luo, Ziqiang Xu, Yuxuan Ouyang, MuRun Yang, DingYang Lin, Kaiyan Chang, Tong Zheng, Bei Li, Peinan Feng, Quan Du, Tong Xiao, JingBo Zhu

ChipSeek: Optimizing Verilog Generation via EDA-Integrated Reinforcement Learning (2026.acl-long)

Challenge:	Large language models have significantly advanced Multilingual Machine Translation (MMT) yet scaling to many languages while maintaining robust performance across directions remains challenging.
Approach:	They propose a strategy to reduce the number of translations in one direction . they propose auxiliary parallel sentences to promote cross-lingual transfer .
Outcome:	The proposed model performs on par with or better than substantially larger baselines.

Copied to clipboard

Zhirong Chen, Kaiyan Chang, Zhuolin Li, Cangyuan Li, Xinyang He, Chujie Chen, Mengdi Wang, Haobo Xu, Yinhe Han, Huawei Li, Ying Wang

MERIT Feedback Elicits Better Bargaining in LLM Negotiators (2026.acl-long)

Challenge:	Existing approaches to optimize Register-Transfer Level (RTL) code fail to simultaneously optimize functional correctness and hardware efficiency metrics such as Power, Performance, and Area (PPA).
Approach:	They propose a hierarchical reward based reinforcement learning framework that integrates direct feedback from EDA simulators and synthesis tools into a reward mechanism.
Outcome:	The proposed framework integrates direct feedback from EDA simulators and synthesis tools into a hierarchical reward based reinforcement learning framework.

Copied to clipboard

Jihwan Oh, Murad Aghazada, Yooju Shin, Se-Young Yun, Taehyeon Kim

Challenge:	Empirical results indicate that baseline LLM strategies diverge from human preferences, while our mechanism substantially improves negotiation performance.
Approach:	They propose a utility feedback centric framework that measures human-aligned, economically grounded metrics that implicitly measure how well the negotiation aligns with human preference.
Outcome:	The proposed framework significantly improves negotiation performance, yielding deeper strategic behavior and stronger opponent awareness.

Collision to Cognition: Hash-Driven Graph Construction for Efficient RAG (2026.acl-long)

Copied to clipboard

Chuang Zhou, Zheng Yuan, Linhao Luo, Zhaozhuo Xu, Yilin Xiao, Junnan Dong, Siyu An, di Yin, Xing Sun, Xiao Huang

WebSynthesis: World Model-Guided Monte Carlo Tree Search for Efficient WebAgent Trajectory Synthesis (2026.acl-long)

Challenge:	Retrieval-augmented generation (RAG) has been used for enhancing large language models with external knowledge.
Approach:	They propose a framework for mining efficient graph structures via hashing to enhance RAG . they adopt an inductive paradigm where global graph structure emerges from local hash collisions .
Outcome:	The proposed framework outperforms existing baselines while requiring no GPU resources or token budget.

Copied to clipboard

Yifei Gao, Junhong Ye, Yifan Yang, Jiaqi Wang, Yi Zhang, Zhang Ruichen, Jitao Sang

EyeMulator: Improving Code Language Models by Mimicking Human Visual Attention (2026.acl-long)

Challenge:	Recent advances in large language models have enabled increasingly capable web agents . however, training such agents at scale still relies on high-quality interaction trajectories that are difficult to obtain at scale.
Approach:	They propose a framework for scalable trajectory synthesis that simulates state transitions without network dependencies and integrates Monte Carlo Tree Search to enable reversible exploration over the simulated state space.
Outcome:	Experiments on WebArena, WebVoyager, and Mind2Web-Online show that agents trained exclusively on synthesized trajectories outperform those trained on real-world data.

Copied to clipboard

Yifan Zhang, Chen Huang, Yueke Zhang, Jiahao Zhang, Toby Jia-Jun Li, Collin McMillan, Kevin Leach, Yu Huang

TPA: Next Token Probability Attribution for Detecting Hallucinations in RAG (2026.acl-long)

Challenge:	Code Language Models learn attention based on statistical input-output token correlations.
Approach:	They propose a model-agnostic technique to align CodeLLM attention with human visual attention without architectural changes.
Outcome:	The proposed model outperforms baselines in three languages, with gains of over 30 CodeBLEU points in translation and up to 22 BERTScore points in summarization.

Copied to clipboard

Pengqian Lu, Jie Lu, Anjin Liu, Guangquan Zhang

GMFL: Efficient Global Masking for Federated LLM Fine-tuning (2026.acl-long)

Challenge:	Existing approaches attribute hallucinations to a binary conflict between internal knowledge stored in FFNs and the retrieved context.
Approach:	They propose a framework which mathematically attributes each next-token probability to seven distinct sources and aggregates source attributions by POS tags to quantify contribution of each model component to the generation of specific linguistic categories within a response.
Outcome:	Extensive experiments show that the proposed framework achieves state-of-the-art performance.

Copied to clipboard

Xin Huang, Yan Hu, Yue-Jiao Gong, Xinglin Zhang

AgentGL: Towards Agentic Graph Learning with LLMs via Reinforcement Learning (2026.acl-long)

Challenge:	Low-Rank Adaptation (LoRA) has emerged as a prominent solution to mitigate the communication and computation costs in federated fine-tuning of Large Language Models (LLMs).
Approach:	They propose a plug-and-play layer freezing mechanism to integrate with existing federated fine-tuning frameworks.
Outcome:	The proposed solution reduces communication overhead and lowers computational costs while preserving the performance of the underlying federated fine-tuning methods.

Copied to clipboard

Yuanfu Sun, Kang Li, Dongzhe Fan, Jiajin Liu, Qiaoyu Tan

MM-JudgeBias: A Benchmark for Evaluating Compositional Biases in MLLM-as-a-Judge (2026.acl-long)

Challenge:	Existing agentic frameworks treat external information as unstructured text and fail to leverage topological dependencies inherent in real-world data.
Approach:	They propose to reframe graph learning as an interleaved process of topology-aware navigation and LLM-based inference.
Outcome:	The proposed framework outperforms strong GraphLLMs and GraphRAG benchmarks in multiple LLM backbones.

Copied to clipboard

Sua Lee, Sanghee Park, Jinbae Im

CSPO: Alleviating Reward Ambiguity for Structured Table-to-LaTeX Generation (2026.acl-long)

Challenge:	Multimodal Large Language Models (MLLMs) are increasingly used as automatic judges . however, their reliability and vulnerabilities to biases remain underexplored .
Approach:	They propose a benchmark to evaluate MLLMs that fail to integrate visual cues . they also introduce a test to evaluate the reliability of MLMLs based on a set of asymmetric evaluation tendencies.
Outcome:	Experiments on 26 state-of-the-art MLLMs reveal modality neglect and asymmetric evaluation tendencies . a standardized model with a benchmark enables a fine-grained diagnosis of nine bias types .

Copied to clipboard

Yunfan Yang, Cuiling Lan, Jitao Sang, Yan Lu

Chart-MRAG: Benchmarking Multimodal Retrieval Augmented Generation on Chart-based Documents (2026.acl-long)

Challenge:	Tables contain rich structured information, but when stored as images their contents remain "locked" within pixels.
Approach:	They propose a framework that disentangles optimization across LaTeX tables components . CSPO assigns component-specific rewards and backpropagates each signal through tokens .
Outcome:	The proposed framework disentangles optimization across LaTeX tables components—structure, style, and content.

Copied to clipboard

null Ymyang, Jiang Zhong, Li Jin, Xiao Sun, Jingwang Huang, null Gaojinpeng, Qing Liu, Yang Bai, Jingyuan Zhang, Rui Jiang, Qin Lei, Kaiwen Wei

Towards Stable and Effective Reinforcement Learning for Mixture-of-Experts (2026.acl-long)

Challenge:	Existing benchmarks focus on simple image-text interactions, overlooking complex visual formats like charts.
Approach:	They propose a semi-automatic framework for generating evaluation samples through multi-modal keypoint extraction, knowledge graph construction, and qa pair synthesis.
Outcome:	The proposed framework generates 4,738 question-answering pairs across 8 domains from real-world documents.

Copied to clipboard

Di Zhang, Xun Wu, Shaohan Huang, Lingjie Jiang, Yaru Hao, Li Dong, Zewen Chi, Zhifang Sui, Furu Wei

AIRCoder: Adaptive Integration of Multi-dimensional Retrieval for Repository-level Code Completion (2026.acl-long)

Challenge:	Reinforcement learning with verifiable rewards (RLVR) training with Mixture-of-Experts policies remains fragile and prone to reward collapse.
Approach:	They propose a router shift-based policy optimization method that computes a per-token router-shift ratio conditioned on the previously activated experts and applies stop-gradient and a lower-bound floor.
Outcome:	The proposed method achieves better performance and greater stability than previous methods.

Copied to clipboard

Chuanqi Shi, Miao Gao, Zhiqiang Gao

ReLook: Vision-Grounded RL with a Multimodal LLM Critic for Agentic Web Coding (2026.acl-long)

Challenge:	Existing methods for retrieving code from large codebases use textual similarity or dependency existence, resulting in inconsistent performance.
Approach:	They propose a retrieval framework that integrates eight complementary metrics across three dimensions: textual similarity, dependency existence, and structural hierarchy.
Outcome:	Experiments on CrossCodeEval and RepoEval show that AIRCoder improves accuracy and performance by 10.2 over baseline methods.

Copied to clipboard

Yuhang Li, Chenchen Zhang, Ruilin Lv, Ao Liu, Ken Deng, Yuanxing Zhang, Jiaheng Liu, Bo Zhou

SafetyMem: Adaptive Jailbreak Defense via Dual-Component Safety Memory (2026.acl-long)

Challenge:	Large Language Models (LLMs) excel at algorithmic code generation, but front-end development is lacking in visual fidelity and interaction.
Approach:	They propose an agentic, vision-grounded reinforcement learning framework that closes a loop by invoking a multimodal LLM as a tool.
Outcome:	The proposed framework outperforms baselines in front-end code generation.

Copied to clipboard

Hao Wang, Ziyi Ni, Huacan Wang, Pin Lyu, Lei Sha

Enhancing Agentic Textual Graph Retrieval with Synthetic Stepwise Supervision (2026.acl-long)

Challenge:	Existing defenses for Large Language Models suffer from a 'memory gap' parameter-modifying methods are computationally expensive and inference-time filters cannot retain or reuse defense knowledge across interactions.
Approach:	They propose a framework that secures Large Language Models through a dual-component safety memory system.
Outcome:	The proposed framework significantly reduces attack success rates while preserving interpretability and efficiency.

Copied to clipboard

Ge Chang, Jinbo Su, Jiacheng Liu, Pengfei Yang, Yuhao Shang, Huiwen Zheng, Hongli Ma, Yan Liang, Yuanchun Li, Yunxin Liu

Cognitive Scaffold: From Fluid Context to Crystallized Memory for Long-Horizon DeepResearch Agents (2026.acl-long)

Challenge:	Existing methods for integrating textual graphs with LLMs are limited by symbolic inference and high annotation costs.
Approach:	They propose a textual graph reasoning framework that integrates textual diagrams with large language models.
Outcome:	The proposed approach achieves 15.6% accuracy and 17.2% in F1 score on three common datasets.

Copied to clipboard

Qiuyuan Ai, Zenghuang Fu, Zhaoyang Li, Ping Jiang, Haoyu Wu, Jie Song, Guannan He

Revisiting Metric Reliability for Fine-grained Evaluation of Machine Translation and Summarization in Indian Languages (2026.acl-long)

Challenge:	Scaling LLM-based agents to long-horizon deep research is constrained by context-noise trade-off . solving a single query may require hundreds of interactions with noisy environments .
Approach:	They propose a factorized memory architecture that decouples the cognitive state into a Fluid Working Context for immediate reasoning and a persistent Knowledge Graph for long-term retention.
Outcome:	The Cognitive Scaffold outperforms baselines on Xbench-DeepSearch, BrowseComp-ZH, and GAIA . it achieves 74.7% Avg@3 and 87.0% Pass@3 on xbench, browseComp, and 88.3% Pass@3.

Copied to clipboard

Amir Hossein Yari, Kalmit Kulkarni, Ahmad Raza Khan, Fajri Koto

Learning to Think on Hypergraph: HyperCoT for Structure-Guided N-ary Knowledge Graph Completion (2026.acl-long)

Challenge:	Existing metrics have been developed and validated for English and other languages . this narrow focus leaves Indian languages largely overlooked, casting doubt on universality of current evaluation practices.
Approach:	They propose a large-scale benchmark that compares 26 automatic metrics with human judgments across six major Indian languages.
Outcome:	ITEM evaluates alignment of 26 automatic metrics with human judgments across six languages . authors: outliers exert significant impact on metric-human agreement, improve fidelity . they say the results offer critical guidance for advancing metric design and evaluation in Indian languages - a global market for machine translation and text summarization systems.

Copied to clipboard

Mengxue Yang, Jinming Li, Chun Yang, Jiaqi Zhu, Jiafan Li, Guanhua Zhang, Ying Li

Two Pathways to Truthfulness: On the Intrinsic Encoding of LLM Hallucinations (2026.acl-long)

Challenge:	Existing methods to solve knowledge hypergraph link prediction problem are limited by their ability to generate chain-of-thought (CoT) representations.
Approach:	They propose a structure-aware approach that models multi-hop structural reasoning as a depth-sensitive progressive evidence accumulation process.
Outcome:	Experiments on three real-world datasets show that HyperCoT outperforms strong n-ary KGC baselines while yielding interpretable multi-hop reasoning traces.

Copied to clipboard

Wen Luo, Guangyue Peng, Wei Li, Shaohang Wei, Feifan Song, Liang Wang, Nan Yang, Xingxing Zhang, Jing Jin, Furu Wei, Houfeng Wang

LLM-ForcedAligner: A Non-Autoregressive and Accurate LLM-Based Forced Aligner for Multilingual and Long-Form Speech (2026.acl-long)

Challenge:	Previous work shows that large language models generate hallucinations, yet the origins and mechanisms of these signals remain unclear.
Approach:	They propose to validate and disentangle two different pathways for truthfulness cues . they also propose to use the same mechanism to derive self-contained evidence from the generated answer .
Outcome:	The proposed applications improve hallucination detection performance by integrating two different inputs.

Copied to clipboard

Bingshen Mu, Xian Shi, Xiong Wang, Hexin Liu, Jin Xu, Lei Xie

Forget What Matters, Keep the Rest: Selective Unlearning of Informative Tokens (2026.acl-long)

Challenge:	Existing methods for forcing alignment are language-specific and prone to temporal shifts.
Approach:	They propose a slot-filling paradigm that uses time indices to predict slot positions.
Outcome:	The proposed method reduces accumulated temporal shifts by 69% compared with prior methods.

Copied to clipboard

Seunghee Koh, Sunghyun Baek, Youngdong Kim, Junmo Kim

Challenge:	Recent studies have explored token-wise loss regularizers that prioritize informative tokens, but rely on ground-truth confidence or external linguistic parsers, which limits their ability to capture contextual information or the model’s overall predictive state.
Approach:	They propose an Entropy-guided Token Weighting (ETW) token-level unlearning regularizer that uses entropy of the predictive distribution as a proxy for token informativeness.
Outcome:	The proposed token-level unlearning regularizer can achieve more effective unlearning while better preserving model utility than existing token-based approaches.

Latent-Condensed Transformer for Efficient Long Context Modeling (2026.acl-long)

Copied to clipboard

Zeng You, Yaofo Chen, Qiuwu Chen, Ying Sun, Shuhai Zhang, Yingjian Li, Yaowei Wang, Mingkui Tan

MedForge: Interpretable Medical Deepfake Detection via Forgery-aware Reasoning (2026.acl-long)

Challenge:	Existing approaches address these bottlenecks separately: Multi-head Latent Attention (MLA) reduces the KV cache by projecting tokens into a low-dimensional latent space, while sparse attention reduces computation.
Approach:	They propose a Latent-Condensed Attention mechanism that performs structured context condensation directly within MLA's latent space.
Outcome:	The proposed approach reduces KV cache size and attention cost without adding parameters.

Copied to clipboard

Zhihui Chen, Kai He, Qingyuan Lei, Bin Pu, Jian Zhang, Yuling Xu, Mengling Feng

CrossGuard: Safeguarding MLLMs against Joint-Modal Implicit Malicious Attacks (2026.acl-long)

Challenge:	Existing defenses against forgery are inadequate for healthcare.
Approach:	They propose a large-scale benchmark for pre-hoc, evidence-grounded medical forgery detection using a doctor inspection guideline and gold edit locations.
Outcome:	Experiments show that the proposed solution can detect and explain medical scans with high fidelity and accuracy.

Copied to clipboard

Xu Zhang, Hao Li, Zhichao Lu

Can Spectral-Clipping Enable Better Learning While Forgetting Less for Low-Rank Adaptation? (2026.acl-long)

Challenge:	Existing methods for MLLMs are weak on explicit attacks, but weak on implicit ones.
Approach:	They propose an automated red-teaming pipeline that leverages reinforcement learning with tailored reward modules to generate diverse implicit samples across 14 domains.
Outcome:	The proposed method outperforms existing methods in implicit and explicit attacks while maintaining high utility.

Copied to clipboard

Hyowon Wi, Noseong Park

Re3: Relevance & Recency Retrieval for Mitigating Temporal Hallucination (2026.acl-long)

Challenge:	In recent years, low-rank adaptation (LoRA) has emerged as a significant paradigm that freezes pre-trained weights and introduces small, learnable adapters instead of fine-tuning the full set of parameters.
Approach:	They propose a low-rank adaptation approach that injects parameterized singular components with spectral clipping into the pre-trained model.
Outcome:	The proposed method improves performance and retains pre-trained knowledge while preserving the weights of the model.

Copied to clipboard

Jiawei Cao, Jie Ouyang, Mingyue Cheng, Zhaomeng Zhou, Chunli Liu, Yupeng Li, Zirui Liu, Shijin Wang

TeCES: Collaborative Geometric Knowledge Representation Framework under Evolving Fact Snapshots (2026.acl-long)

Challenge:	Existing retrievers suffer from temporal-semantic misalignment and outdated-document interference . Existing frameworks suffer from both temporal validity and outdated factual versions .
Approach:	They propose a framework that mitigates temporal hallucinations by embedding heterogeneous temporal signals into the semantic space to ensure retrieval fidelity.
Outcome:	Experiments show that Re3 outperforms baselines by 9.7% in generation accuracy . the framework outperformed strongest baselines on challenging dynamic tasks .

Copied to clipboard

Jiujiang Guo, Zhengliang Guo, Kai Wang, Meiyang Wang, Dehua Peng, Shaozu Yuan, Chengyin Hu, Shuan Ai, Yiwei Wei

MicroC-KT: Modeling Community Effect via Learning Micro-Environment for Evidence-Grounded Explainable Knowledge Tracing (2026.acl-long)

Challenge:	Existing knowledge graphs represent static facts but lack collaborative modeling of both . e.g., existing knowledge graph models lack a framework for integrating snapshots into knowledge graph.
Approach:	They propose a framework for high-fidelity modeling of evolving snapshots using concept of snapshots.
Outcome:	The proposed framework outperforms existing models on six benchmarks.

Copied to clipboard

Zhiyi Duan, Zixing Shi, Bing Jia, Qi Wang

ImF: Embedding an Implicit Fingerprint in Your Large Language Models (2026.acl-long)

Challenge:	Existing graph-based methods focus on exercise-concept relations, but lack the broader context of group references and contrastive evidence.
Approach:	They propose a framework that incorporates learning micro-environments to provide social-cognitive anchors for KT by extracting contrastive group evidence.
Outcome:	The proposed framework outperforms state-of-the-art models on four public datasets while providing more reliable and evidence-based explanations.

Copied to clipboard

Jiaxuan Wu, Wanli Peng, Hang fu, Xue Yiming, Juan Wen

Temp-R1: A Unified Autonomous Agent for Complex Temporal KGQA via Reverse Curriculum Reinforcement Learning (2026.acl-long)

Challenge:	Training and serving large language models (LLMs) is resource-intensive, making reliable intellectual property protection and black-box ownership verification increasingly important.
Approach:	They propose a method to inject a small set of secret query–response behaviors into model fingerprinting . they encode ownership information into a natural-looking target response and derive a semantically aligned query .
Outcome:	The proposed fingerprints improve stealthiness and remain verifiable under model updates and deployment-time prompt interventions.

Copied to clipboard

Zhaoyan Gong, Zhiqiang Liu, Songze Li, Xiaoke Guo, Yuanxiang Liu, Xinle Deng, Zhizhen Liu, Lei Liang, Huajun Chen, Wen Zhang

S2G-RAG: Structured Sufficiency and Gap Judging for Iterative Retrieval-Augmented QA (2026.acl-long)

Challenge:	Existing methods rely on fixed workflows and expensive closed-source APIs, limiting flexibility and scalability.
Approach:	They propose a temporal reasoning agent that trains on difficult questions first . they expand the action space with specialized internal actions alongside external action .
Outcome:	The proposed agent improves 19.8% over baselines on complex questions and multi-tasks.

Copied to clipboard

Minghan Li, Junjie Zou, Xinxuan Lv, Chao Zhang, Guodong Zhou

Question Answering

Challenge:	Retrieval-augmented generation grounds language models in external evidence, but multi-hop question answering remains difficult . iterative pipelines must control what to retrieve next and when evidence is adequate.
Approach:	They propose an iterative framework with an explicit controller, S2G-Judge . they map structured gap items into the next retrieval query to produce stable retrieval trajectories .
Outcome:	Experiments on TriviaQA, HotpotQA, and 2WikiMultiHopQA show that S2G-RAG improves multi-hop QA performance and robustness under multi-turn retrieval.

Jailbreaking Multimodal Large Language Models using Multi-Clip Video (2026.acl-long)

Copied to clipboard

Choongwon Kang, Seungjong Sun, Hyunmin Jun, Jang Hyun Kim

HAG: Hierarchical Demographic Tree-based Agent Generation for Topic-Adaptive Simulation (2026.acl-long)

Challenge:	Existing studies show that video inputs can bypass safety alignment, yet it remains unclear which properties of video input induce this vulnerability.
Approach:	They propose a simple image-based defense that mitigates the vulnerability of MLLMs by analyzing video inputs.
Outcome:	The proposed defense leverages the relative robustness of the image modality.

Copied to clipboard

Rongxin Chen, Tianyu Wu, Bingbing Xu, JiaTang Luo, Xiucheng Xu, Huawei Shen

Challenge:	Existing approaches to Agent-Based Modeling fail to adapt to unseen topics absent from data.
Approach:	They propose a Hierarchical Agent Generation framework that formalizes population generation as a two-stage decision process.
Outcome:	The proposed framework outperforms baseline models in a multi-domain benchmark and comprehensive evaluation framework.

Stable-RAG: Mitigating Retrieval-Permutation-Induced Hallucinations in Retrieval-Augmented Generation (2026.acl-long)

Copied to clipboard

Qianchi Zhang, Hainan Zhang, Liang Pang, Hong-Wei Zheng, Zhiming Zheng

From Logical to Computational Sparsity: Structure-Aware Block-Sparse Attention for Long-Code Completion (2026.acl-long)

Challenge:	Existing RAG methods focus on enhancing LLM robustness to low-quality retrieval, but neither address permutation sensitivity.
Approach:	They propose a method that exploits permutation sensitivity to mitigate hallucinations in Large Language Models.
Outcome:	The proposed model improves answer accuracy, reasoning consistency, and generalization across datasets, retrievers, and input lengths compared with strong baselines.

Copied to clipboard

Yanli Wang, Yanlin Wang, Bowen Zhang, Yiwei Zhang, Daya Guo, Jiachi Chen, Hongyu Zhang, Zibin Zheng

CogEvolve: A Multimodal Benchmark for Evaluating Relational Reasoning in Semantic Extension (2026.acl-long)

Challenge:	Existing sparse attention methods for long-context generation pose high latency . general sparsity methods cause excessive accuracy degradation without considering code structure .
Approach:	They propose a training-free Structure-aware block-sparse attention mechanism that bridges the gap between logical and computational sparsity.
Outcome:	The proposed method reduces TTFT by 45-55% while maintaining accuracy within 3% of dense attention.

Copied to clipboard

Jingjie Zeng, Huayang Li, Liang Yang, Yuanyuan Sun, Shaowu Zhang, Hongfei Lin

Towards Privacy-Preserving Large Language Model: Text-free Inference Through Alignment and Adaptation (2026.acl-long)

Challenge:	a gap exists between human embodied logic and machine statistical learning . authors: models internalize statistical patterns or mimic static recognition .
Approach:	They propose a cognitive linguistic benchmark to test whether large language models internalize statistical logic or not . they find that models function as "Super-Associators" expert at static recognition yet fail at causal reasoning .
Outcome:	The proposed model fails at causal reasoning and has a high-fidelity concept representation but lacks transformational operators essential for true relational understanding.

Copied to clipboard

Jeongho Yoon, Chanhee Park, Yongchan Chun, Hyeonseok Moon, Heuiseok Lim

VisRet: Visualization Improves Knowledge-Intensive Text-to-Image Retrieval (2026.acl-long)

Challenge:	Existing LLMs require users to submit raw text regardless of its sensitivity, resulting in substantial computational overhead and degrade model performance.
Approach:	They propose a new training pipeline that allows a client-side encoder to condition on k-pooled prompt embeddings instead of raw text and a server-side projection module to fine-tune the projection module and LLM on private, domain-specific data using noise-injected embeddables.
Outcome:	The proposed approach eliminates the need for transmitting raw prompt text while maintaining a favorable balance between privacy preservation and model utility for both clients and service providers.

Copied to clipboard

Di Wu, Yixin Wan, Kai-Wei Chang

Selective Knowledge Distillation: Fusing LLM Semantic Strengths with DNN Efficiency for Binary Code Similarity Detection (2026.acl-long)

Challenge:	Text-to-image retrieval is challenging because of cross-modal embeddings are bags of concepts, underrepresenting structured visual relationships.
Approach:	They propose a retrieval paradigm that embeds textual queries into the image modality via T2I generation and performs retrieval within the image mode to bypass weaknesses of cross-modal retrievers in recognizing subtle visual-spatial features.
Outcome:	The proposed retrieval paradigm outperforms previous approaches in visual-spatial retrieval benchmarks.

Copied to clipboard

Shize Zhou, Peiyu Liu, Lirong Fu, Tong Ye, Wenhai Wang

Mixture-of-Experts with Intermediate CTC Supervision for Accented Speech Recognition (2026.acl-long)

Challenge:	BinSKD is a binary code similarity detection technique that can be used in bug detection, patch analysis, and malware detection.
Approach:	They propose to leverage an LLM-based BCSD method as the teacher model and transfer its knowledge of high-level program semantics to various DNN-based student models.
Outcome:	The proposed method yields Recall@1 improvements of 14.5%–91.2% for DNN-based BCSD methods and enables HermesSim to match the teacher’s performance with orders-of-magnitude efficiency.

Copied to clipboard

Wonjun Lee, Hyounghun Kim, Gary Lee

TransLLM: A Unified Multi-Task Large Language Model for Urban Transportation via Learnable Prompting (2026.acl-long)

Challenge:	Accented speech remains a persistent challenge for automatic speech recognition (ASR) Accent-agnostic approaches improve robustness but struggle with heavily accented or unseen varieties .
Approach:	They propose a Mixture-of-Experts architecture with intermediate CTC supervision that promotes expert specialization and generalization.
Outcome:	Experiments show that the proposed architecture improves on accented speech . the proposed framework is based on a mixture-of-experts architecture with intermediate supervision .

Copied to clipboard

Jiaming Leng, Yunying Bi, Chuan Qin, Zhenya Huang, Bing Yin, Haojie Ren, Yanyong Zhang, Chao Wang

CoreCodeBench: Decoupling Code Intelligence via Fine-Grained Repository-Level Tasks (2026.acl-long)

Challenge:	Existing models lack generalization capabilities and lack structured spatiotemporal data.
Approach:	They propose a unified multi-task framework that synergizes spatiotemporal encoding with LLM reasoning through learnable prompt composition.
Outcome:	The proposed framework outperforms baseline models on seven datasets and three tasks on supervised and zero-shot settings with excellent generalization and robustness.

Copied to clipboard

Lingyue Fu, Hao Guan, Bolun Zhang, Haowei Yuan, Yaoming Zhu, Lin Qiu, ZongYu Wang, Xuezhi Cao, Xunliang Cai, Weiwen Liu, Weinan Zhang, Yong Yu

Challenge:	Existing large language models for software engineering rely on coarse-grained pass rates obscuring specific cognitive bottlenecks.
Approach:	They propose a repository-level benchmark that dissects coding capabilities through atomized tasks.
Outcome:	The proposed framework achieves a 78.55% validity yield, surpassing the 31.7% retention rate of SWE-bench-Verified.

Stop Fixating on Prompts: Reasoning Hijacking and Constraint Tightening for Red-Teaming LLM Agents (2026.acl-long)

Copied to clipboard

Yanxu Mao, Peipei Liu, Tiehan Cui, Congying Liu, Mingzhe Xing, Datao You

Attention Basin: Why Contextual Position Matters in Large Language Models (2026.acl-long)

Challenge:	Existing red-team methods rely on modifying user prompts, which lack adaptability to new data and may impact the agent’s performance.
Approach:	They propose a framework that implicitly manipulates the agent’s reasoning trajectory and memory retrieval with three key stages: Trigger Extraction, Reasoning Hijacking, and Constraint Tightening.
Outcome:	The proposed framework shows outstanding performance in cross-model and cross-scenario environments.

Copied to clipboard

Zihao Yi, Zhenqing Ling, Delong Zeng, Haohao Luo, Zhe Xu, Wei Liu, Jian Luan, Wanxia Cao, Ying Shen

RRAtention: Dynamic Block Sparse Attention via Per-Head Round-Robin Shifts for Long-Context Inference (2026.acl-long)

Challenge:	Large Language Models (LLMs) are sensitive to the contextual position of information in input.
Approach:	They introduce Attention-Driven Reranking (AttnRank) which estimates a model’s intrinsic positional attention preferences using a small calibration set and reorders retrieved documents or few-shot examples to align the most salient content with these high-attention positions.
Outcome:	Experiments on multi-hop QA and few-shot in-context learning tasks show that AttnRank achieves substantial improvements across 10 large language models of varying architectures and scales, without modifying model parameters or training procedures.

Copied to clipboard

Siran Liu, Guoxia Wang, Sa Wang, Jinle Zeng, Haoyang Xie, Siyu Lou, Jiabin Yang, Dianhai Yu, Haifeng Wang, Chao Yang

Beyond Surface Features: Advancing Medical Vision-Language Alignment via Dynamic Evidence-Guided Preference Optimization (2026.acl-long)

Challenge:	Existing approaches to dynamic sparse attention require preprocessing, lack global evaluation, violate query independence, or incur high computational overhead.
Approach:	They propose a dynamic sparse attention method that achieves all desirable properties through a head round-robin (RR) sampling strategy.
Outcome:	Experiments on natural language understanding and multimodal video comprehension show that the proposed method achieves 2.4 speedup at 128K context length outperforming existing methods.

Copied to clipboard

Zixuan Huang, Zhihong Zhu, Xiaolong Liu, Yanchao Hao, Manman Zhang, Zheng Wei, Bowen Xing, Xian Wu, Ye Li, Fen Miao, Yefeng Zheng

SAFE-QAQ: End-to-End Slow-Thinking Audio-Text Fraud Detection via Reinforcement Learning (2026.acl-long)

Challenge:	Existing preference-based methods for medical large vision-Language Models face limitations in medical settings . existing methods are limited by overfitting to superficial cues and pseudo convergence of the preference signal.
Approach:	They propose a framework that enables evidence-aware and adaptive preference learning for Med-LVLMs.
Outcome:	The proposed framework improves evidence-aware and adaptive preference learning for Med-LVLMs.

Copied to clipboard

Peidong Wang, Zhiming Ma, Xin Dai, YongKang Liu, Shi Feng, Xiaocui Yang, Wenxing Hu, Zhihao Wang, Mingjun Pan, Li Yuan, Daling Wang

MaDS: Long-Horizon GUI Automation via Synergizing Dual-Layer Memory and Multi-Round Debate (2026.acl-long)

Challenge:	Existing methods for fraud detection rely on transcribed text, lacking acoustic cues . a proposed framework for audio-based slow-thinking fraud detection eliminates transcription errors .
Approach:	They propose a framework for audio-based slow-thinking fraud detection that eliminates transcription errors and rewards slow-thought reasoning by capturing fine-grained audio details.
Outcome:	The proposed method improves accuracy, inference efficiency, and real-time processing capabilities.

Copied to clipboard

Pengchen Chen, Shi Chen, Qiming Ye, Xinli Chen, Xinran Li, Wei Xiang

Challenge:	Current methods struggle to distinguish targets in low Signal-to-Noise Ratio environments and lack sufficient pre-execution verification to prevent error accumulation.
Approach:	They propose a Memory-augmented Debate System to ensure precise grounding across diverse interfaces and handle irreversible errors in extended workflows.
Outcome:	The proposed system achieves a 90.23% task success rate on MaDS-Benchmark and strong performance on public benchmarks including AITW, AITZ, CAGUI, and GUIOdyssey.

CIRAG: Construction–Integration Retrieval and Adaptive Generation for Multi-hop Question Answering (2026.acl-long)

Copied to clipboard

Zili Wei, Yilin Wang, Xiaocui Yang, Shi Feng, Weidong Bao, Daling Wang, Zihan Wang, Yifei Zhang

TinyJudge: Unverifiable Constraint Alignment via Lightweight Specialist Ensembles (2026.acl-long)

Challenge:	Existing methods for iterative retrieval-augmented generation (iRAG) suffer from greedy single-path expansion and granularity–demand mismatch .
Approach:	They propose a model that constructs candidate triples and history-conditionally integrates them to distill core triples to generate the next-hop query.
Outcome:	The proposed model mitigates the greedy single-path expansion and granularity–demand mismatch by preserving multiple plausible evidence chains.

Copied to clipboard

Yirong Zeng, Yufei Liu, Xiao Ding, Yutai Hou, Yuxian Wang, Wu Ning, Haonan Song, Dandan Tu, Qixun Zhang, Yuxiang He, Bibo Cai, Ting Liu

Challenge:	Instruction Following (IF) is a core capability of LLMs, requiring strict adherence to diverse constraints.
Approach:	They propose a framework that uses tiny language models to evaluate instruction following . they propose to use a set of specialized tiny language model to provide rewards for soft constraints.
Outcome:	The proposed framework outperforms baseline models by 12% and speeds up training time by 3.

Dynamic Infilling Anchors for Format-Constrained Generation in Diffusion Large Language Models (2026.acl-long)

Copied to clipboard

Boyan Han, Yiwei Wang, Yi Song, Yujun Cai, Chi Zhang

Compiling Activation Steering into Weights via Null-Space Constraints for Stealthy Backdoors (2026.acl-long)

Challenge:	Diffusion large language models (dLLMs) offer bidirectional attention and parallel generation . fixed anchors can enforce constraints, but they often impose rigid spans, leading to truncated reasoning .
Approach:	They propose a method that dynamically estimates end-anchor positions to adjust generation length before iterative infilling.
Outcome:	The proposed method improves format compliance and answer accuracy on GSM8K and MATH.

Copied to clipboard

Rui Yin, Tianxu Han, Naen Xu, Changjiang Li, Ping He, Chunyi Zhou, Jun Wang, Zhihui Fu, Tianyu Du, Jinbao Li, Shouling Ji

Inhibitory Attacks on Backdoor-based Fingerprinting for Large Language Models (2026.acl-long)

Challenge:	Existing methods to inject safety-aligned large language models rely on token-level mappings, which do not guarantee sustained harmful output.
Approach:	They propose a method that directly modifies model weights to map a trigger to an attacker-specified response.
Outcome:	The proposed method achieves high triggered attack success while maintaining non-triggered safety and general utility.

Copied to clipboard

Hang fu, Wanli Peng, Yinghan Zhou, Jiaxuan Wu, Juan Wen, Xue Yiming

TimeSAF: Towards LLM-Guided Semantic Asynchronous Fusion for Time Series Forecasting (2026.acl-long)

Challenge:	Backdoor-based LLM fingerprinting is a promising solution for intellectual property protection . however, the vulnerability of existing LLMs for the ensemble scenario is unexplored .
Approach:	They propose two new fingerprinting attack methods to assess the robustness of LLM fingerprinting by token filter attack and sentence verification attack.
Outcome:	The proposed methods inhibit the fingerprint response while maintaining ensemble performance.

Copied to clipboard

Fan Zhang, Shiming Fan, Hua Wang

Pruning Unsafe Tickets: A Resource-Efficient Framework for Safer and More Robust LLMs (2026.acl-long)

Challenge:	Existing time series forecasting methods use a deep synchronous fusion strategy . high-level abstract semantics are inappropriately entangled with low-level temporal dynamics .
Approach:	They propose a framework based on hierarchical asynchronous fusion that decouples unimodal feature learning from cross-modal interaction.
Outcome:	The proposed framework outperforms state-of-the-art approaches on long-term forecasting benchmarks.

Copied to clipboard

Wai Man Si, Mingjie Li, Michael Backes, Yang Zhang

EQUIP: EQUivariant preserving In-Place updates for Efficient Token Pruning (2026.acl-long)

Challenge:	Empirical evaluations on ML models show substantial reductions in unsafe generations and improved robustness against jailbreak attacks.
Approach:	They propose a resource-efficient pruning framework that directly identifies unsafe behaviors while preserving model utility.
Outcome:	The proposed framework reduces unsafe generations and improves robustness against jailbreak attacks with minimal utility loss.

Copied to clipboard

Arun Ramachandran, R Govindarajan, Murali Annavaram, Prakash Raghavendra

Exons-Detect: Identifying and Amplifying Exonic Tokens via Hidden-State Discrepancy for Robust AI-Generated Text Detection (2026.acl-long)

Challenge:	Token-pruning methods cause "holes" in KV tensors, posing major challenges . equip reduces recomputation of rotation operations through in-place update, caching and re-indexing .
Approach:	They propose an EQUIP-based in-place token update mechanism that preserves the equivariance property of the operations performed in the attention computation.
Outcome:	EQUIP reduces recomputation of rotation operations and reduces eviction overheads . it achieves geomean speedups of 1.62 (or 1.47) over StreamingLLM and 3.45 ( or 1.86)

Copied to clipboard

Xiaowei Zhu, Yubing Ren, Fang Fang, Shi Wang, Yanan Cao, Li Guo

Reasoning Fails Where Step Flow Breaks (2026.acl-long)

Challenge:	Existing methods for AI-generated text detection assume uniform token contributions, making them less robust under short sequences or localized token modifications.
Approach:	They propose a training-free method for AI-generated text detection based on an exon-aware token reweighting perspective.
Outcome:	The proposed method achieves state-of-the-art detection performance and robustness to adversarial attacks and varying input lengths.

Copied to clipboard

Xiaoyu Xu, Yulan Pan, Xiaosong Yuan, Zhihong Shen, Minghao Su, Yuanhao Su, Xiaofeng Zhang

Cat-MoD: Accelerating Multimodal Alignment via Caption Token Guided Asymmetric Mixture-of-Depths (2026.acl-long)

Challenge:	Existing analysis tools struggle with long chain of thought traces.
Approach:	They propose a saliency-inspired test-time intervention that adjusts shallow saliencies to improve accuracy on math, science, and coding tasks.
Outcome:	The proposed model improves accuracy on math, science, and coding tasks without retraining.

Copied to clipboard

YiJie Huang, Xiaocui Yang, Shi Feng, Wen Zhang, Kaisong Song, Yifei Zhang, Daling Wang

Beyond Scaling: Measuring and Predicting the Upper Bound of Knowledge Retention in Language Model Pre-Training (2026.acl-long)

Challenge:	Existing query-based alignment modules enforce uniform cross-attention across all layers, leading to computational redundancy.
Approach:	They propose a framework that allows for asynchronous query-based alignment with large-scale visual features.
Outcome:	The proposed framework matches or surpasses baseline performance while reducing alignment FLOPs by approximately 37% during training and inference.

Copied to clipboard

Changhao Jiang, Ming Zhang, Yifei Cao, Junjie Ye, Xiaoran Fan, Shihan Dou, Zhiheng Xi, Jiajun Sun, Yi Dong, Yujiong Shen, Jingqi Tong, Baoyu Fan, Tao Gui, Qi Zhang, Xuanjing Huang

Efficient Paths and Dense Rewards: Probabilistic Flow Reasoning for Large Language Models (2026.acl-long)

Challenge:	Existing methods to predict performance of large language models are lacking . authors propose a size-dependent mutual information predictor for closed-book question answering accuracy .
Approach:	They propose a size-dependent mutual information predictor that integrates knowledge frequency, knowledge specificity, and model size to forecast closed-book question answering accuracy.
Outcome:	The proposed method outperforms baseline models and achieves R2 > 0.7 in predicting QA accuracy without additional training.

Copied to clipboard

Yan Liu, Feng Zhang, Zhanyu Ma, Jun Xu, Jiuchong Gao, Jinghua Hao, Renqing He, Han Liu, Yangdong Deng

TriPlay-RL: Tri-Role Self-Play Reinforcement Learning for LLM Safety Alignment (2026.acl-long)

Challenge:	Existing approaches to mitigate inference inefficiency and optimization difficulty are fragmented and constrained by inherent trade-offs.
Approach:	They propose a framework that reconceptualizes discrete reasoning steps as a continuous probabilistic flow, quantifying the contribution of each step toward the ground-truth answer.
Outcome:	The proposed framework achieves a superior balance between inference efficiency and reasoning performance on challenging benchmarks.

Copied to clipboard

Zhewen Tan, Wenhan Yu, Jianfeng Si, Tongxin Liu, Kaiqi Guan, Huiyan Jin, Jiawen Tao, Xiaokun Yuan, Xiangzheng Zhang, Duohe Ma, Tong Yang, Lin Sun

Cross-Modal Coreference Alignment: Enabling Reliable Information Transfer in Omni-LLMs (2026.acl-long)

Challenge:	Existing approaches to safety alignment of large language models rely on costly manual annotations or human review.
Approach:	They propose a closed-loop reinforcement learning framework called TriPlay-RL that enables iterative collaboration among three roles with near-zero manual annotation.
Outcome:	The proposed framework achieves 20%–50% improvement in adversarial effectiveness while preserving high output diversity while achieving 10%–30% gains in safety performance without degrading general reasoning capability.

Copied to clipboard

Hongcheng Liu, Yuhao Wang, Zhe Chen, Pingjie Wang, Zhiyuan Zhu, Yixuan Hou, Yanfeng Wang, Yu Wang

Inflated Excellence or True Performance? Rethinking Medical Diagnostic Benchmarks with Dynamic Evaluation (2026.acl-long)

Challenge:	Experiments on 13 Omni-LLMs reveal systematic weaknesses in cross-modal coreference . cross-module coreference is a crucial missing piece for advancing robust omni-modal reasoning.
Approach:	They propose a cross-modal coreference problem to evaluate and enhance Omni-LLMs' reasoning capabilities.
Outcome:	Experiments on 13 Omni-LLMs show they lack coreference-aware thinking patterns . the CROSSOMNI dataset yields significant performance gains and generalizes well to collaborative reasoning tasks.

Copied to clipboard

Xiangxu Zhang, Lei Li, Yanyun Zhou, Xiao Zhou, Yingying Zhang, Xian Wu

Challenge:	Current evaluations of large language models (LLMs) are limited in capturing key challenges of clinical diagnostic scenarios.
Approach:	They propose a dynamic benchmark for medical diagnostics that provides a stress test of diagnostic robustness.
Outcome:	The proposed model provides a stress test of diagnostic robustness and veracity, helpfulness and consistency.

TrendFact: A Benchmark Towards Hotspot Perception in Automatic Fact-Checking (2026.acl-long)

Copied to clipboard

Xiaocheng Zhang, Xi Wang, Yifei Lu, Jianing Wang, Zhuangzhuang Ye, Mengjiao Bao, Peng Yan, Xiaohong Su

MPBoCo: Multimodal Prompt-based Boundary-enhanced Continual Framework for Joint Entity and Relation Extraction (2026.acl-long)

Challenge:	Existing benchmarks lack social metadata and evaluation framework to meet this urgent evaluation needs.
Approach:	They propose a benchmark capable of evaluating HPA and three fact-checking tasks.
Outcome:	The proposed framework improves HPA and computational efficiency for RLM-driven systems.

Copied to clipboard

Guanglu Sun, Xinyu Liu, Lili Liang, Yang Yu, Fei Lang, Suxia Zhu, Ming Liu

ConfSpec: Efficient Step-Level Speculative Reasoning via Confidence-Gated Verification (2026.acl-long)

Challenge:	Existing methods struggle to balance real-time adaptability and computational efficiency in continual learning scenarios.
Approach:	They propose a Continual Multimodal Entity and Relation Joint Extraction task and a Multimodal Prompt-based Boundary-enhanced Continuum framework that stores task-specific knowledge via learnable multimodal prompts.
Outcome:	The proposed framework outperforms baseline methods in real-world scenarios by 5.5% and 7.2%.

Copied to clipboard

Siran Liu, Zane Cao, Yongchao He

Personalizing LLMs with Binary Feedback: A Preference-Calibrated Optimization Framework (2026.acl-long)

Challenge:	Existing approaches to chain-of-thought reasoning incur high inference latency due to long generation traces.
Approach:	They propose a confidence-gated cascaded verification framework that reduces the trade-off between generation and verification.
Outcome:	The proposed framework achieves 2.24 speedups while matching target-model accuracy.

Copied to clipboard

Xilai Ma, Liye Zhao, Weijun Yao, Haibing Di, Wenya Wang, Jing Li

Challenge:	Existing methods focus on isolated user histories, neglecting the essential role of inter-user differences.
Approach:	They propose a framework that personalizes Large Language Models via preference-calibrated binary signals.
Outcome:	The proposed framework outperforms baselines in a variety of personalization tasks and backbone LLMs.

Localized Low-Rank Adaptation within Clustered Parameter Subspaces (2026.acl-long)

Copied to clipboard

Jiahao Xiong, Yihe Liu, Xianming Hu, Hongbo Zhao, Nuoyi Chen, Jie Zhang, Kai Zhang

SGVEF-LOOP: Coverage-Guided Progressive Topological Exploration and Fact-Grounded Metamorphic Evaluation for MCP Agents (2026.acl-long)

Challenge:	Low-Rank Adaptation (LoRA) for large language models has been successful in various domains.
Approach:	They propose to perform low-rank updates within clustered parameter subspaces . they group rows/columns of update matrix into locally coherent, uncorrelated subspace blocks .
Outcome:	Empirical results show that low-rank Adaptation (LoRA) is better than global adaptations in various domains.

Copied to clipboard

Zenghao Liu, Yansong Zhang

Challenge:	Existing frameworks for modeling agents are inadequate for comprehensive evaluation . evaluation of 8 diverse MCP Agents reveals capability stratification and behavioral anomalies .
Approach:	They propose a coverage-guided framework for progressive topological exploration and fact-augmented metamorphic testing that exploits sparse regions using adaptive sampling and dual-constraint validation.
Outcome:	The proposed framework achieves 100% coverage and 80.54% of the theoretical transition bound.

MentalSeek-Dx: Towards Progressive Hypothetico-Deductive Reasoning for Real-world Psychiatric Diagnosis (2026.acl-long)

Copied to clipboard

Xiao Sun, null Ymyang, Xinyi Jiang, Yu Tian, Junnan Zhu, Jiang Zhong, Qin Lei, Jingwang Huang, Haoyang Zeng, Xinyu Zhou, Xin Xiao, Kaiwen Wei

Think Parallax: Solving Multi-Hop Problems via Multi-View Knowledge-Graph-Based Retrieval-Augmented Generation (2026.acl-long)

Challenge:	Mental health disorders represent a burgeoning global public health challenge . lack of ecological validity and fine-grained diagnostic supervision limits their utility .
Approach:	They propose a medical-specialized LLM trained to internalize clinical reasoning process through supervised trajectory construction and curriculum-based reinforcement learning.
Outcome:	The proposed model achieves state-of-the-art with only 14B parameters, establishing a clinically grounded framework for reliable psychiatric diagnosis.

Copied to clipboard

Jinliang Liu, Jiale Bai, Shaoning Zeng

AIM-CoT: Active Information-driven Multimodal Chain-of-Thought for Vision-Language Reasoning (2026.acl-long)

Challenge:	Existing KG-RAG systems collapse all reasoning hops into a single representation, flat embedding space, suppressing this implicit structure and causing noisy or drifted path exploration.
Approach:	They propose a symmetric multi-view framework that decouples queries and KGs into aligned, head-specific retrieval spaces.
Outcome:	The proposed framework achieves state-of-the-art retrieval and QA performance on WebQSP and CWQ, and significantly reduces hallucination.

Copied to clipboard

Xiping Li, Jianghong Ma

Benchmarking Egocentric Clinical Intent Understanding Capability for Medical Multimodal Large Language Models (2026.acl-long)

Challenge:	Existing methods for I-MCoT fail to capture dynamic needs of vision-language models . existing methods rely on attention signals, which are unreliable under severe granularity imbalance between brief textual query and informative image.
Approach:	They propose a framework that integrates specially selected visual evidence into the context of Vision-Language Models (VLMs) they propose 'AIM-CoT' to improve evidence selection and insertion triggering .
Outcome:	Experiments across three benchmarks and four backbones demonstrate the proposed framework’s consistent superiority.

Copied to clipboard

Shaonan Liu, Guo Yu, Xiaoling Luo, Shiyi Zheng, Jie Liu, Wenting Chen, Linlin Shen

Data Pollination: An Emergent Ecological Process Driving AI Population Evolution (2026.acl-long)

Challenge:	Existing benchmarks fail to evaluate egocentric clinical intent understanding of medical multimodal large language models.
Approach:	They propose a benchmark leveraging clinician gaze as a Cognitive Cursor to assess intent understanding across surgery, emergency simulation and diagnostic interpretation.
Outcome:	The proposed benchmark addresses challenges of visual homogeneity of anatomical structures, strict temporal-causal dependencies in clinical workflows, and implicit adherence to safety protocols.

Copied to clipboard

Shufang Xie, Qizhi Pei, Ang Lv, Jingyang Hu, Lijun Wu, Rui Yan

Extending First-Order Logic for Factual Reasoning over Knowledge Graphs (2026.acl-long)

Challenge:	evidence from deployed systems suggests that language models interact through a shared data ecosystem.
Approach:	They propose to use data pollination to investigate stability dynamics under synthetic data training to investigate model collapse.
Outcome:	The proposed model can mitigate model collapse observed in recursive training, and improve performance across benchmarks.

Copied to clipboard

Yuanzhen Hao, Desheng Wu

Diffuse Thinking: Exploring Diffusion Language Models as Efficient Thought Proposers for Reasoning (2026.acl-long)

Challenge:	Existing methods for factual reasoning over knowledge graphs lack support for multiple quantifiers and connectives.
Approach:	They propose an extended FOL -structure over knowledge graphs that incorporates comparison predicates and counting quantifiers.
Outcome:	The proposed method achieves state-of-the-art on Fact-FOLX-KG, while previous methods experience performance drop on claims requiring comparison and counting.

Copied to clipboard

Chenyang Shao, Sijian Ren, Fengli Xu, Yong Li

Does Memory Need Graphs? A Unified Framework and Empirical Analysis for Long-Term Dialog Memory (2026.acl-long)

Challenge:	Large language models (LLMs) have demonstrated strong capabilities in complex reasoning tasks, yet their autoregressive generation paradigm makes it computationally prohibitive to explore diverse reasoning paths.
Approach:	They propose a framework that combines diffusion-based generation with autoregressive evaluation to efficiently generate diverse intermediate reasoning thoughts and employ LLMs as evaluators to assess and select candidates based on their plausibility and correctness.
Outcome:	The proposed framework improves inference efficiency while maintaining competitive or superior reasoning accuracy.

Copied to clipboard

Sen Hu, Yuxiang Wei, Jiaxin Ran, Xueran Han, Zhiyuan Yao, Huacan Wang, Ronghao Chen, Lei Zou

False Friends in the Shell: Unveiling the Emoticon Semantic Confusion in Large Language Models (2026.acl-long)

Challenge:	Existing literature on dialog memory systems is inconsistent on their effectiveness . empirical findings on graph structures are difficult to attribute to specific design choices .
Approach:	They propose a framework that decomposes dialog memory systems into core components . they conduct stage-wise experiments on LongMemEval and HaluMeM, and compare implementation details .
Outcome:	The proposed framework compares graph-based and non-graph memory architectures on long-term dialog memory systems.

Copied to clipboard

Weipeng Jiang, Xiaoyu Zhang, Juan Zhai, Shiqing Ma, Chao Shen, Yang Liu

DocLens: A Tool-Augmented Multi-Agent Framework for Long Visual Document Understanding (2026.acl-long)

Challenge:	Emoticons are widely used in digital communication to convey affective intent, yet their safety implications for Large Language Models (LLMs) remain largely unexplored.
Approach:	They propose to use ASCII-based emoticons to perform unintended actions in large language models (LLMs) This vulnerability is pervasive, with an average confusion ratio exceeding 38%, and 90% of confused responses yield 'silent failures' authors call on the community to recognize this emerging vulnerability and develop effective mitigation methods to uphold the safety and reliability of human-LLM interactions.
Outcome:	The proposed framework exploits emoticon semantic confusion in six LLMs and demonstrates that existing prompt-based mitigations are ineffective.

Copied to clipboard

Dawei Zhu, Rui Meng, Jiefeng Chen, Sujian Li, Tomas Pfister, Jinsung Yoon

Stability Implies Redundancy: Delta Attention Selective Halting for Efficient Long-Context Prefilling (2026.acl-long)

Challenge:	Existing approaches to localizing evidence from long visual documents fail on a fundamental challenge: evidence localization.
Approach:	They propose a tool-augmented multi-agent framework that “zooms in” on evidence like a lens.
Outcome:	The proposed framework achieves state-of-the-art performance on MMLongBench-Doc and FinRAGBench-V, surpassing even human experts.

Copied to clipboard

Yujie Chen, Tailai Chen, Yifeng Gao, Zoe Wanying He, Yijue Xu, Shaobo Wang, Linfeng Zhang

Beyond Self-Report: Bridging the Intention-Behavior Gap in Critical Thinking Assessment via Interpretable Multi-Agent System (2026.acl-long)

Challenge:	Existing methods to reduce sequence length rely on heuristics that break compatibility with hardware-efficient kernels like FlashAttention.
Approach:	They propose a method that selectively halts stabilized tokens by monitoring layer-wise update dynamics of the self-attention mechanism.
Outcome:	The proposed method can reduce prefill complexity while preserving model accuracy and hardware efficiency.

Copied to clipboard

Zekun Li, Jifan Yu, Haoxuan Li, Ye He, Daniel Zhang-Li, Shangqing Tu, Joy Jia Yin Lim, Yikun Jiang, Jiaxin Yuan, Yu Zhang

LLM Reasoning as Trajectories: Step-Specific Representation Geometry and Correctness Signals (2026.acl-long)

Challenge:	Accurate assessment of critical thinking is limited by the Intention Behavior Gap in psychology . evaluators that measure self-reported competence are limited by multiagent architectures .
Approach:	They propose a framework that operationalizes cognitive assessment into an interpretable multi-agent workflow with Assessment Chain-of-Thought.
Outcome:	The proposed framework aligns better with human expert ratings than gold-standard inventories on large-scale simulations and human participants.

Copied to clipboard

Lihao Sun, Hang Dong, Bo Qiao, Qingwei Lin, Dongmei Zhang, Saravan Rajmohan

A Data-Efficient Path to Multilingual LLMs: Language Expansion via Post-training PARAM𝛥 Integration into Upcycled MoE (2026.acl-long)

Challenge:	Existing models generate tokens by updating high-dimensional representations and decoding from them at each timestep.
Approach:	They propose a framework that allows reasoning correction and length control based on derived ideal trajectories.
Outcome:	The proposed model can predict correctness and length control based on ideal trajectories.

Copied to clipboard

Hao Zhou, Tianhao Li, Zhijun Wang, Shuaijie She, Linjuan Wu, Hao-Ran Wei, Baosong Yang, Jiajun Chen, Shujian Huang

The Evolution of Thought: Tracking LLM Overthinking via Reasoning Dynamics Analysis (2026.acl-long)

Challenge:	Large Language Models (LLMs) are expensive and require extensive Continued Pre-Training and data-intensive alignment to expand.
Approach:	They propose a method which upcycles a dense model into a Mixture-of-Experts architecture, allocating different experts to different languages.
Outcome:	Experiments show that the proposed model upcycles a dense model into a Mixture-of-Experts(MoE) architecture, allocating different experts to different languages.

Copied to clipboard

Zihao Wei, Liang Pang, Jiahao Liu, Wenjie Shi, Jingcheng Deng, Shicheng Xu, Zenghao Duan, Jingang Wang, Fei Sun, Huawei Shen, Xueqi Cheng

CoCoNUTS: Concentrating on Content while Neglecting Uninformative Textual Styles for AI-Generated Peer Review Detection (2026.acl-long)

Challenge:	Explicit reasoning trajectories increase performance but often trigger overthinking . despite its importance, this study examines how each step of reasoning affects the final outcome .
Approach:	They propose a Reasoning Completion Point Detector that detects the RCP by monitoring rank dynamics of termination tokens.
Outcome:	The proposed method reduces token usage by up to 44% while preserving accuracy.

Copied to clipboard

Yihan Chen, Jiawei Chen, Guozhao Mo, Xuanang Chen, Ben He, Xianpei Han, Le Sun

Challenge:	Existing detectors rely on stylistic cues to distinguish between surface-level language refinement and genuine content generation.
Approach:	They propose a content-based detection paradigm to detect substantive AI-generation . they propose 'CoCoDet' detector that can detect surface-level language refinement .
Outcome:	The proposed detector achieves a macro F1 score of 98.24% on permissible machine-polished reviews and maintains 3.89% false positive rate on real-world reviews.

Can LLMs Estimate Cognitive Complexity of Reading Comprehension Items? (2026.acl-long)

Copied to clipboard

Seonjeong Hwang, Hyounghun Kim, Gary Lee

ReCoQA: A Benchmark for Tool-Augmented and Multi-Step Reasoning in Real Estate Question and Answering (2026.acl-long)

Challenge:	Existing methods for estimating the cognitive complexity of reading comprehension items are expensive, time-consuming, and subject to rater variability.
Approach:	They propose to use two dimensions to estimate cognitive complexity of RC items to focus on evidence Scope and transformation level to estimate the cognitive complexity.
Outcome:	The proposed models can estimate the cognitive complexity of items by focusing on two dimensions—Evidence Scope and Transformation Level—that indicate the degree of cognitive burden involved in reasoning about the answer.

Copied to clipboard

Yindong Zhang, Wenmian Yang, Yiquan Zhang, Weijia Jia

Human or LLM as Standardized Patients? A Comparative Study in Medical Education (2026.acl-long)

Challenge:	Real estate agents are labor-intensive, difficult to scale, and prone to interest-driven bias.
Approach:	They propose a large-scale benchmark of 29,270 real-estate instances with machine-verifiable supervision for intermediate steps . they propose 'hIRE-Agent' framework that integrates heterogeneous evidence into an understand–plan–execute architecture as a strong baseline .
Outcome:	Experiments show that HIRE-Agent integrates heterogeneous evidence . the framework is able to integrate a front-end parser, planning Supervisor, and execution Specialists .

Copied to clipboard

Bingquan Zhang, Xiaoxiao Liu, Yuchi Wang, Zhou Lei, Qianqian Xie, Benyou Wang

Harmonizing the Past, Present, and Future: A Null-Space Constrained Region-Specific Method for Continual Learning in LLMs (2026.acl-long)

Challenge:	Standardized patients (VSPs) are indispensable for clinical skills training but remain expensive and difficult to scale.
Approach:	They propose a multi-agent VSP framework that separates case-grounded information disclosure from response generation to support stable, inquiry-conditioned patient behavior.
Outcome:	The proposed framework more closely matches human SP behavior than existing VSPs, particularly in case consistency and controlled disclosure.

Copied to clipboard

Jinhui Chen, Shizhu He, Xingchang Yang, Huanxuan Liao, Yequan Wang, Xiangwen Liao, Wenhao Teng, Kang Liu, Jun Zhao

Attention Weights as an Indicator: Analyzing and Improving Document Utilization in Retrieval-Augmented Generation (2026.acl-long)

Challenge:	Existing continual learning paradigms prioritize instant performance through dense updates, leading to catastrophic forgetting and rapid exhaustion of model capacity.
Approach:	They propose a method that preserves previously acquired knowledge and acquires new task-specific skills while preserving sufficient parameter capacity for subsequent adaptation.
Outcome:	The proposed method is based on the brain's functional partitioning and can be used to map tasks between specialized and generalist neurons.

Copied to clipboard

Jing Jin, Yuhan Song, Wen Luo, Houfeng Wang

Inertia in Moral and Value Judgments of Large Language Models (2026.acl-long)

Challenge:	In traditional RAG models, documents are grouped into categories based on their quality and order, and the quality of inputs is variable due to ineffective retrievers or misalignment between the retriever and generator.
Approach:	They propose to use attention weights to enhance document utilization from three perspectives: document ranking, placement, and filtering.
Outcome:	The proposed method outperforms baselines and improves document utilization effectiveness in a training-free manner.

Copied to clipboard

Bruce W. Lee, Yeongheon Lee, Hyunsoo Cho

LiveCultureBench: a Multi-Agent, Multi-Cultural Benchmark for Large Language Models in Dynamic Social Simulations (2026.acl-long)

Challenge:	Large Language Models behave non-deterministically, and prompting is a common method for steering their outputs.
Approach:	They use role-play at scale to study the value orientation and inertia of Large Language Models.
Outcome:	The proposed model keeps values skewed in one direction across persona settings.

Copied to clipboard

Viet Thanh Pham, Lizhen Qu, Thuy-Trang Vu, Gholamreza Haffari, Dinh Phung

Enabling Agents to Communicate Entirely in Latent Space (2026.acl-long)

Challenge:	Large language models (LLMs) are increasingly deployed as autonomous agents . evaluations focus primarily on task success rather than cultural appropriateness or reliability.
Approach:	They propose a multi-cultural, dynamic benchmark that embeds large language models as agents in a simulated town and evaluates them on task completion and adherence to socio-cultural norms.
Outcome:	The proposed model evaluates LLMs on task completion and adherence to socio-cultural norms across models and cultural profiles.

Copied to clipboard

Zhuoyun Du, Runze Wang, Huiyu Bai, Zouying Cao, Xiaoyong Zhu, Yu Cheng, Bo Zheng, Wei Chen, Haochao Ying

Challenge:	Natural language is the de facto communication medium for LLM-based agents, but it presents a fundamental constraint . natural language downsampling limits the depth and nuance of information that can be transmitted . et al.: inter-agent latent space communication is a promising paradigm for solving complex tasks .
Approach:	They propose a paradigm that leverages the last hidden states of an LLM as a representation of its thought for direct communication.
Outcome:	The proposed paradigm outperforms fine-tuned chain-of-thought prompting and single-agent baselines even across heterogeneous models.

DR-Arena: an Automated Evaluation Framework for Deep Research Agents (2026.acl-long)

Copied to clipboard

Yiwen Gao, Ruochen Zhao, Yang Deng, Wenxuan Zhang

Challenge:	Existing benchmarks for evaluating deep research capabilities rely on static datasets.
Approach:	They propose a fully automated evaluation framework that pushes DR agents to their capability limits through dynamic investigation.
Outcome:	DR-Arena achieves a Spearman correlation of 0.94 with the LMSYS Search Arena leaderboard.

Question Tells You Where the Answer Is: Intention-aware Long-Context KV Cache Compression (2026.acl-long)

Copied to clipboard

Liang Zhao, Xiaocheng Feng, Weihong Zhong, Lei Huang, Kun Zhu, Baoxin Wang, Dayong Wu, Guoping Hu, Ting Liu, Bing Qin

Challenge:	Recent methods to reduce the KV cache size fail to identify crucial KVs for generation while excluding others accurately, resulting in severe information loss.
Approach:	They propose an intention-aware KV cache eviction method that identifies and retains crucial KVs according to the attention distribution of intention, which semantically reflects the user’s goal and determines which part of the context is relevant.
Outcome:	The proposed method can maintain the model performance while reducing the KV cache size from 128K to 2K, leading to a 6.3x increase in decoding speed and 7.8x enhancement in memory efficiency compared to the default setting.

LADR: Locality-Aware Dynamic Rescue for Efficient Text-to-Image Generation with Diffusion Large Language Models (2026.acl-long)

Copied to clipboard

Chenglin Wang, Yucheng Zhou, Shuang Chen, Tao Wang, Kai Zhang

PACE: Predictive Adaptive Context Extraction for Long-Horizon LLM Agents (2026.acl-long)

Challenge:	Existing methods for inference are expensive and lack spatial redundancy . Discrete Diffusion Language Models are a promising paradigm for multimodal generation .
Approach:	They propose a locality-aware dynamic rescue method that exploits spatial Markov property of images.
Outcome:	The proposed method achieves an approximate 4 speedup over baselines on four text-to-image generation benchmarks.

Copied to clipboard

Lei Wei, Xiao Peng, null Tt, Guannan Zhang, Chenhao Jiang, Hongyu Li, Lanbo Lin, Yuanwu Xu, Jiayao Liu, Kesu Wang, Bin Wang

PaCoRe: Learning to Scale Test-Time Compute with Parallel Coordinated Reasoning (2026.acl-long)

Challenge:	Large Language Model (LLM) agents struggle with ultra-long-horizon tasks requiring hundreds or thousands of interaction steps.
Approach:	They propose a framework that reconceptualizes context management as a Next Step Prediction problem.
Outcome:	The proposed framework improves task success rates and robust cross-lingual performance.

Copied to clipboard

Jingcheng Hu, Yinmin Zhang, Shijie Shang, Xiaobo Yang, Yue Peng, Zhewei Huang, Hebin Zhou, Xin Wu, Jie Cheng, Fanqi Wan, Xiangwen Kong, Chengyuan Yao, Kaiwen Yan, Ailin Huang, Hongyu Zhou, Qi Han, Zheng Ge, Xiangyu Zhang, Heung-Yeung Shum

Draft, Verify, Restore: Self-Refining Historical Inscription Restoration with a Unified MLLM (2026.acl-long)

Challenge:	Parallel Coordinated Reasoning (PaCoRe) overcomes a central limitation of contemporary language models: their inability to scale test-time compute (TTC) far beyond sequential reasoning under a fixed context window.
Approach:	They propose a training-and-inference framework to overcome a central limitation of language models: their inability to scale test-time compute (TTC) under a fixed context window.
Outcome:	The proposed model scales to multi-million-token effective TTC without exceeding context limits.

Copied to clipboard

Yuyi Zhang, Junle Liu, Peirong Zhang, Jianliang Liu, Zhenhua Yang, Lianwen Jin

Challenge:	Existing methods for end-to-end historical inscription restoration rely on task-separated pipelines with irreversible error accumulation and patch-based generation that sacrifices page-level consistency.
Approach:	They propose a unified MLLM for end-to-end historical inscription restoration that integrates draft-guided localization and Hierarchical self-refinement to enable accurate damage localization.
Outcome:	The proposed model achieves superior performance in both text restoration accuracy and appearance restoration quality.

SAM3-I: Segment Anything with Instructions (2026.acl-long)

Copied to clipboard

Jingjing Li, Yue Feng, Yuchen Guo, Jincai Huang, Wei Ji, Qi Bi, Yongri Piao, Miao Zhang, Xiaoqi Zhao, Qiang Chen, Shihao Zou, Huchuan Lu, Li Cheng

Stop When Enough: Adaptive Early-Stopping for Chain-of-Thought Reasoning (2026.acl-long)

Challenge:	Existing methods for concept-level grounding and instruction-level reasoning use coarse representations and iterative mask filtering.
Approach:	They propose an instruction-following extension of the Segment Anything Model 3 family that unifies concept-level grounding and instruction-level reasoning within a single segmentation framework.
Outcome:	Experiments show that SAM3-I achieves appealing performance across referring and reasoning-based segmentation while maintaining its strong concept recall ability.

Copied to clipboard

Renliang Sun, Wei Cheng, Dawei Li, Haifeng Chen, Wei Wang

Improving Long-Context Translation via Self-Supervised Dual Learning (2026.acl-long)

Challenge:	Chain-of-Thought reasoning has driven recent gains of large language models (LLMs) on reasoning-intensive tasks by externalizing intermediate steps.
Approach:	They propose a training-free framework that adaptively determines when to stop reasoning to mitigate overthinking.
Outcome:	The proposed framework reduces token usage by 20-55% while maintaining or improving accuracy compared to standard CoT prompting.

Copied to clipboard

Shanbo Cheng, Shuaijie She, Yu Bao, Jianbing Zhang, Jiajun Chen, Shujian Huang

CaRL-EM: Cost-Aware Reinforcement Learning for Entity Matching with LLMs (2026.acl-long)

Challenge:	Large language models with long context windows suffer from catastrophic information distortion, undermining the strict faithfulness required for translation.
Approach:	They propose a self-supervised post-training framework that improves long-document translation reliability via round-trip consistency.
Outcome:	The proposed framework improves long-document translation reliability via round-trip consistency.

Copied to clipboard

Chaohui Guo, Michel C. A. Klein, Zhisheng Huang

Challenge:	Entity matching (EM) requires fine-grained contextual understanding and domain knowledge.
Approach:	They propose a reinforcement learning controller that manages LLM operations by combining multiple operators and a set of model capacities.
Outcome:	The proposed controller can be reused with different LLM backends at inference time without retraining.

Audio Jailbreak: An Open Comprehensive Benchmark for Jailbreaking Large Audio-Language Models (2026.acl-long)

Copied to clipboard

Zirui Song, Qian Jiang, Mingxuan Cui, Mingzhe Li, Lang Gao, Zeyu Zhang, Zixiang Xu, Yanbo Wang, Guangxian Ouyang, Zhenhao Chen, Xiuying Chen

When Personalization Legitimizes Risks: Uncovering Safety Vulnerabilities in Personalized Dialogue Agents (2026.acl-long)

Challenge:	a recent study evaluated large audio-language models against jailbreak attacks . a new benchmark is being developed to evaluate LAM safety against jailbreaking attacks based on temporal and semantic nature of speech .
Approach:	They propose a benchmark to evaluate LAM jailbreak vulnerabilities in adversarial audio prompts . they use a dataset of 1,495 adversarials to evaluate their performance .
Outcome:	The proposed benchmark evaluates state-of-the-art LAMs against jailbreak attacks . it demonstrates that even small, semantically preserved perturbations can reduce safety .

Copied to clipboard

Jiahe Guo, Xiangran Guo, Yulin Hu, Zimo Long, Xingyu Sui, Xuda Zhi, Yongbo Huang, Hao He, Weixiang Zhao, Yanyan Zhao, Bing Qin

AlphaContext: An Evolutionary Tree-based Psychometric Context Generator for Creativity Assessment (2026.acl-long)

Challenge:	Existing research on personalized LLM agents focuses on the effectiveness of personalized responses.
Approach:	They propose a benchmark to quantify intent legitimation in personalized interactions . they propose 'detection-reflection' method that detects intent legititimation from internal representation space .
Outcome:	The proposed method reduces safety degradation by using internal representation space.

Copied to clipboard

Yixuan Wang, Yue Huang, Hong Qian, Yunzhao Wei, Yifei Ding, Wenkai Wang, Zhi Liu, Zhongjing Huang, Aimin Zhou, Jiajun Guo

Challenge:	Existing LLM-based tools struggle with insufficient assessment cues, weak narrative coherence, limited stylistic diversity, and poor support for creative thinking.
Approach:	They propose an evolutionary tree-based psychometric context generator that integrates rule-guided outline planning, sentence-level MCTS generation, MAP-Elites quality-diversity optimization and assessment-guide refiner simulation.
Outcome:	The proposed tool outperforms strong LLMs and structured frameworks on 7 evaluation dimensions and shows higher alignment with expert-designed contexts.

Demystifying Data Organization for Enhanced LLM Training (2026.acl-long)

Copied to clipboard

Yalun Dai, Yangyu Huang, Tongshen Yang, Yonghan Wang, Xin Zhang, Wenshan Wu, Qihao Zhao, Hao Li, Yuanyuan Gao, Kim-Hui Yap, Scarlett Li

Beyond Single View: A Comprehensive Benchmark for Medical Multimodal Large Language Models on Multi-Image Understanding (2026.acl-long)

Challenge:	Large Language Models (LLMs) have revolutionized various fields, yet their training efficiency is heavily reliant on effective data curation.
Approach:	They propose to reuse pre-computed sample-level scores originally generated for data efficiency and introduce two new data ordering methods to improve LLM training.
Outcome:	The proposed methods improve the stability and performance of LLM training.

Copied to clipboard

Dexuan Xu, Yuan Jiayin, Jianing Wang, Yanyuan Chen, Hanpin Wang, Yu Huang

When Seeing Is not Enough: Revealing the Limits of Active Reasoning in MLLMs (2026.acl-long)

Challenge:	Existing benchmarks for multimodal large language models are limited to multiview diagnostics.
Approach:	They propose a benchmark specifically designed for medical multi-image understanding that evaluates MLLMs across four dimensions.
Outcome:	The proposed model performs better in multi-image contexts than open-source models . the model perform better when processing increased visual loads than closed-source ones .

Copied to clipboard

Hongcheng Liu, Pingjie Wang, Yuhao Wang, Siqu Ou, Yanfeng Wang, Yu Wang

CuMA: Aligning LLMs with Sparse Cultural Values via Demographic-Aware Mixture of Adapters (2026.acl-long)

Challenge:	Existing evaluations of multimodal large language models focus on passive inference, where seeing is not enough.
Approach:	They propose a benchmark to evaluate active reasoning in multimodal large language models . they propose to acquire missing evidence and iteratively refine decisions under incomplete information .
Outcome:	The proposed model performs better on active reasoning than on passive inference settings.

Copied to clipboard

Ao Sun, Xiaoyu Wang, Zhe Tan, Yu Li, Zhu Jiachen, Yuheng Jia, Shu Su

Neo-Classic: A Benchmark for Evaluating Linguistic-Aesthetic Reasoning in Classical Chinese Poetry (2026.acl-long)

Challenge:	Large Language Models (LLMs) have a global audience, so alignment must extend to cultural resonance.
Approach:	They propose a framework that frames alignment as a conditional capacity separation problem.
Outcome:	The proposed framework outperforms both dense baselines and semantic-only MoEs on three large language models.

Copied to clipboard

Han Zhang, Zihan Gu, Zhiyuan Wang, Tianyi Ma, Jiacheng Lu, Xinyan Zhang, Yuhao Wei, Cheng Hua

A Multi-Agent Framework for Feature-Constrained Difficulty Control in Reading Comprehension Item Generation (2026.acl-long)

Challenge:	Large Language Models (LLMs) achieve high accuracy on established Classical Chinese Poetry benchmarks, but it remains challenging to distinguish transferable Linguistic-Aesthetic Reasoning from reliance on familiar pre-training patterns.
Approach:	They propose a benchmark that combines a constructionist Out-of-Sample dataset with reverse understanding probes to evaluate large-scale large-format models.
Outcome:	The proposed model performs well on classical Chinese poetry benchmarks, but a performance gap persists . the model can complete famous couplets and can be used to understand a variety of texts.

Copied to clipboard

Seonjeong Hwang, Jun Seo, Hyounghun Kim, Gary Lee

Machine Reading Comprehension

Challenge:	Existing methods for difficulty-controlled reading comprehension item generation rely on a single agent prompting approach.
Approach:	They propose a multi-agent framework for Feature-constrained Item Generation where multiple LLM agents collaborate to generate and iteratively revise items based on intended constraints.
Outcome:	The proposed method generates items with monotonically increasing difficulty at higher rates than baselines.

ToMMeR - Efficient Entity Mention Detection from Large Language Models (2026.acl-long)

Copied to clipboard

Victor Morand, Nadi Tomeh, Josiane Mothe, Benjamin Piwowarski

Shuttle Between Symbolic Instructions and Neural Parameters of Large Language Models (2026.acl-long)

Challenge:	Existing methods to detect text spans that refer to entities are often conflated with entity typing in a single joint task.
Approach:	They propose a lightweight model that probes mention detection capabilities from early LLM layers.
Outcome:	The proposed model achieves 93% recall zero-shot with 90% precision under human-calibrated LLM-judge protocol .

Copied to clipboard

Wangtao Sun, Haotian Xu, Huanxuan Liao, Xuanqing Yu, Zhongtao Jiang, Shizhu He, Jun Zhao, Kang Liu

Mitigating Spurious Correlations in Text Classification Using Latent Space Geometry (2026.acl-long)

Challenge:	Despite their distinct external representations, a deeper analysis reveals their intrinsic nature: instructions serve as a natural language compression devised by humans for data governing specific mapping patterns, whereas parameters act as 'neuro compression' of the same task data.
Approach:	They propose a neural network framework to model and learn the bi-directional mappings between instructions and parameters of large language models by evaluating it on the tasks of instruction deduction and induction.
Outcome:	The proposed framework can map one of the instructions/parameters to the other by evaluating it on the tasks of instruction deduction and induction.

Copied to clipboard

Jiasen Gao, Xiaoliang Chen, Duoqian Miao, Xu Gu, Xianyong Li, Yajun Du

Efficient Hyperparameter Optimization for LLM Reinforcement Learning (2026.acl-long)

Challenge:	Existing models rely on predictive shortcuts that hold in training data but break under distribution shifts, leading to large performance drops for minority groups.
Approach:	They propose a framework that transforms abstract biases into interpretable geometric anchors without auxiliary classifiers by manipulating latent space geometry.
Outcome:	The proposed framework outperforms state-of-the-art baselines and improves worst-group accuracy by over 20% on the CivilComments dataset.

Copied to clipboard

Minping Chen, Bowen Xiao, Du Liang, Chuxuan Zeng, Zeyi Wen

Hetero-Designer: Automated Design of Multi-Agent Systems with Heterogeneous LLMs (2026.acl-long)

Challenge:	Existing hyperparameter optimization methods are inefficient in reinforcement learning due to model scale and resource-intensive training cycles.
Approach:	They propose a hyperparameter optimization method that adapts both model size and training budget as fidelity.
Outcome:	The proposed method significantly improves the computational efficiency of each trial (up to 14.9) over existing HPO methods.

Copied to clipboard

Zhiheng Zhang, Yuanzhe Zhang, Bohan Yu, Daojian Zeng, Kang Liu, Jun Zhao

Challenge:	Existing approaches to design LLM-based Multi-agent systems are constrained by homogeneous LLMs.
Approach:	They propose an automated design of heterogeneous-LLMs-based MAS with a binary-star transformer and an autoregressive graph generation pipeline.
Outcome:	The proposed pipeline is high-performing on various benchmarks and extensible to unseen LLMs and roles.

Achieving Multi-Hop Calculation and Safe Abstention in Financial Numerical Reasoning by Metric Graph Constrained LLMs (2026.acl-long)

Copied to clipboard

Aoyuan Jiang, Liang Hong, Haoxuan Liu, Rui Wang

SOCIA-EVO: Automated Simulator Construction via Dual-Anchored Bi-Level Optimization (2026.acl-long)

Challenge:	Large language models (LLMs) are prone to forced generation when confronting ambiguous evidence or complex recursive dependencies.
Approach:	They propose a framework that imposes semantic and structural constraints via a financial metric knowledge graph.
Outcome:	a neuro-symbolic framework outperforms existing models on financial metric knowledge graphs.

Copied to clipboard

Yuncheng Hua, Sion Weatherhead, Mehdi Jafari, Hao Xue, Flora D. Salim

Challenge:	Large Language Models (LLMs) demonstrate strong capabilities in translating natural language into code, but applying them to this domain remains challenging.
Approach:	They propose a dual-anchored evolutionary framework that combines a static blueprint and a bi-level optimization to decouple structural refinement from parameter calibration.
Outcome:	The proposed framework identifies two failure modes in long-horizon LLM agents: contextual drift and optimization instability arising from conflating structural and parametric errors.

SoundBreak: A Systematic Study of Audio-Only Adversarial Attacks on Trimodal Models (2026.acl-long)

Copied to clipboard

Aafiya Shamshad Hussain, Gaurav Srivastava, Alvi Md Ishmam, Zaber Ibn Abdul Hakim, Chris Thomas

Trust Within? Seek Beyond? Knowledge Boundary Aware Policy Optimization for Agentic Search (2026.acl-long)

Challenge:	Recent advances in multimodal large language models have increased their vulnerability to adversarial manipulation.
Approach:	They propose to target audio-only adversarial attacks on multimodal audio–video–language models . they show that attacks can be successful at low perceptual distortions .
Outcome:	The proposed models achieve up to 96% success rate under realistic conditions . the proposed models are more robust to noise than to noise and distortion than to speech recognition systems .

Copied to clipboard

Tao Feng, Xinke Jiang, Xinyan Hu, Yonggang Zhang, Zhen Tao, Wentao Zhang, Boyang Liu, Wenhao Jiang, Chao Wu

RPC-Bench: A Fine-grained Benchmark for Research Paper Comprehension (2026.acl-long)

Challenge:	Existing approaches to augment large language models with external knowledge suffer from a lack of calibration regarding the model’s knowledge boundary.
Approach:	They propose a reinforcement learning framework that explicitly aligns retrieval decisions with quantified knowledge states.
Outcome:	The proposed framework outperforms strong baselines while exhibiting reduced hallucination rates.

Copied to clipboard

Yelin Chen, Fanjin Zhang, Suping Sun, Yunhe Pang, Yuanchun Wang, Jian Song, XiaoYan Li, Lei Hou, Shu Zhao, Jie Tang, Juanzi Li

Challenge:	Existing benchmarks for understanding research papers offer limited fine-grained evaluation at scale.
Approach:	They propose a large-scale question-answering benchmark built from review–rebuttal exchanges of high-quality computer science papers.
Outcome:	The proposed model is based on human-verified QA pairs and contains 15K questions.

FusionFlow: Enabling Deep Structural Exploration for Automated Agentic Workflow Generation (2026.acl-long)

Copied to clipboard

Xiang Wang, Zongtao Yang, Zhuojian Hong, Shuhao Zhang, Wei Wei

CompTab: A Comprehensive Benchmark for Real-World TableQA with Complex Reasoning and Irregular Tables (2026.acl-long)

Challenge:	Existing workflow generation methods rely on incremental refinement or tree-based search over a single evolving workflow.
Approach:	They propose a framework centered on workflow fusion that synthesizes multiple independently evolved workflows and allows exploration of deeper regions of the workflow space within a finite budget.
Outcome:	Experiments show that FusionFlow outperforms existing workflow generation methods on six reasoning benchmarks.

Copied to clipboard

Zhen Yang, Wei Du, Jie Wang, Wenze Zhou, Xiangfeng Meng, Zhengyang Wang, Suping Sun, Ziwei Du, Haodong Zou, Jie Chen, Yongbin Liu, Shicheng Tan, Jiahao Ying, Shu Zhao

Jianzhu Bao, Haozhen Zhang, Kuicai Dong, Bozhi Wu, Sarthak Ketanbhai Modi, Zi Pong Lim, Yon Shin Teo, Wenya Wang

Challenge:	Existing benchmarks focus on well-structured tables and fail to reflect irregular structures and complex reasoning commonly encountered in real-world scenarios.
Approach:	They propose a benchmark to evaluate TableQA under complex reasoning and irregular table conditions.
Outcome:	The proposed framework improves generalization and realism of large language models under complex and irregular table conditions.

Copied to clipboard

Video-MMMU: Evaluating Knowledge Acquisition from Multidisciplinary Professional Videos (2026.acl-long)

Challenge:	Chart understanding is a critical capability for vision-language models, serving as a cornerstone for automated data analysis, document understanding, and scientific research.
Approach:	They propose a chart-efficient training framework to enhance counterfactual sensitivity by code modification and a similarity-based data selection strategy.
Outcome:	The proposed framework achieves superior or comparable performance to strong chart-specific VLMs while using significantly less training data.

Copied to clipboard

Kairui Hu, Penghao Wu, Fanyi Pu, Wang Xiao, Xiang Yue, Bo Li, Yuanhan Zhang, Ziwei Liu

Semantically Comprehensive Token Pruning in LVLMs via Maximizing Concept Coverage (2026.acl-long)

Challenge:	Existing video benchmarks do not evaluate the knowledge acquisition capabilities of Large Multimodal Models (LMMs) existing video benchmark focuses on static, general visual understanding tasks, without evaluating whether models can acquire knowledge dynamically.
Approach:	They propose a multi-modal, multi-discipline, multitrack benchmark that evaluates Large Multimodal Models’ ability to acquire knowledge from college-level, educational videos.
Outcome:	The proposed benchmark reveals a substantial gap between human learners and current Large Multimodal Models (LMMs) and focuses on improving their learning efficiency.

Copied to clipboard

Xueting Li, Qi Liu, Chenghao Xu, Xu Yang, Guangtao Lyu, Jiahua Li, Cheng Deng

OptiCo: Adaptive Distributed Training Optimization via Collaborative Agent Reasoning (2026.acl-long)

Challenge:	Existing visual token pruning methods leverage simple metrics derived from human experience, such as attention or similarity, to rank and select tokens within a highly entangled feature space.
Approach:	They propose a novel visual token pruning method that uses a concept-driven paradigm to quantify the Marginal Semantic Gain of each token's contribution to uncovered concepts.
Outcome:	The proposed method outperforms state-of-the-art methods in a concept-driven model while maintaining semantic completeness.

Copied to clipboard

Sheng Chen, Tang Zhe, Weixing Zhang, Fei Yang, Yuanyuan. Wang, Tianlin Li, Yang Liu

MemBuilder: Reinforcing LLMs for Long-Term Memory Construction via Attributed Dense Rewards (2026.acl-long)

Challenge:	Existing distributed training frameworks are plagued by over-reliance on prior profiling and poor generalization across models/hardware.
Approach:	They propose a model-driven multi-agent framework that leverages Large Language Models to enable automatic and explainable distributed training strategy configuration.
Outcome:	The proposed framework outperforms expert-designed training strategies within 20 iterations.

Copied to clipboard

Zhiyu Shen, Ziming Wu, Fuming Lai, Shaobing Lian, Yanghui Rao

Towards Fine-grained Audio Captioning with Multimodal Contextual Fusion (2026.acl-long)

Challenge:	Memory-augmented frameworks fail to capture temporal evolution of historical states, limiting consistency in long-term dialogues.
Approach:	They propose a framework that trains models to orchestrate multi-dimensional memory construction with attributed dense rewards.
Outcome:	The proposed framework outperforms state-of-the-art closed-source models and generalizes well to OOD benchmarks.

Copied to clipboard

Shunian Chen, Xinyuan Xie, Zheshu Chen, Owen Lee, Liyan Zhao, Zhan Su, Qilin Sun, Benyou Wang

Probing Semantic Alignment, Lexical Invariance, and Syntactic Influence in LLM Metaphor Processing (2026.acl-long)

Challenge:	Existing methods for audio captioning lack fine-grained detail and contextual accuracy due to limited unimodal or superficial information.
Approach:	They propose a two-stage automated pipeline that uses pretrained models to extract contextual cues from video . a large language model synthesizes these inputs to generate detailed and context-aware captions .
Outcome:	The proposed method is scalable and generates detailed and context-aware captions on large-scale audio datasets.

Copied to clipboard

Fengying Ye, Shanshan Wang, Lidia S. Chao, Derek F. Wong

MathCanvas: Intrinsic Visual Chain-of-Thought for Multimodal Mathematical Reasoning (2026.acl-long)

Challenge:	Large language models (LLMs) achieve strong performance on metaphor detection and interpretation tasks, yet it remains unclear what such success actually reveals about metaphor processing.
Approach:	They propose to probing semantic attribute alignment, lexical invariance, and syntactic sensitivity to examine the limits of behavioral evidence for metaphor processing.
Outcome:	The proposed model can exhibit semantic drift relative to reference attributes, stable lexical anchors persist across contextual conditions, potentially supporting conventional metaphors while biasing novel metaphors requiring contextual integration.

Copied to clipboard

Weikang Shi, Aldrich Yu, Rongyao Fang, Houxing Ren, Ke Wang, Aojun Zhou, Changyao Tian, Xinyu Fu, Yuxuan Hu, Zimu Lu, Linjiang Huang, Si Liu, Rui Liu, Hongsheng Li

Rethinking the Role of Entropy in Optimizing Tool-Use Behaviors for Large Language Model Agents (2026.acl-long)

Challenge:	Existing approaches to visual chain-of-thought are limited by external tools or fail to generate high-fidelity diagrams.
Approach:	They propose a framework to enable large multimodal models with VCoT capabilities . they pre-train a model on a 15.2M-pair corpus and teach it how to leverage visual aids .
Outcome:	The proposed framework unlocks complex, human-like visual reasoning in large language models . it pre-trains the model on a 15.2M-pair corpus and fine-tunes it on MathCanvas-Instruct .

Copied to clipboard

Zeping Li, Hongru Wang, Yiwen Zhao, Guanhua Chen, Yixia Li, Keyang Chen, Yixin Cao, Guangnan Ye, Hongfeng Chai, Zhenfei Yin

Collaboration of Fusion and Independence: Hypercomplex-driven Robust Multi-Modal Knowledge Graph Completion (2026.acl-long)

Challenge:	Large language models excel in mathematical reasoning and multi-hop question answering tasks, but in long trajectories, agents often invoke tools excessively or inappropriately, increasing computation cost and derailing the reasoning process.
Approach:	They propose to use entropy reduction as a supervisory signal to reduce tool calls . they propose to design two reward strategies to address the needs of optimizing tool-use behavior.
Outcome:	The proposed reward strategies reduce tool calls by 72.07% and improve performance by 22.27%.

Copied to clipboard

Zhiqiang Liu, Yichi Zhang, Mengshu Sun, Lei Liang, Wen Zhang

LLM-MC-Affect: LLM-Based Monte Carlo Modeling of Affective Trajectories and Latent Ambiguity for Interpersonal Dynamic Insight (2026.acl-long)

Challenge:	Existing multi-modal knowledge graphs lack modality-specific information and are limited in their ability to capture nuanced semantic interplay between modalities.
Approach:	They propose a multi-modal knowledge graph completion method which integrates both paradigms . they use a fine-grained Entity Representation Factorization module and a Robust Relation-aware Modality Fusion module to obtain robust representations for three independent modalities and one fused modality.
Outcome:	The proposed method achieves coexistence and collaboration of fused and independent modality representations while maintaining modality-specific information.

Copied to clipboard

Yu-Zheng Lin, Bono Po-Jen Shih, John Paul Martin Encinas, Elizabeth Victoria Achom, Karan Patel, Jesus Horacio Pacheco, Sicong Shao, Jyotikrishna Dass, Soheil Salehi, Pratik Satam

Multilingual and Cross-Lingual Citation Needed Detection on Wikipedia for Lower-Resource Languages (2026.acl-long)

Challenge:	Emotional coordination is a core property of human interaction that shapes relational meaning . prior approaches treat sentiment as a deterministic point estimate for individual speakers . scalable and deployable approach extends beyond education to broader social and behavioral research .
Approach:	They propose a probabilistic framework that characterizes emotion as a latent probability distribution defined over an affective space.
Outcome:	The proposed framework characterizes emotion as a latent probability distribution defined over affective space.

Copied to clipboard

Gerrit Quaremba, Amy Rechkemmer, Elizabeth Black, Denny Vrandečić, Elena Simperl

Process Reward Models Meet Planning: Generating Precise and Scalable Datasets for Step-Level Rewards (2026.acl-long)

Challenge:	Existing research has largely overlooked lower-resource languages for automated fact-checking.
Approach:	They propose a multilingual CND corpus spanning 18 languages across three resource levels and a small decoder-based language model for CND.
Outcome:	The proposed model outperforms prompted LLMs in cross-lingual CND across languages.

Copied to clipboard

Raffaele Pisano, Roberto Navigli

UniSonate: A Unified Model for Speech, Music, and Sound Effect Generation with Text Instructions (2026.acl-long)

Challenge:	Existing PRM datasets are expensive to construct and limited to the mathematical domain.
Approach:	They propose a method to generate a corpus of one million reasoning steps using the Planning Domain Definition Language.
Outcome:	The proposed model generates a corpus of approximately one million reasoning steps across various PDDL domains and trains them.

Copied to clipboard

Chunyu Qiang, Xiaopeng Wang, Kang Yin, Yuzhe Liang, Yuxin Guo, Teng Ma, Ziyu Zhang, Tianrui Wang, Cheng Gong, Yushen Chen, Ruibo Fu, Longbiao Wang, Jianwu Dang

AgentAsk: Multi-Agent Systems Need to Ask (2026.acl-long)

Challenge:	Generative audio modeling has been fragmented into specialized tasks such as text-to-speech (TTS), text- to-music (TTM), and text-ta (TTA) specialized models require reference audio for timbre cloning and strict phoneme alignment, whereas TTA models generate unstructured textures from open-ended captions.
Approach:	They propose a unified flow-matching framework capable of synthesizing speech, music, sound effects . they propose 'token injection mechanism' that projects unstructured environmental sounds into structured temporal latent space .
Outcome:	The proposed framework achieves state-of-the-art performance in instruction-based TTS and TTM while maintaining competitive fidelity in TTA.

Copied to clipboard

Bohan Lin, Kuo Yang, Zelin Tan, Yingchuan Lai, Chen Zhang, Guibin Zhang, Xinlei Yu, Miao Yu, Xu Wang, Yudong Zhang, Yang Wang

HopWeaver: Cross-Document Synthesis of High-Quality and Authentic Multi-Hop Questions (2026.acl-long)

Challenge:	Multi-agent systems fail to consistently outperform strong single-a agent baselines due to error propagation at inter-aggent message handoffs.
Approach:	They propose an edge-level error taxonomy that identifies four main errors in multi-agent interactions as data gaps, signal corruption, referential drift and capacity gaps as primary sources of failure.
Outcome:	The proposed module outperforms existing systems on five benchmarks and is architecture-agnostic.

Copied to clipboard

Zhiyu Shen, Jiyuan Liu, Yunhe Pang, Yanghui Rao, Fu Lee Wang, Jianxing Yu

REST: Stress Testing Large Reasoning Models by Asking Multiple Problems at Once (2026.acl-long)

Challenge:	Multi-Hop Question Answering (MHQA) is a critical benchmark for evaluating the model’s ability to integrate information from diverse sources.
Approach:	They propose a framework that synthesizes authentic multi-hop questions without manual annotation without the need for manual guidance.
Outcome:	The proposed framework synthesizes bridge and comparison questions without human intervention and achieves comparable or superior quality to human-annotated datasets at a lower cost.

Copied to clipboard

Zhuoshi Pan, Qizhi Pei, Yu Li, Zinan Tang, QiYao Sun, H. Vicky Zhao, Conghui He, Lijun Wu

You Can Have a Second Chance: Unbiased and Multi-bit Watermarking for Diffusion Language Models with Regret-based Remasking (2026.acl-long)

Challenge:	Recent Large Reasoning Models (LRMs) lack a narrow evaluation paradigm . a single-question evaluation setup suffers from two major limitations .
Approach:	They propose a stress-testing framework that exposes LRMs to multiple problems simultaneously.
Outcome:	The proposed framework outperforms existing models on reasoning benchmarks and state-of-the-art models.

Copied to clipboard

Ke Yang, Dongyang Liang, Jing Yu, Shuguang Yuan, Chi Chen

VCORE: Variance-Controlled Optimization-based Reweighting for Chain-of-Thought Supervision (2026.acl-long)

Challenge:	Existing sequential LLMs cannot be directly applied to DLMs, as their generation order is arbitrary.
Approach:	They propose a stability-aware constraint that allows watermarking only in stable contexts and a bit-controlled, unbiased modulation to preserve the original DLM output distribution.
Outcome:	The proposed scheme achieves stable watermarking with minimal quality impact while maintaining high detection accuracy and multi-bit capacity.

Copied to clipboard

Xuan Gong, Senmiao Wang, Hanbo Huang, Ruoyu Sun, Shiyu Liang

MAGNET: Towards Adaptive GUI Agents with Memory-Driven Knowledge Evolution (2026.acl-long)

Challenge:	Empirical evaluations demonstrate that VCORE achieves the strongest overall average performance, with especially clear gains on lower-capacity models.
Approach:	They propose a framework that reformulates supervision as a constrained optimization problem.
Outcome:	Empirical evaluations show that VCORE achieves the strongest overall average performance, with especially clear gains on lower-capacity models.

Copied to clipboard

Libo Sun, Jiwen Zhang, Siyuan Wang, Zhongyu Wei

Challenge:	Mobile GUI agents powered by large foundation models can perform tasks autonomously, but frequent updates that alter UI appearance and reorganize workflows cause agents trained on historical data to fail.
Approach:	They propose a memory-driven adaptive agent framework with stationary memory that links visual features to stable functional semantics and procedural memory that captures stable task intents across varying workflows.
Outcome:	The proposed framework improves performance over memory-augmented baselines and offline benchmarks on AndroidWorld.

VLN-MME: Diagnosing MLLMs as Language-guided Visual Navigation Agents (2026.acl-long)

Copied to clipboard

Xunyi Zhao, Gengze Zhou, Qi Wu

ImplicitMemBench: Measuring Unconscious Behavioral Adaptation in Large Language Models (2026.acl-long)

Challenge:	Multimodal Large Language Models have demonstrated remarkable capabilities across vision-language tasks, but their performance as embodied agents needs further exploration.
Approach:	They propose a framework to evaluate multimodal large language models as zero-shot agents . they find that enhancing prevalent agents with Chain-of-Thought reasoning and self-reflection leads to an unexpected performance decrease.
Outcome:	The proposed framework enables comparisons and component-level ablations across diverse MLLM architectures, agent designs, and navigation tasks.

Copied to clipboard

Chonghan Qin, Xiachong Feng, Weitao Ma, Xiaocheng Feng, Lingpeng Kong

Semantic Hardness Is Not Visual Hardness: Sign-Aware Hard Negative Mining for Sign Language Retrieval (2026.acl-long)

Challenge:	Existing memory benchmarks for LLMs evaluate explicit recall of facts, yet overlook implicit memory where experience becomes automated behavior without conscious retrieval.
Approach:	They propose a benchmark that evaluates implicit memory using three constructs from non-declarative memory.
Outcome:	The new benchmark reframes evaluation from "what agents recall" to "what they automatically enact" no model exceeds 66% overall, with top performers far below human baselines .

Copied to clipboard

Junmyeong Lee, Chan Hur, ChangSu Choi, Sukmin Cho, Fitsum Gaim, Eui Jun Hwang, Hoyun Song, KyungTae Lim

Difference in Task Performance on Sparse Speech Representations (2026.acl-long)

Challenge:	Existing methods for sign language retrieval fail to capture visual ambiguity . semantically distinct yet visually confusable signs are rarely treated as hard negatives .
Approach:	They propose a method that constructs hard negatives based on visual confusability rather than linguistic similarity.
Outcome:	The proposed method significantly improves fine-grained retrieval performance while preserving coarse-grain accuracy.

Copied to clipboard

Wenjie Peng, Chen Chen, Thomas Hain

VerilogLAVD: LLM-Aided Pattern Generation for Verilog CWE Detection (2026.acl-long)

Challenge:	Existing methods for learning speech representations that are useful for a variety of downstream tasks have been extensively investigated in different domains.
Approach:	They propose to train Autoencoders with varying sparsity levels using three SSL features and evaluate them on six tasks of SUPERB: speech enhancement, speaker identification, speech Emotion Recognition, phone recognition, automatic speech recognition and slot filling.
Outcome:	The proposed model can be used to learn speech representations that are useful for a variety of downstream tasks.

Copied to clipboard

Xiang Long, Yingjie Xia, Li Kuang, Yao Wan, ZiHao Liu

Challenge:	Existing static analysis tools focus on functional correctness and depend heavily on manual rules.
Approach:	They propose a framework that generates executable Traversal Detection Patterns (TDPs) to help detect hardware vulnerabilities.
Outcome:	The proposed framework improves the F1 score by 133% compared to LLM-based methods.

Temporal Sampling for Forgotten Reasoning in LLMs (2026.acl-long)

Copied to clipboard

Yuetai Li, Zhangchen Xu, Fengqing Jiang, Bhaskar Ramasubramanian, Luyao Niu, Bill Yuchen Lin, Xiang Yue, Radha Poovendran

Conflict-Aware Memory for Embodied Agents: Enhancing Vector Data Quality via Detection Rules (2026.acl-long)

Challenge:	a new metric measures the percentage of questions that were answered incorrectly during fine-tuning .
Approach:	They propose a decoding strategy that draws outputs from multiple checkpoints along the training trajectory.
Outcome:	The proposed method improves reasoning performance and consistency across benchmarks.

Copied to clipboard

Kexin Ma, Haotian Wang, Shenglin Chen, Yishuai Cai, null Huangyuyu, Ruochun Jin

Benchmarking Deflection and Hallucination in Large Vision-Language Models (2026.acl-long)

Challenge:	Embodied agents have successfully leveraged large language models (LLMs) to better transform human instructions and images into executable task plans.
Approach:	They propose Conflict Detection Rules to identify and manage data quality issues in vector knowledge bases and correct the index structure.
Outcome:	Experimental results show that planners with Conflict Detection Rules exceed the basic LLM planner by 15.25% and 14.25% in grammatical accuracy (GA) and interpretation accuracy (IA) on average.

Copied to clipboard

Nicholas Moratelli, Christopher Davis, Leonardo F. R. Ribeiro, Bill Byrne, Gonzalo Iglesias

STAPO: Selective Trajectory-Aware Policy Optimization for LLM Agent Training (2026.acl-long)

Challenge:	Existing benchmarks overlook conflicts between visual and textual evidence and the importance of generating deflections when incomplete knowledge is retrieved.
Approach:	They propose a dynamic curation pipeline that preserves benchmark difficulty over time . they propose 'vlm-DeflectionBench' benchmark to probe model behaviour under conflicting evidence .
Outcome:	The proposed benchmarks overlook conflicts between visual and textual evidence and are prone to obsolescence . the proposed benchmark is based on 2,775 samples spanning diverse retrieval settings .

Copied to clipboard

Qiuyi Qi, Tian Liang, Mutian Bao, Jinjian Zhang, Dongnan Liu, Wei Zhou, Linjian Mo, Ming Kong, Jie Liu, Feng Zhang, Qiang Zhu

Read As Human: Compressing Context via Parallelizable Close Reading and Skimming (2026.acl-long)

Challenge:	Prior work has explored step-level supervision using Shannon-entropy-based uncertainty signals, which conflate inherent state complexity with agent confidence.
Approach:	They propose a hierarchical group-based RL framework that leverages normalized entropy to locate outlier steps associated with trajectory neglect and optimizes them via a mechanism of trajectory-aware reward and trajectory-independent penalty.
Outcome:	Experiments on ALFWorld, WebShop, and Search-Augmented QA show that STAPO achieves state-of-the-art performance while substantially alleviating trajectory neglect.

Copied to clipboard

Jiwei Tang, Shilei Liu, Zhicheng Zhang, Qingsong Lv, Runsong Zhao, Tingwei Lu, Langming Liu, Haibin Chen, Yujin Yuan, Hai-Tao Zheng, Wenbo Su, Bo Zheng

Challenge:	Existing task-aware methods require loading the entire input sequence at once for compression, which suffer from computational inefficiency.
Approach:	They propose a framework that adopts an adaptive hybrid reading strategy to reduce computational inefficiency and redundant information in long-context scenarios.
Outcome:	Experiments show that RAM outperforms baselines on multiple question answering and summarization benchmarks while delivering up to a 12x speedup on long inputs.

OASIS: Mitigating Harmful Fine-tuning Attacks on LLMs via Orthogonal and Adaptive Safety Alignment Strategy (2026.acl-long)

Copied to clipboard

Jiayu Tang, Guowei Peng, Qiuhao Xie, Yuning Yang, Xiurui Xie, Guisong Liu

KoCo-Bench: Can Large Language Models Leverage Domain Knowledge in Software Development? (2026.acl-long)

Challenge:	Existing methods to decouple safety enforcement from harmful feature acquisition rely on perturbation directions that conflict with harmful gradients . harmful fine-tuning attacks pose a significant challenge for service providers aiming to uphold rigorous safety standards.
Approach:	They propose an orthogonal and ad hoc safety alignment strategy to decouple safety enforcement from harmful feature acquisition.
Outcome:	Experiments on four large language models show that OASIS reduces the Harmful Score by 60% compared to baselines while maintaining stable task utility.

Copied to clipboard

Xue Jiang, Ge Li, Jiaru Qian, Xianjie Shi, Chenjie Li, Hao Zhu, Ziyu Wang, Jielun Zhang, Zeyu Zhao, Kechi Zhang, Jia Li, Wenpin Jiao, Zhi Jin, Yihong Dong

Enhancing Two Steps Textual Anomaly Detection through Anisotropy Mitigation (2026.acl-long)

Challenge:	Existing domain-specific code benchmarks focus on assessing what knowledge LLMs possess rather than how they acquire and apply new knowledge.
Approach:	They propose a benchmark to evaluate domain specialization methods in real-world software development.
Outcome:	KOCO-bench is a new benchmark for evaluating domain specialization methods in real-world software development.

Copied to clipboard

Pierre Fihey, Matthieu Labeau, Pavlo Mozharovskyi

InfiniteWeb: Scalable Web Environment Synthesis for GUI Agent Training (2026.acl-long)

Challenge:	Recent approaches to anomaly detection focus on embeddings from pre-trained models . however, the geometric properties of pre-training embedders can hinder detection algorithms .
Approach:	They propose to apply anomaly detection algorithms to embeddings from pre-trained models to improve accuracy.
Outcome:	The proposed approach improves similarity-trained models by adapting embeddings to assumptions made by classical detection algorithms.

Copied to clipboard

Ziyun Zhang, Zezhou Wang, Xiaoyi Zhang, Zongyu Guo, Jiahao Li, Bin Li, Yan Lu

Challenge:	Existing GUI agent benchmarks are manually constructed and lack scale and diversity as training environments.
Approach:	They propose a GUI agent training system that automatically generates web environments at scale.
Outcome:	The proposed system outperforms commercial GUI agents at realistic website construction and improves on OSWorld and Online-Mind2Web.

MagicBench: Diagnosing Visual Agency Loss and Semantic Dependency in Multimodal LLMs (2026.acl-long)

Copied to clipboard

Tang Da Huang, Weidong Tang, Wen Qi Xu, Xianpeng Guo

Bloom-Eval: A Hierarchical Evaluation Benchmark for Automatic Survey Generation Based on Bloom’s Taxonomy (2026.acl-long)

Challenge:	MLLMs assume linguistic context invariably enhances visual understanding . a diagnostic benchmark is used to evaluate ML models under hierarchical linguistic interference .
Approach:	They propose a diagnostic benchmark to evaluate MLLMs under hierarchical linguistic interference.
Outcome:	The proposed benchmark compared 402 videos with a physical constraint set to evaluate MLLMs under hierarchical linguistic interference.

Copied to clipboard

Fei Zhang, Zhe Zhao, HaiBin Wen, Tianshuo Wei, Zaixi Zhang, Chao Yang, Ye Wei

Cultivating Forensic Reasoning for Generalizable Multimodal Manipulation Detection (2026.acl-long)

Challenge:	Existing evaluation methods suffer from cognitive dimensional simplification and methodological unreliability due to the ”LLM-as-a-Judge” approach.
Approach:	They propose a six-tiered benchmark that evaluates ASG systems by prioritizing deterministic algorithms and introducing a GRADE approach for abstract abilities.
Outcome:	The proposed method provides the ASG field with a systematic, reproducible, and theoretically grounded benchmark to guide future research.

Copied to clipboard

Yuchen Zhang, Yaxiong Wang, Kecheng Han, Yujiao Wu, Lianwei Wu, Li Zhu, Zhedong Zheng

Fair-CCD: Mitigating Bias in Large Language Models for Tabular Classification Through Context-Contrastive Decoding (2026.acl-long)

Challenge:	Existing methods for manipulation detection and grounding focus on manipulator type classification under result-oriented supervision.
Approach:	They propose a reasoning-driven framework that shifts learning from outcome fitting to process modeling.
Outcome:	The proposed framework achieves state-of-the-art with superior generalization on large-scale datasets.

Copied to clipboard

Donghan Liu, Han Sun, Zhaohui Wang, Qin Li, Min Zhang

DVI-DTM: Dual-View Representation Learning for Interpretable Short Text Dynamic Topic Modeling (2026.acl-long)

Challenge:	Prior work to mitigate fairness issues often employs subjective demonstration selection, leading to low controllability and limited stability across different models and tasks.
Approach:	They propose to use in-context learning to insert social biases into large language models to create a structured and controllable representation of the relationship between sensitive attributes and predicted labels.
Outcome:	Extensive experiments show that Fair-CCD consistently improves fairness metrics without degrading task accuracy.

Copied to clipboard

Di Liu, Zheng Fang, Bin Wu

GenesisFunc: Multi-Agent Data Generation for Accurate and Generalizable Function-Calling (2026.acl-long)

Challenge:	Existing dynamic topic modeling methods face semantic ambiguity and interpretation ambiguities when applied to short texts.
Approach:	They propose a Dual-View representation learning-based Interpretable short text Dynamic Topic Model to address semantic ambiguity and interpretation ambiguities.
Outcome:	The proposed model outperforms the state-of-the-art models in topic alignment and dynamic topic quality metrics while producing highly interpretable topic descriptions.

Copied to clipboard

Hao-Xiang Xu, Chong Deng, Jiaqing Liu, Wen Wang, Qian Chen, Lujia Bao, Xiangang Li, Zhen-Hua Ling

Challenge:	Large Language Models (LLMs) extend their capabilities through function-calling (FC) however, obtaining and annotating real function-called data is challenging, and synthetic data from existing pipelines suffers from unreliable APIs, limited tool scalability, insufficient diversity, and weak quality control.
Approach:	They propose a pipeline for generating FC training data using reliable tools and a multi-agent framework that supports a dialogue generation system that produces conversations spanning diverse scenarios.
Outcome:	The proposed pipeline outperforms open-source models in in-domain FC performance and out-of-domain generalization while reaching FC capabilities comparable to some of the latest API-based models.

Selective Test-Time Debiasing for CLIP via Reward Gating (2026.acl-long)

Copied to clipboard

Jaeho Han, Jisoo Yang, Hyeondong Woo, Mingyu Jeon, Sunjae Yoon, Junyeong Kim

CURE: Critique-Driven Unified Reinforcement Learning for Test-Time Self-Improvement (2026.acl-long)

Challenge:	Existing methods for debiasing use uniform bias corrections across all input queries . weak debiases retains bias in sensitive queries, while weak dealiases in biased ones .
Approach:	They propose a framework that selectively applies debiasing based on input sensitivity . RG-TTA adaptively triggers fairness regularization based upon bias sensitivity of each input .
Outcome:	Experiments show that debiasing improves zero-shot performance while maintaining fairness . weak debiased queries distort semantically meaningful information while weak ones fail to mitigate stereotypes .

Copied to clipboard

Guirong Chen, Shuqi Ye, Wenkai Yang, Shiqi Shen, Guangyao Shen, Yankai Lin

Challenge:	Existing critique-guided methods fail to equip models with the autonomous improvement capabilities required for test-time scaling.
Approach:	They propose a framework that jointly optimizes a single policy for standard solving, critiquing, and guided re-exploration.
Outcome:	The proposed framework maintains competitive single-turn performance and unlocks effective inference-time scaling.

LGSA: Label Geometry Structuring and Aligning for Hierarchical Text Classification (2026.acl-long)

Copied to clipboard

Shuai Zhang, Weibo Xu, Jiahao Nie, Kecheng Huang

FactVerse: A Benchmark for Factual Consistency in Interleaved Image–Text Generation (2026.acl-long)

Challenge:	Existing hierarchical text classification methods use prompt tuning or contrastive learning to implicitly learn label embeddings for classification, but this method fails to model hierarchy-aware geometric relations among labels.
Approach:	They propose a two-stage framework that transforms the label hierarchy from an implicit prior into an explicit embedding by using a general orthogonal frame.
Outcome:	The proposed framework outperforms existing state-of-the-art methods on three real-world HTC datasets.

Copied to clipboard

Yubo Shan, Kun Zhang, Qiming Xu, Liping Cao, Yingying Cao, Jian Zhang, Yu Wang, Jingyuan Li, Yuanzhuo Wang

GMSA: Enhancing Context Compression via Group Merging and Layer Semantic Alignment (2026.acl-long)

Challenge:	Existing benchmarks lack effective mechanisms to evaluate factual consistency in interleaved image-text generation.
Approach:	They propose a benchmark dedicated to evaluating factual consistency in interleaved image-text generation.
Outcome:	The proposed framework outperforms existing evaluation methods in evaluating factual consistency in interleaved image-text generation.

Copied to clipboard

Jiwei Tang, Zhicheng Zhang, Shunlong Wu, Jingheng Ye, Lichen Bai, Zitai Wang, Tingwei Lu, Lin Hai, Yiming Zhao, Hai-Tao Zheng, Hong-Gee Kim

SpidR-Adapt: A Universal Speech Representation Model for Few-Shot Adaptation (2026.acl-long)

Challenge:	Large Language Models (LLMs) have achieved remarkable performance across NLP tasks . however, in long-context scenarios, they face high computational cost and information redundancy.
Approach:	They propose an encoder-decoder context compression framework that generates a compact sequence of soft tokens for downstream tasks.
Outcome:	Experiments show that GMSA outperforms baselines on multiple long-context question answering and summarization benchmarks while maintaining low end-to-end latency.

Copied to clipboard

Mahi Luthra, Jiayi Shen, Maxime Poli, Angelo Ortiz Tandazo, Yosuke Higuchi, Youssef Benchekroun, Martin Gleize, Charles-Éric Saint-James, Dongyan Lin, Phillip Rust, Angel Villar-Corrales, null Surya, Vanessa Stark, Rashel Moritz, Juan Pino, Yann LeCun, Emmanuel Dupoux

EIFFEL: a novel benchmark to measure bias of English heavy training on French idiomatic expressions (2026.acl-long)

Challenge:	Empirically, SpidR-Adapt achieves rapid gains in phonemic discriminability and downstream spoken language modeling scores . current self-supervised learning models require thousands of hours of training data to learn meaningful linguistic representations.
Approach:	They propose a bi-level optimization framework for rapid adaptation of speech units to new languages using minimal unlabeled data.
Outcome:	The proposed model achieves rapid gains in phonemic discriminability and spoken language modeling scores . it surpasses in-domain toplines after training on less than 1h of target-language audio .

Copied to clipboard

Charlotte Noel, Nicholas Asher, Olivier Gouvert, Farah Benamara, Julie Hunter

Latent Attention Denoising: A Training-Free Energy-Based Framework for Mitigating Hallucinations in Vision-Language Models (2026.acl-long)

Challenge:	Mainstream multilingual models are generally trained on a much higher proportion of English data . this raises questions about their ability to capture linguistic features specific to non-English languages .
Approach:	They propose a benchmark to test multilingual LLMs' ability to capture linguistic features in other languages.
Outcome:	The proposed benchmark shows that multilingual models can capture features in non-English languages and cultural norms.

Copied to clipboard

Zhiwen Luo, Siyu Jiang, Weilong Jiang, Kun He

From Knowing to Teaching: Scaffolding Pedagogical Decisions for LLM Agent (2026.acl-long)

Challenge:	Existing models for large vision-Language Models lack reliable visual hallucination . despite advances in multimodal perception, they are still limited in real-world applications .
Approach:	They propose a framework that recasts attention calibration as a one-step score-based denoising process.
Outcome:	The proposed framework achieves superior performance on generative and discriminative tasks while maintaining efficiency comparable to standard decoding.

Copied to clipboard

Yucheng Wang, Shen Yang, Jifan Yu, Haoxuan Li, Joy Jia Yin Lim, Daniel Zhang-Li, Huiqin Liu, Lei Hou, Juanzi Li, Bin Xu

Learn Like Humans: Use Meta-cognitive Reflection for Efficient Self-Improvement (2026.acl-long)

Challenge:	Large language models produce content lacking pedagogical depth when asked to generate lessons .
Approach:	They propose a framework that allows teachers to select content according to pedagogical intent and sequence topics so foundations precede applications.
Outcome:	The framework achieves 67.8% win rate in human evaluation and 79.6% in LLM-based evaluation against eight baselines.

Copied to clipboard

Xinmeng Hou, Bohao Qu, Wuqi Wang, Peiliang Gong, Qing Guo, Yang Liu

Think Faster Than Words: Efficient LLM Chain-of-Thought Reasoning via Dynamic Shortcut Decoding (2026.acl-long)

Challenge:	Existing self-improving frameworks rely on inefficient, multi-turn recursive loops that incur high computational costs.
Approach:	They propose a framework that achieves efficient self-evolution within a single recurrence cycle.
Outcome:	The proposed framework outperforms state-of-the-art self-evolving systems while significantly reducing computational overhead.

Copied to clipboard

Fan Liu, Yanhao Wang, Min Zhang, Zhikang Chen, Zeyuan Li, Lewei He, Jiahui Pan

AEA: Adaptive Expert Allocation Improves Sentence Embeddings from Mixture-of-Experts LLM (2026.acl-long)

Challenge:	Existing methods that prune or employ early stopping to reduce latency often compromise reasoning reliability.
Approach:	They propose a shortcut decoding framework that integrates probes over internal hidden states with step-level entropy to detect convergence of reasoning during generation and adaptively selects between a fast-exit path and a stability-verified path to remove redundant steps while preserving answer correctness.
Outcome:	The proposed framework reduces token usage by approximately 35% and maintains accuracy comparable to full CoT decoding.

Copied to clipboard

Shufan Yang, Zifeng Cheng, Zhiwei Jiang, Qingfeng Qi, Yafeng Yin, Cong Wang, Ao Zhou, Qing Gu

It’s High Time: A Survey of Temporal Question Answering (2026.acl-long)

Challenge:	Existing methods to improve embeddings from Mixture-of-Experts models allocate a fixed number of experts uniformly across all layers and tokens, ignoring inter-layer and inter-token heterogeneity.
Approach:	They propose an Adaptive Expert Allocation framework that performs layer-wise and token-wise expert allocation to enhance embedding quality.
Outcome:	The proposed method improves embedding quality across multiple MoE models.

Copied to clipboard

Bhawna Piryani, Abdelrahman Abdallah, Jamshid Mozafari, Avishek Anand, Adam Jatowt

SDAR-VL: Stable and Efficient Block-wise Diffusion for Vision-Language Understanding (2026.acl-long)

Challenge:	Temporal Question Answering (TQA) is a research area that focuses on answering questions involving temporal constraints or context.
Approach:	They present a comprehensive overview of Temporal Question Answering (TQA) this research area focuses on answering questions involving temporal constraints or context .
Outcome:	The proposed frameworks are compared against a range of datasets, tasks, and approaches.

Copied to clipboard

Shuang Cheng, Yuhua Jiang, Zineng Zhou, Dawei Liu, Tao Wang, Linfeng Zhang, Biqing Qi, Bowen Zhou

N-GLARE: An Non-Generative Latent Representation-Efficient LLM Safety Evaluator (2026.acl-long)

Challenge:	Existing block-wise discrete diffusion models lack robust autoregressive (AR) decoders.
Approach:	They propose a block-wise discrete diffusion framework for large-scale vision-language understanding with a progressive beta noise curriculum.
Outcome:	The proposed framework improves training efficiency, convergence stability, and task performance over conventional block diffusion.

Copied to clipboard

Zheyu Lin, Jirui Yang, Yukui Qiu, Yubing Bao, Hengqi Guo, Yao Guan

Don’t Act Blindly: Robust GUI Automation via Action-Effect Verification and Self-Correction (2026.acl-long)

Challenge:	Evaluating the safety robustness of LLMs is critical for their deployment.
Approach:	They propose to use latent representations to characterize hidden layer dynamics by analyzing the APT of latent models and introducing the JSS metric.
Outcome:	The proposed method exploits the APT (Angular-Probabilistic Trajectory) of latent representations and introduces the JSS (Jensen-Shannon Separability) metric.

Copied to clipboard

Yuzhe Zhang, Xianwei Xue, Xingyong Wu, Mengke Chen, Chen Liu, Xinran He, Run Shao, Feiran Liu, Huanmin Xu, Qiutong Pan, Haiwei Wang

Beyond Static Persona Consistency: Dynamic Persona Coherence in LLM Role-Playing (2026.acl-long)

Challenge:	Existing GUI agents assume deterministic environment responses, generating actions without verifying whether previous operations succeeded.
Approach:	They propose a GUI agent that explicitly models action outcomes and recovery under noisy environments.
Outcome:	The proposed agent reduces failure loops and improves recovery success in noisy environments while maintaining competitive standard task performance.

Copied to clipboard

Yirui QI, Xiaoming Zhang, Ruilin Zeng, Mengyao Liu, Ziyi Zhou, Dezhuang Miao, Bingyu Yan, Zhenyu Guan

Challenge:	Existing LLMs conflate identity consistency with emotional rigidity . Existing models exhibit either robotic repetition or persona drift .
Approach:	They propose a framework that decouples Identity-Layer Stability from Adaptive-Layer Appropriateness to achieve persona coherence repair.
Outcome:	Experiments on GPT-4o, Claude-3.5-Sonnet, and DeepSeek-V3.2 show consistent improvements (+16–84% gains)

What If Consensus Lies? Selective-Complementary Reinforcement Learning at Test Time (2026.acl-long)

Copied to clipboard

Dong Yan, Jian Liang, Yanbo Wang, Shuo Lu, Ran He, Tieniu Tan

Exploring and Distilling Multi-Dimensional Clues for Interpretable Social Bot Detection (2026.acl-long)

Challenge:	Existing TTRL methods rely on positive pseudo-labeling strategies to enhance reasoning capabilities.
Approach:	They propose a test-time reinforcement learning framework that mitigates label noise amplification by deriving pseudo-rewards from majority voting consensus.
Outcome:	The proposed framework mitigates label noise amplification by implementing selective positive pseudo-labeling and entropy-gated negative p-labeled pruning.

Copied to clipboard

Yi Han, Haiqi Lu, Lizi Liao, Shuhan Zhou, Yuanxing Liu, Weinan Zhang, Ting Liu

Why Are We Moral? An LLM-based Agent Simulation Approach to the Study of Moral Evolution (2026.acl-long)

Challenge:	Existing research on social bot detection results directly without corresponding supportive explanations, making it difficult to assess the extent to which such predictions are trustworthy.
Approach:	They propose a four-dimensional clue framework that uses outcome-reward reinforcement learning to train inspectors to generate faithful, grounded clues from user information, semantic features, interactive situation, and behavioral pattern.
Outcome:	The proposed framework outperforms baselines in detection performance and significantly improves the performance of large language models.

Copied to clipboard

Zhou Ziheng, Huacong Tang, Mingjie Bi, Wanying He, Fang Sun, Yizhou Sun, Ying Nian Wu, Demetri Terzopoulos, Yipeng Kang, Fangwei Zhong

Systematicity between Forms and Meanings across Languages Supports Efficient Communication (2026.acl-long)

Challenge:	Existing models of moral evolution must abstract away cognitive processes . et al. (2017): evolution of morality presents a puzzle: natural selection favors selfish .
Approach:	They propose an LLM-based agent simulation framework that manipulates cognitive factors to understand moral evolution.
Outcome:	The proposed model exploits cognitive realism to explore moral evolution in a hunter-gatherer society.

Copied to clipboard

Doreen Osmelak, Yang Xu, Michael Hahn, Kate McCurdy

On the (In-)Security of the Shuffling Defense in the Transformer Secure Inference (2026.acl-long)

Challenge:	Languages vary in how meanings map to word forms, but this theory does not account for systematic relations within word forms.
Approach:	They propose a model that measures the learnability of meaning-to-form mappings by inverse of simplicity.
Outcome:	The proposed model captures fine-grained regularities in linguistic form, allowing better discrimination between attested and unattested systems.

Copied to clipboard

Zhengyi Li, Yakai Wang, Jingwen Leng, Kang Yang, Yu Yu, Jiaping Gui, Yu Feng, Ning Liu, Minyi Guo

LaCo: Layer-wise Compensation for Pruned Large Language Models (2026.acl-long)

Challenge:	Existing work reveals only randomly permuted activations to the client, allowing adversaries to extract model weights.
Approach:	They propose an attack that aligns differently shuffled activations to a common permutation and exploits them to extract model weights.
Outcome:	The proposed attack can align shuffled activations to a common permutation and exploit them to extract model weights with a query cost of approximately $1.

Copied to clipboard

Yingen Liu, Fan Wu, null Panxuyan, Ruihui Li, Zhuo Tang, Kenli Li

QiMeng-PRepair: Precise Code Repair via Edit-Aware Reward Optimization (2026.acl-long)

Challenge:	Existing methods for predicting performance degradations of Large Language Models (LLMs) neglect the structural distortions caused by sparsity.
Approach:	They propose a framework that reorients the recovery paradigm from global adaptation to hierarchical representation alignment by sequentially optimizing each layer to reconstruct the model's hidden states.
Outcome:	The proposed framework surpasses parameter-efficient baselines in perplexity reduction and zero-shot reasoning.

Copied to clipboard

Changxin Ke, Rui Zhang, Jiaming Guo, Yuanbo Wen, Li Ding, Shuo Wang, Xuyuan Zhu, Xiong Peng, Di Huang, Zidong Du, Xing Hu, Qi Guo, Yunji Chen

Location Not Found: Exposing Implicit Local and Global Biases in Multilingual LLMs (2026.acl-long)

Challenge:	Existing approaches to program repair are based on correctness alone.
Approach:	They propose a framework that mitigates over-editing and improves repair accuracy by generating buggy programs and re-edits.
Outcome:	The proposed framework improves repair precision by 31.4% under fix1@1, a metric that considers repair correctness and extent, and significantly increases decoding throughput when combined with speculative editing.

Copied to clipboard

Guy Mor-Lan, Omer Goldman, Matan Eyal, Adi Mayrav Gilady, Sivan Eiger, Idan Szpektor, Avinatan Hassidim, Yossi Matias, Reut Tsarfaty

Incorporating Temporal Coherence to Cross-Document Event Coreference Resolution (2026.acl-long)

Challenge:	Multilingual large language models have minimized the fluency gap between languages, but they are exposed to the risk of biases as knowledge and norms may propagate across languages.
Approach:	They propose a test set with 2,156 questions in 12 languages to quantify models' biases . they show a global bias towards answers relevant to the US-locale .
Outcome:	The proposed model can answer locale-ambiguous questions in 12 languages.

Copied to clipboard

Xinyu Chen, Peifeng Li, Qiaoming Zhu

LLMs in Sarcasm Detection? It’s elementary! (Or is it?) (2026.acl-long)

Challenge:	Existing approaches focus on enhancing semantic coherence between event mentions, but they overlook the critical aspect of temporal coherency.
Approach:	They propose a Temporal Cohorence-driven event coreference framework that explicitly models temporal constraints by constructing a temporal event graph and a GNN to resolve conflicts.
Outcome:	Experiments on the ECB+, GVC, WEC, and ECb+META datasets show that CohTP outperforms state-of-the-art methods.

Copied to clipboard

Priyanshu Mahato, Aniket Santosh Mishra, Kripabandhu Ghosh

Beyond Itinerary Planning—A Real-World Benchmark for Multi-Turn and Tool-Using Travel Tasks (2026.acl-long)

Challenge:	Large Language Models (LLMs) are often cited for their sophisticated pragmatic reasoning, but they collapse to random guessing on organic human speech.
Approach:	They propose that LLMs have near-human competence in sarcasm detection . authors propose that this proficiency may be deceptive .
Outcome:	The proposed model performance on synthetic leaderboards is a statistical mirage of competence.

Copied to clipboard

Xiang Cheng, Yulan Hu, Xiangwen Zhang, Lu Xu, Lide Tan, Zheng Pan, Xin Li, Yong Liu

TLoRA: Task-aware Low Rank Adaptation of Large Language Models (2026.acl-long)

Challenge:	Existing studies on LLM performance on travel planning have shown that existing settings are limited due to limited domain coverage, insufficient modeling of users’ implicit preferences in multi-turn conversations, and a lack of evaluation of agents’ capability boundaries.
Approach:	They propose a benchmark to evaluate LLMs' planning and tool-use abilities in real-world settings by collecting user queries, user preferences, and tools from real scenarios.
Outcome:	The proposed benchmark evaluates agents' capabilities in real-world settings and shows that even advanced models exhibit imbalanced performance across different capabilities.

Copied to clipboard

Weicheng Lin, Yi Zhang, Jiawei Dang, Liang-Jie Zhang

MARS-RA: Rank Aggregation for Credit Assignment via Multimodal Comparisons in Embodied Multi-Agent Cooperation (2026.acl-long)

Challenge:	Existing low-rank Adaptation (LoRA) methods address only one factor, often at the cost of increased training complexity or reduced practical efficiency.
Approach:	They propose a low-rank Adaptation framework that optimizes initialization and resource allocation at the outset of training.
Outcome:	The proposed framework performs excellently across various tasks while reducing the number of trainable parameters.

Copied to clipboard

Dawei Wang, Di Zhao, Xinyuan Liu, Marci Chi Ma, Xiaoyang Liu, Chengming Zhou, Gary Ushaw, Richard Davison

Persona-E²: A Human-Grounded Dataset for Personality-Shaped Emotional Responses to Textual Events (2026.acl-long)

Challenge:	Embodied AI systems are open, where agents may leave or enter mid-task due to hardware failures or task-related errors.
Approach:	They propose a framework that reformulates credit assignment as a rank aggregation problem using contribution-based pairwise comparisons among agents generated by large multimodal models.
Outcome:	The proposed framework can guide agents toward effective cooperation in complex tasks of different types.

Copied to clipboard

Yuqin Yang, Haowu Zhou, Haoran Tu, Zhiwen Hui, Shiqi Yan, HaoYang Li, Dong She, Xianrong Yao, Yang Gao, Zhanpeng Jin

Challenge:	A critical bottleneck is the lack of ground-truth human data to link personality traits to emotional shifts.
Approach:	They propose a large-scale dataset to capture reader-based emotional variations across news, social media, and life narratives.
Outcome:	The proposed model captures reader-based emotional variations across news, social media, and life narratives.

Interleaved Latent Visual Reasoning with Selective Perceptual Modeling (2026.acl-long)

Copied to clipboard

Shuai Dong, Siyuan Wang, Xingyu Liu, Chenglin Li, Haowen Hou, Zhongyu Wei

Think before Go: Hierarchical Reasoning for Image-goal Navigation (2026.acl-long)

Challenge:	Existing approaches to interleaved reasoning are limited by the cost of re-encoding pixel-dense images.
Approach:	They propose a framework that unifies dynamic state evolution with precise perceptual modeling.
Outcome:	The proposed framework outperforms existing approaches on multimodal reasoning benchmarks.

Copied to clipboard

Pengna Li, Kangyi Wu, Shaoqing Xu, Fang Li, Lin Zhao, Long Chen, Zhi-Xin Yang, Nanning Zheng

LLMs as Knowledge Graph Refiners: Mitigating Factual Inconsistencies in Generative Knowledge Extraction (2026.acl-long)

Challenge:	Existing methods for image-goal navigation fail to extract informative visual cues, leading agents to wander around.
Approach:	They propose a framework that decomposes image-goal navigation into high-level planning and low-level execution.
Outcome:	The proposed method is superior to existing methods in both simulation and real-world environments.

Copied to clipboard

Donghyun Kim, Hyeongjun Yang, Seokju Hwang, Kyong-Ho Lee, Chanhee Lee

SILO-BENCH: A Scalable Environment for Evaluating Distributed Coordination in Multi-Agent LLM Systems (2026.acl-long)

Challenge:	Knowledge graphs (KGs) represent real-world entities and their relations in a structured form.
Approach:	They propose a framework that performs triple-level refinement on KGs constructed via GKE.
Outcome:	The proposed framework improves KG quality from diverse perspectives.

Copied to clipboard

Yuzhe Zhang, Feiran Liu, Yi Shan, Xinyi Huang, Xin Yang, Yueqi Zhu, Xuxin Cheng, Cao Liu, Ke Zeng, Terry Jingchen Zhang, Wenyuan Jiang

Challenge:	Existing benchmarks conflate coordination ability with role-based priors.
Approach:	They propose a role-free benchmark for evaluating free-form collaboration under information silos.
Outcome:	The proposed benchmark systematically probes coordination capabilities under information silos using 54 configurations and 3 frontier LLMs.

Mitigating Structural Knowledge Collapse in Domain-Specific LLMs via Morpheme-Aware KV-Aggregation (2026.acl-long)

Copied to clipboard

Yuxuan Si, Zheqi Lv, Chengxi Zang, Zhengyu Chen, Fei Wu

Zero-shot Jianzi Recognition as Structured Visual Information Extraction in Open Compositional Symbolic Systems (2026.acl-long)

Challenge:	Existing tokenizers over-fragment domain terms, disrupting morpheme semantics.
Approach:	They propose a lightweight tokenizer that dynamically consolidates fragments without tokenizer changes.
Outcome:	The proposed adapter outperforms vocabulary adaptation baselines on medical and legal terms by 3.2–4.6% and 7.9% on high-fragmentation terms.

Copied to clipboard

Zehan Li, Fu Zhang, Zhijun Liu, Jingwei Cheng

Whose Facts Win? LLM Source Preferences under Knowledge Conflicts (2026.acl-long)

Challenge:	Optical character recognition (OCR) is a relatively new form of tablature recognition, but its accuracy is limited due to its unbounded composition and manuscript-level variability.
Approach:	They propose a method that predicts component sequences under a zero-shot split and synthesize manuscript-like training images via component-wise style recomposition and manuscript-domain noise modeling.
Outcome:	The proposed method achieves 63.02% sequence accuracy on real-world Jianzi benchmark, surpassing Gemini-3-Pro by 35.11%.

Copied to clipboard

Jakob Schuster, Vagrant Gautam, Katja Markert

CLARity: Reasoning Consistency Alone Can Teach Reinforced Experts (2026.acl-long)

Challenge:	Existing studies on the role of the source of knowledge conflicts have not investigated the role .
Approach:	They propose a framework that reduces repetition bias by up to 79.2% while maintaining at least 72.5% of original preferences.
Outcome:	The proposed method reduces repetition bias by up to 79.2% while maintaining at least 72.5% of original preferences.

Copied to clipboard

Jiuheng Lin, Cong Jiang, Zirui Wu, Jiarui Sun, Yansong Feng

EGSS: Entropy-guided Stepwise Scaling for Reliable Software Engineering (2026.acl-long)

Challenge:	Existing solutions to supervise the reasoning process are prohibitively expensive.
Approach:	They propose a cost-effective reinforcement learning framework that enhances reasoning quality using a small, general-purpose LLM only.
Outcome:	Experiments show that CLARity improves reasoning quality by 16.5% over standard outcome-based reinforcement learning (RL) human evaluations confirm substantial gains in factual correctness and reasoning coherence, leading to more trustworthy model outputs.

Copied to clipboard

Chenhui Mao, Yuanting Lei, Zhixiang Wei, Ming Liang, Zhixiang Wang, Jingxuan Xu, Dajun Chen, Wei Jiang, Yong Li

Detecting What Queries Seek: Steering LLM Safety with FFN Output Activation Monitoring (2026.acl-long)

Challenge:	Entropy-Guided Stepwise Scaling (EGSS) is a novel TTS framework for software engineering tasks.
Approach:	They propose an entropy-guided stepwise scaling framework that balances efficiency and effectiveness through entropic-guide encoding and robust test-suite augmentation.
Outcome:	EGSS boosts performance by 5–10% across all evaluated models, and reduces inference-time token usage by over 28% . compared to existing methods, EGS reduces token usage and reduce inference time by over 20% .

Copied to clipboard

Xiaohao Luo, Ying Wei, Rui Zhao

Temporal Evidence Chain for Temporal Knowledge Graph Question Answering with Large Language Models (2026.acl-long)

Challenge:	Existing methods for detecting malicious queries rely on residual stream activations, resulting in limited discriminative power and unreliable interventions.
Approach:	They propose to use feed-forward networks (FFNs) to generate more discriminative signals for intervention, since these activations more explicitly reflect the intent of a query.
Outcome:	Experiments show that the proposed approach achieves state-of-the-art defense performance against various jailbreak attacks while maintaining the model's original performance on benign tasks.

Copied to clipboard

Shihao Liu, Xiaofei Zhou, Bo Wang, Geyuan Zhang

Modal Dependency Parsing as Structured Prediction over Source-Cue Scope (2026.acl-long)

Challenge:	Temporal Knowledge Graph Question Answering (TKGQA) aims to answer temporal questions using knowledge from Temporal knowledge graphs.
Approach:	They propose a framework to construct temporal evidence chains for LLM reasoning using Temporal Knowledge Graphs.
Outcome:	TECQA outperforms existing methods on MultiTQ and CronQuestions.

Copied to clipboard

Jayeol Chun, Nianwen Xue

DEFT: Demystifying VLN Failures via a Unified Dual-View Explainability Framework for LLM-based Agents (2026.acl-long)

Challenge:	Existing work on identifying sources only focuses on defining source-introducing cues . a structured model focuses learning at the source-cue level and constrains event-level decisions to a small, scope-defined candidate set.
Approach:	They propose a framework that leverages large language models to explicitly identify source-cue pairs and their respective scope to define modal contexts.
Outcome:	The proposed framework surpasses state-of-the-art results by 3 and 4% for English and Chinese datasets.

Copied to clipboard

Yawen Wang, Yihan Dai, Jianming Chen, Junjie Wang, Qing Wang

GLIER: Generative Legal Inference and Evidence Ranking for Legal Case Retrieval (2026.acl-long)

Challenge:	Existing interpretability methods isolate temporal criticality from feature salience, creating an alignment gap and failing to account for the behavioral instability of black-box agents.
Approach:	They propose a unified dual-view framework that jointly analyzes when a decision is pivotal and what visual evidence grounds it.
Outcome:	Extensive experiments on MatterPort3D show that DEFT outperforms baselines in both temporal and feature fidelity.

Copied to clipboard

Minghan Li, Tianrui Lv, Chao Zhang, Guodong Zhou

LLMs (Almost) Never Abstain Under Medical Uncertainty (2026.acl-long)

Challenge:	Existing dense retrieval methods neglect the explicit legal logic that underpins legal relevance.
Approach:	They propose a framework that reformulates retrieval as an inference process over latent legal variables.
Outcome:	GLIER outperforms strong baselines like SAILER and KELLER in a legal case-based retrieval task . the framework exhibits exceptional data efficiency even when trained with only 10% of the data .

Copied to clipboard

Alessio Cocchieri, Luca Ragazzi, Giuseppe Tagliavini, Gianluca Moro

Challenge:	Medical multiple-choice question answering (MCQA) benchmarks assume large language models should always commit to an answer.
Approach:	They propose a benchmark to evaluate medical abstention under uncertainty . they remove the gold answer and introduce an explicit "I abstain" option . results highlight abstinence as a critical but overlooked dimension of medical decision-making evaluation .
Outcome:	The new benchmark evaluates medical abstention under uncertainty.

TOWER+: Bridging Generality and Translation Specialization in Multilingual LLMs (2026.acl-long)

Copied to clipboard

Ricardo Rei, Nuno M Guerreiro, José Pombal, João Alves, Amin Farajian, Pedro Teixeirinha, Andre Martins

Culture-Aware Machine Translation in Large Language Models: Benchmarking and Investigation (2026.acl-long)

Challenge:	Large Language Models (LLMs) are emerging as the de facto solution for multilingual machine translation.
Approach:	They propose a suite of LLMs that can be fine-tuned to deliver strong performance on translation and multilingual general-purpose text capabilities.
Outcome:	The proposed models outperform existing models on translation and general-purpose tasks.

Copied to clipboard

Zekun Yuan, Yangfan Ye, Xiaocheng Feng, Baohang Li, Qichen Hong, Yunfei Lu, Dandan Tu, Bing Qin

Learning from Near-Misses: Error-Aware Contrastive Few-Shot Learning for NL2Formula (2026.acl-long)

Challenge:	Existing studies on large language models focus on literal-level translation quality, such as adequacy and fluency.
Approach:	They propose a Culture-Aware Novel-Driven Parallel Dataset for Machine Translation and a multi-dimensional evaluation framework for assessing cultural translation quality.
Outcome:	The proposed model improves evaluation reliability in LLM-as-a-judge scenarios under culture-aware constraints.

Copied to clipboard

Zhihao Shuai, Yiyun Chen, Maolin Ma, Yutong Chen, Hanjia Qiu, Jing Xu, Ziye Chen, Weikai Yang

Calibrated Speculative Decoding: Frequency-Guided Candidate Selection for Efficient Inference (2026.acl-long)

Challenge:	Existing spreadsheet formulas often produce near-miss outputs due to an incorrect function, operator, or reference.
Approach:	They propose an abstract syntax tree-based error taxonomy that organizes common error modes by the kind of decision that goes wrong in the parse tree.
Outcome:	The proposed framework improves Exact Match (EM) by 6.4 points over supervised fine-tuning and matches self-consistency (SC@5) accuracy.

Copied to clipboard

null Zhouxuwen, Fangxin Liu, Chao Wang, Xiao Zheng, Hao Zheng, Min He, Li Jiang, Haibing Guan

Learning from Mistakes: Negative Reasoning Samples Enhance Out-of-Domain Generalization (2026.acl-long)

Challenge:	Speculative decoding (SD) is a powerful and efficient way to accelerate autoregressive generation.
Approach:	They propose a training-free framework that recovers valid tokens discarded by standard verification . they use online correction memory and Semantic Consistency Gating to analyze rejections .
Outcome:	The proposed framework outperforms existing methods and achieves peak throughput speedup of 2.33x.

Copied to clipboard

Tian Xueyun, MingHua Ma, Bingbing Xu, Nuoyan Lyu, Wei Li, Heng Dong, Zheng Chu, Yuanzhuo Wang, Huawei Shen

LongTutor: Benchmarking Large Language Models for Long-term Personalized Tutoring (2026.acl-long)

Challenge:	Recent studies show that supervised fine-tuning (SFT) is a common approach for reasoning in large language models.
Approach:	They propose to use supervised fine-tuning (SFT) on chain-of-thought trajectories demonstrations . they find that incorporating negative traxories yields substantial OOD generalization gains .
Outcome:	The proposed scheme yields 5.51% OOD gain over positive-only training.

Copied to clipboard

Ning Li, Zheng Zhang, Zhenya Huang, Rui Li, Yi Zhan, Yinbo Luo, Qi Liu, Enhong Chen

HiChunk: Evaluating and Enhancing Retrieval Augmented Generation with Hierarchical Chunking (2026.acl-long)

Challenge:	Existing evaluations focus on isolated, short-term interactions, overlooking the inherently long-term nature of learning.
Approach:	They propose a benchmark for long-term personalized tutoring based on an annotated learning log . they propose an automated generator–verifier pipeline to enable benchmark expansion .
Outcome:	The proposed benchmarks evaluate LLMs across three progressive tasks: evidence acquisition, knowledge state diagnosis, and adaptive teaching action.

Copied to clipboard

Wensheng Lu, Keyu Chen, Zhifeng Shen, Ruizhi Qiao, Xing Sun

How Good is Your Wikipedia? Auditing Data Quality for Low-resource and Multilingual NLP (2026.acl-long)

Challenge:	Existing evaluation benchmarks for document chunking are inadequate due to evidence sparsity . evaluators are unable to evaluate different chunking methods due to the evidence sparing .
Approach:	They propose a QA benchmark for document chunking and a hierarchical document structuring framework for it.
Outcome:	The proposed framework improves document chunking quality within reasonable time consumption.

Copied to clipboard

Kushal Tatariya, Artur Kulmizev, Wessel Poelman, Esther Ploeger, Marcel Bollmann, Johannes Bjerva, Jiaming Luo, Heather Lent, Miryam de Lhoneux

Understanding and Mitigating Political Stance Cross-topic Generalization in Large Language Models (2026.acl-long)

Challenge:	Wikipedia’s perceived high quality and broad language coverage have established it as a fundamental resource in NLP.
Approach:	They propose a data filtering procedure which removes a large percentage of Wikipedia's data and a 4-level quality ranking of the site.
Outcome:	The results show that the proposed filtering procedure outperforms the raw Wikipedia models in three language modelling scenarios.

Copied to clipboard

Jiayi Zhang, Shu Yang, Junchao Wu, Derek F. Wong, Di Wang

MPR-GUI: Benchmarking and Enhancing Multilingual Perception and Reasoning in GUI Agents (2026.acl-long)

Challenge:	Recent studies have focused on the internal representations of large language models and the mechanisms that lead to unintended cross-topic generalization.
Approach:	They propose a method that uses inhibition to localize political neurons and a technique that uses topic-specific blocking to mitigate the cross-topic generalization.
Outcome:	The proposed method reduces cross-topic generalization by 20% while preserving topic-specific performance.

Copied to clipboard

Ruihan Chen, Qiming Li, Xiaocheng Feng, Weihong Zhong, Xiaoliang Yang, Yuxuan Gu, Zekun Zhou, Yunfei Lu, Haoyu Ren, Kun Chen, Dandan Tu, Bing Qin

CoVerRL: Breaking the Consensus Trap in Label-Free Reasoning via Generator-Verifier Co-Evolution (2026.acl-long)

Challenge:	Existing GUI benchmarks lack fine-grained diagnostics to identify which capabilities lead to task failures.
Approach:	They propose a multilingual P R GUI Benchmark to assess LVLMs' language capabilities . they propose XLI to align non-English hidden states with English ones during inference .
Outcome:	The proposed benchmark reveals consistent gaps between English and non-English settings . it reduces the cross-lingual gaps with an average gain of 6.5% in non- English settings compared to static benchmarks .

Copied to clipboard

Teng Pan, Yuchen Yan, Zixuan Wang, Ruiqing Zhang, Guiyang Hou, Wenqi Zhang, Weiming Lu, Jun Xiao, Yongliang Shen

From Weights to Activations: Is Steering the Next Frontier of Adaptation? (2026.acl-long)

Challenge:	Label-free reinforcement learning enables large language models to improve reasoning capabilities . but as training maximizes self-consistency, output diversity collapses, authors say . authors propose a framework where a single model alternates between generator and verifier roles .
Approach:	They propose a framework where a model alternates between generator and verifier roles, bootstrapping each other.
Outcome:	Experiments show that CoVerRL outperforms label-free baselines on reasoning benchmarks . the framework can be used to improve reasoning abilities without ground-truth supervision .

Copied to clipboard

Simon Ostermann, Daniil Gurgurov, Tanja Baeumel, Michael A. Hedderich, Sebastian Lapuschkin, Wojciech Samek, Vera Schmitt

Challenge:	Pre-trained large language models are the basis of a wide range of NLP tasks.
Approach:	They propose to use parameter updates and parameter-efficient adaptation to modify behavior of large language models.
Outcome:	The proposed method enables local and reversible behavioral change without parameter updates.

Can LLMs Act as Historians? Evaluating Historical Research Capabilities of LLMs via the Chinese Imperial Examination (2026.acl-long)

Copied to clipboard

Lirong Gao, Zeqing Wang, Yuyan Cai, Jiayi Deng, Yanmei Gu, Yiming Zhang, Jia Zhou, Yanfei Zhang, Junbo Zhao

Less Languages, Less Tokens: An Efficient Unified Logic Cross-lingual Chain-of-Thought Reasoning Framework (2026.acl-long)

Challenge:	Existing benchmarks assess basic knowledge breadth or lexical understanding, failing to capture higher-order skills that are central to historical research.
Approach:	They propose a benchmark anchored in the Chinese Imperial Examination system that assesses historical knowledge and lexical understanding.
Outcome:	The new benchmark aims to assess the ability of LLMs to process historical materials and documents.

Copied to clipboard

Chenyuan Zhang, Qiguang Chen, Xie Chen, Zhuotao Tian, Bowen Xing, Meishan Zhang, Libo Qin, Baotian Hu, Min Zhang

Embracing Anisotropy: Turning Massive Activations into Interpretable Control Knobs for Large Language Models (2026.acl-long)

Challenge:	Existing methods for cross-lingual chain-of-thought (XCoT) with self-consistency are costly due to extensive sampling of full trajectories across languages.
Approach:	They propose a cross-lingual chain-of-thought framework that minimizes redundancy in token usage and latency.
Outcome:	Experiments on polymath show that UL-XCoT reduces decoding token costs and latency by 50% . UL XCot also aggregates remaining high-quality reasoning paths via voting .

Copied to clipboard

Youngji Roh, Hyunjin Cho, Jaehyung Kim

From Language to Driving: A Dual-Loop SLM-Enhanced Framework for Multi-Planner Scheduling via a Domain-Specific Language (2026.acl-long)

Challenge:	Prior work shows that Large Language Models exhibit highly anisotropic internal representations . prior work shows specialized dimensions capture domain-specific features .
Approach:	They propose a simple magnitude-based criterion to identify Domain-Critical Dimensions in a training-free manner.
Outcome:	The proposed method outperforms whole-dimension steering in domain adaptation and jailbreaking scenarios.

Copied to clipboard

Jiawei Liu, Xun Gong, Muli Yang, Xingrui Yu, Fen Fang, Xulei Yang, Ivor Tsang, Yunfeng hu, Hong Chen, Qing Guo

Challenge:	Recent large language model-based AD research offers new avenues to address this challenge.
Approach:	They propose a small language model (SLM) for high-level semantic reasoning and schedule generation, while an inner loop performs low-level, high-frequency schedule execution and vehicle control.
Outcome:	The proposed framework improves instruction completion rates while maintaining high safety and compliance relative to multiple baselines.

One Cognitive Loop Is Enough: SODA unlocks Pure-Text Spatial Reasoning in Large Language Models (2026.acl-long)

Copied to clipboard

Shunwen Bai, Jiahuan Zhang, Haoran Huang, Yurun Wang, Jiale Liu, Yanxi Wu, Ningzhe Yu, Yudong Gao, Mingjun Cheng

Compatibility-Aware Dynamic Fine-Tuning for Large Language Models (2026.acl-long)

Challenge:	Existing large language models (LLMs) lack visual input, leading to errors in basic numerical comparisons.
Approach:	They propose a spatial OODA framework that integrates the OODAC cognitive loop into multiple control tasks and integrates it into LLMs.
Outcome:	The proposed model significantly improves the spatial reasoning capabilities of large language models across multiple scenarios including SPOD-Bench, SPACE and applications.

Copied to clipboard

Yucheng Zhou, Junwei Sheng, Qianning Wang, Jianbing Shen

Spectral Characterization and Mitigation of Sequential Knowledge Editing Collapse (2026.acl-long)

Challenge:	Recent work attributes optimization instability to the low probability of demonstrations being incompatible with the sample level.
Approach:	They propose a Dynamic Fine-Tuning extension of DFT that controls sample-level optimization variance.
Outcome:	The proposed model can generalize token-level stabilization to the sample level while remaining fully supervised and free of reward modeling.

Copied to clipboard

Chi Zhang, Mengqi Zhang, Xiaotian Ye, Runxi Cheng, Zisheng Zhou, Ying Zhou, Pengjie Ren, Zhumin Chen

Challenge:	Existing approaches to reducing the effects of knowledge editing are insufficiently understood.
Approach:	They propose a plug-and-play framework that preserves the dominant subspace of the original weights and analyzes parameter updates in the spectral basis of the weights.
Outcome:	The proposed framework improves editing efficacy while preserving general abilities under long-horizon sequential editing, including extreme settings with up to 20,000 edits.

SSA: Improving Performance With a Better Scoring Function (2026.acl-long)

Copied to clipboard

Omar Naim, Swarnadeep Bhar, Jerome Bolte, Nicholas Asher

Think Better, Not Longer: Token-Level Marginal Utility for Efficient Reasoning in Large Reasoning Models (2026.acl-long)

Challenge:	Despite the success of in-context learning, recent studies have identified systematic limitations in its generalization behavior.
Approach:	They propose a new attention scoring function that mitigates failures in transformer models . they use Scaled Signed Averaging to train the scoring function instead of Softmax .
Outcome:	The proposed scoring function outperforms transformer models with Softmax on NLP benchmarks and linguistic probing tasks.

Copied to clipboard

Jiawei Li, Yang Gao, Huashan Sun, Chong Feng

AgentSlimming: Towards Efficient and Cost-Aware Multi-Agent Systems (2026.acl-long)

Challenge:	Large reasoning models (LRMs) generate explicit Chain-of-Thought rationales, but often suffer from "overthinking".
Approach:	They propose a unified training framework to synthesize concise reasoning chains by identifying tokens that reduce the model’s likelihood of the correct answer.
Outcome:	Experiments on deepSeek-R1-Distill-Qwen backbones show that MUTO yields better efficiency-accuracy Pareto frontier.

Copied to clipboard

Yulang Chen, Haoxuan Peng, Jinyan Liu, Zichen Wen, Dongrui Liu, Linfeng Zhang

Reusable Experiences: Latent Routing and Modular Composition in LLMs (2026.acl-long)

Challenge:	Automated expansion methods often result in bloated structures with redundant agents, leading to excessive token consumption.
Approach:	They propose a plug-and-play compression framework for graph-structured multi-agent workflows . they estimate the importance score of each agent and remove redundant agents .
Outcome:	Experiments show that AgentSlimming reduces average token cost by 78.9% with negligible performance degradation.

Copied to clipboard

Shuai Ling, Lizi Liao, Dongmei Jiang, Weili Guan

Communicating in Emergent Language with an Induced Morphological Phrasebook (2026.acl-long)

Challenge:	Existing approaches represent accumulated experience as explicit textual artifacts in prompts or implicitly within model weights via fine-tuning. Existing methods are limited by context windows and cannot internalize knowledge.
Approach:	They propose a framework that treats latent experiences as fundamental units for LLM specialization.
Outcome:	Experiments on multi-task NLP benchmarks show that this approach outperforms standard fine-tuning, yielding improved generalization through flexible skill reuse.

Copied to clipboard

Brendon Boldt, David R. Mortensen

Controllable Contamination Detection for Reliable LLM Evaluation with Statistical Guarantees (2026.acl-long)

Challenge:	a major challenge in studying emergent languages is interpreting how they convey meaning-neural networks may invent communication systems lacking features of human language.
Approach:	They build rule-based emergent language agents using form-meaning mappings induced from ELs and test their communicative performance in the EL environment.
Outcome:	The proposed model shows that EL agents rely on repetition and morpheme ordering to convey meaning.

Copied to clipboard

Zheng Zhang, Qi Liu, Siyuan Liang, Ning Li, Zirui Hu, Weibo Gao, Rui Li, Zhenya Huang, Leszek Rutkowski, Baosheng Yu, Dacheng Tao

From 1,000,000 Users to Every User: Scaling Up Personalized Preference for User-level Alignment (2026.acl-long)

Challenge:	Existing training data detectors fail to detect clean samples from contaminated test sets . existing methods fail to identify clean samples due to black-box nature of LLMs .
Approach:	They propose a framework that detects and filters contaminated evaluation data . they propose 'failure detection' to reduce the proportion of contaminated samples mistakenly retained .
Outcome:	The proposed framework reduces false discovery rate (FDR) under valid FDR control while maintaining evaluation consistency.

Copied to clipboard

Jia-Nan Li, Jian Guan, Songhao Wu, Wei Wu, Rui Yan

Beyond Atomic Characters: Glyph-Aware Sub-character Alignment for Low-Resource Multilingual OCR (2026.acl-long)

Challenge:	Current approaches to align large language models assume uniform human preferences, overlooking the diversity inherent in human populations.
Approach:	They propose a framework for scalable personalized alignment of large language models . they establish a preference space characterizing psychological and behavioral dimensions .
Outcome:	The proposed framework improves on existing methods with an average of 17.06% accuracy gain across four benchmarks and a strong adaptation capability to novel preferences.

Copied to clipboard

Mengxiao Zhu, Haixu Chen, Jiu Sha, Jie Liu, Ge Shi

Why Supervised Fine-Tuning Fails to Learn: A Systematic Study of Incomplete Learning in Large Language Models (2026.acl-long)

Challenge:	Low-resource multilingual OCR models struggle with complex script structures and data scarcity.
Approach:	They propose a framework for multilingual character recognition that integrates visual and linguistic backbones with a novel glyph-aware interface.
Outcome:	The proposed framework improves on high-resolution visual and language backbones with glyph-aware interface.

Copied to clipboard

Chao Xue, Yao Wang, Mengqiao Liu, Di Liang, Xingsheng Han, Peiyang Liu, Xianjie Wu, Chenyao Lu, Lei Jiang, Yu Lu, Haibo Shi, Shuang Liang, Minlong Peng, Flora D. Salim

KnowMe-Bench: Benchmarking Person Understanding for Lifelong Digital Companions (2026.acl-long)

Challenge:	Incomplete learning is widespread and heterogeneous in large language models . authors identify five recurrent sources of incomplete learning: missing prerequisite knowledge, conflicts between SFT supervision and pre-training knowledge, internal inconsistencies within SFT data, left-side forgetting during sequential fine-tuning, and insufficient optimization for rare or complex patterns.
Approach:	They propose a diagnostic-first framework that maps incomplete learning to causes . they identify five recurrent sources of incomplete learning: missing prerequisite knowledge, conflicts between supervision and pre-training knowledge, internal inconsistencies, left-side forgetting during sequential fine-tuning, and insufficient optimization for rare or complex patterns.
Outcome:	The proposed framework maps incomplete learning to causes using observable training and inference signals.

Copied to clipboard

Tingyu Wu, Zhisheng Chen, Ziyan Weng, Shuhe Wang, Shuo Zhang, Sen Hu, Silin Wu, Qizhen Lan, Huacan Wang, Ronghao Chen

PersonalityDBench: A Dataset for Personality Disorders - from Modeling to Controlled Generation (2026.acl-long)

Challenge:	Existing long-horizon memory benchmarks use multi-turn dialogues or synthetic user histories . despite rapid progress on long-term memory evaluation, there are gaps in existing benchmarks .
Approach:	They propose a long-form autobiographical narrative benchmark that reconstructs each narrative into a flashback-aware, time-anchored stream and evaluates models with evidence-linked questions.
Outcome:	The proposed benchmarks build from long-form autobiographical narratives . they show that retrieval-augmented systems improve factual accuracy while errors persist on temporally grounded explanations and higher-level inferences.

Copied to clipboard

Federico Ravenda, Seyed Ali Bahrainian, Daniele Montagnani, Antonietta Mira, Andrea Raballo

Challenge:	Personality disorders are chronic, rigid patterns of thinking, behavior, and emotions that deviate from cultural norms and persist in social settings.
Approach:	They propose a large-scale, clinically grounded dataset that supports multidimensional study of personality pathology and standardized evaluation of LLM steering toward clinically ground behavioral targets.
Outcome:	The PersonalityDBench dataset supports multidimensional study of personality pathology and evaluation of LLM steering toward clinically grounded behavioral targets.

Towards Interpretable Tabular Reasoning: Enhancing LLM Reasoning on Tabular Data with Pre-Constructed Logic Graph (2026.acl-long)

Copied to clipboard

Lirong Gao, Zewei Yu, Zhongrui Yin, Qi Zhang, Yuke Zhu, Bo Zheng, Haobo Wang, Junbo Zhao, Gang Chen, Sheng Guo

Beyond Explicit Refusals: Soft-Failure Attacks on Retrieval-Augmented Generation (2026.acl-long)

Challenge:	Tabular data is used in fields such as finance and healthcare due to its heterogeneity and complexity.
Approach:	They propose a Logic-Graph-Enhanced LLM Reasoning framework that integrates the strengths of tree-based models and LLMs to improve their interpretability.
Outcome:	The proposed framework outperforms tree-based models and state-of-the-art LLMs on tabular prediction tasks, achieving superior accuracy and interpretability.

Copied to clipboard

Wentao Zhang, Yan Zhuang, ZhuHang Zheng, Mingfei Zhang, Jiawen Deng, Fuji Ren

Compressing LLM Knowledge into Graph Representations for Text-attributed Graphs Learning (2026.acl-long)

Challenge:	Existing jamming attacks on RAG systems typically induce explicit refusals or denial-of-service behaviors.
Approach:	They propose a black-box attack framework that exploits safety-aligned behaviors of large language models to trigger soft failures.
Outcome:	The proposed framework exploits safety-aligned behaviors of large language models to induce soft failures.

Copied to clipboard

Runhuai Chen, Dian Shen, Dandan Zhang, Kaihong Huang, Linghui Meng, Beilun Wang

End-to-End Optimization of LLM-Driven Multi-Agent Search Systems via Heterogeneous-Group-Based Reinforcement Learning (2026.acl-long)

Challenge:	Existing GNN-LLM approaches use large language models at inference time for processing text attributes, resulting in costly deployment.
Approach:	They propose a framework that internalizes LLM knowledge within graph models and supports inference-efficient TAG learning.
Outcome:	The proposed framework internalizes LLM knowledge within graph models and supports inference-efficient TAG learning.

Copied to clipboard

Guanzhong Chen, Shaoxiong Yang, Chao Li, Wei Liu, Jian Luan, Zenglin Xu

Translation with Thought: Difficulty-Adaptive Reasoning via Reinforcement Learning for Multi-Domain Machine Translation (2026.acl-long)

Challenge:	Existing multi-agent reinforcement learning methods depend on large critic networks to evaluate joint actions, leading to instability and high memory costs.
Approach:	They propose a method to optimize large language models for agent-specific roles . they propose combining agent-based frameworks with retrieval-augmented generation .
Outcome:	Experiments show that multi-agent group policy optimization outperforms baselines in task performance and computational efficiency.

Copied to clipboard

Yongshi Ye, Biao Fu, Chongxuan Huang, Yidong Chen, Xiaodong Shi

Segment, Embed, and Align: A Universal Recipe for Aligning Subtitles to Signing (2026.acl-long)

Challenge:	Multi-domain machine translation (MDMT) is a unique challenge due to varying levels of linguistic complexity across domains.
Approach:	They propose a resource-rational framework that learns to modulate inference between intuitive and deliberate reasoning.
Outcome:	Evaluated on 15 benchmarks spanning in-domain and out-of-domain settings, as well as 3 seen and 59 unseen languages, with ablations across three backbone models, TwT-7B and Twt-14B outperform much larger SOTA reasoning models in translation quality, while reducing token usage by 32–60%.

Copied to clipboard

Zifan Jiang, Youngjoon Jang, Liliane Momeni, Gül Varol, Sarah Ebling, Andrew Zisserman

Understanding Emergent Misalignment via Feature Superposition Geometry (2026.acl-long)

Challenge:	Existing approaches for aligning spoken language text to sign language videos rely on end-to-end training tied to a specific language or dataset.
Approach:	They propose a universal approach for aligning spoken language text with corresponding timestamps to sign language videos using a lightweight dynamic programming procedure.
Outcome:	The proposed method can be used on four sign language datasets and is highly efficient on CPU.

Copied to clipboard

Gouki Minegishi, Hiroki Furuta, Takeshi Kojima, Yusuke Iwasawa, Yutaka Matsuo

MavenCoder: Competitive Code Generation via Model Adaptive Planning Strategies and Multi-Perspective Verification Enhancement (2026.acl-long)

Challenge:	Emergent misalignment is a problem for large language models (LLMs) fine-tuning on narrow tasks can induce harmful behaviors despite no explicit supervision.
Approach:	They propose a mechanistic account based on the geometry of feature superposition . they propose to use sparse autoencoders to identify misalignment-inducing features .
Outcome:	The proposed model outperforms random removal and stronger mitigations than LLM-as-a-judge filtering.

Copied to clipboard

ZhenChun Xu, Yi Cai, Dajun Zheng, li Yuan, Mengchen Zhao, Qixiang Wang, Jiexin Wang

Evidence-Augmented Policy Optimization with Reward Co-Evolution for Long-Context Reasoning (2026.acl-long)

Challenge:	Recent advances in large language models (LLMs) have significantly enhanced automated program synthesis.
Approach:	They propose a model-adaptive and verification–enhanced framework for competition-level code generation that leverages adaptive assessment aligned with the model’s capabilities to select planning strategies while providing timely feedback and correction via multi-perspective verification.
Outcome:	The proposed framework outperforms existing state-of-the-art approaches on livecodebench, humanEval+, MBPP+, and codecontests, and achieves pass@1 results exceeding 3%–40%.

Copied to clipboard

Xin Guan, Zijian Li, Shen Huang, Pengjun Xie, Jingren Zhou, Jiuxin Cao

Generating Attribution Reports for Manipulated Facial Images: A Dataset and Baseline (2026.acl-long)

Challenge:	Evidence-Augmented Policy Optimization (EAPO) improves long-context reasoning performance . Xu et al., 2025): large language models are a critical part of NLP .
Approach:	They propose an Evidence-Augmented Reasoning paradigm that uses a group-relative reward to improve evidence quality.
Outcome:	EAPO significantly improves long-context reasoning performance compared to baselines.

Copied to clipboard

Jingchun Lian, Lingyu Liu, Yaxiong Wang, Yujiao Wu, Lianwei Wu, Li Zhu, Zhedong Zheng

Optimizing RAG Rerankers with LLM Feedback via Reinforcement Learning (2026.acl-long)

Challenge:	Existing facial forgery detection methods focus on binary classification or pixel-level localization, providing little semantic insight into the nature of the manipulation.
Approach:	They propose a multimodal task that localizes forged regions and generates natural language explanations grounded in editing process.
Outcome:	The proposed task localizes forged regions and generates natural language explanations grounded in editing process.

Copied to clipboard

Yuhang Wu, Xiangqing Shen, Fanfan Wang, Cangqi Zhou, Zhen Wu, Xinyu Dai, Rui Xia

Explainable and Fine-Grained Safeguarding of LLM Multi-Agent Systems via Bi-Level Graph Anomaly Detection (2026.acl-long)

Challenge:	Current reranking models are optimized on static human annotations in isolation, decoupled from the downstream generation process.
Approach:	They propose a reinforcement learning framework that directly aligns reranking with LLM's generation quality.
Outcome:	Experiments on knowledge-intensive benchmarks show that RRPO outperforms strong baselines.

Copied to clipboard

Junjun Pan, Yixin Liu, Rui Miao, Kaize Ding, Yu Zheng, Quoc Viet Hung Nguyen, Alan Wee-Chung Liew, Shirui Pan

Understanding and Mitigating Bias Inheritance in LLM-based Data Augmentation on Downstream Tasks (2026.acl-long)

Challenge:	Existing graph anomaly detection methods rely on coarse sentence-level information and overlook fine-grained lexical cues, limiting their reliability and real-world applicability.
Approach:	They propose an explainable and fine-grained safeguarding framework for detecting malicious agents in multi-agent systems (MAS) to incorporate both coarse and fine lexical information for anomalous agent identification.
Outcome:	Extensive experiments across diverse MAS topologies and attack scenarios demonstrate robust detection performance and strong interpretability of XG-Guard.

Copied to clipboard

Miaomiao Li, Hao Chen, Yang Wang, Tingyuan Zhu, Weijia Zhang, Kaijie Zhu, Kam-Fai Wong, Jindong Wang

Beyond the Panorama: Training-Free Hierarchical Perception-Reasoning for Fine-Grained Vision in MLLMs (2026.acl-long)

Challenge:	Generating synthetic datasets via large language models (LLMs) has emerged as promising approach to improve LLM performance.
Approach:	They propose three mitigation strategies to mitigate bias inheritance in LLMs by analyzing real and LLM-augmented data.
Outcome:	The proposed methods can work differently on different tasks and biases.

Copied to clipboard

Xiaoyang Yi, Jing Chen, Li Peng, Yuru Bao, Jian Zhang

To Judge or Not to Judge: Can Large Language Models Leverage the Dispute Focus in Legal Judgment? (2026.acl-long)

Challenge:	Existing multimodal large language models (MLLMs) face challenges in fine-grained visual tasks.
Approach:	They propose a training-free hierarchical perception-reasoning framework that enhances fine-grained visual understanding by simulating human perception mechanisms.
Outcome:	The proposed framework enhances fine-grained visual understanding by simulating human perception mechanisms.

Copied to clipboard

Luoming Hu, Liang Yang, Jingjie Zeng, Zijie Xing

CPR-RAG: Clinical Prior-Regularized Retrieval for Anatomy-Aware 3D CT Report Generation (2026.acl-long)

Challenge:	Existing research on large language models for legal judgment prediction fails to address the complexity of civil judicial cases.
Approach:	They propose a framework that leverages the dispute focus to guide LLMs through a structured, judge-like cognitive workflow.
Outcome:	The proposed framework can guide LLMs through a structured, judge-like cognitive workflow.

Copied to clipboard

Sungkyu Yang, Kang-Min Kim, Mansu Kim

Evaluating Answer Leakage Robustness of LLM Tutors against Adversarial Student Attacks (2026.acl-long)

Challenge:	Existing approaches to grounding radiology reports from 3D volumetric data are limited due to visual-semantic ambiguity and lack of "normal" context.
Approach:	They propose a model-agnostic retrieval-augmented generation framework that integrates clinical priors into the retrieval process.
Outcome:	The proposed model improves clinical efficacy across state-of-the-art models.

Copied to clipboard

Jin Zhao, Marta Knežević, Tanja Käser

Expect the Unexpected? Testing the Surprisal of Salient Entities (2026.acl-long)

Challenge:	Large Language Models (LLMs) are increasingly used in education, yet their default usefulness conflicts with pedagogical principles.
Approach:	They propose an adversarial student agent that they fine-tune to jailbreak LLM-based tutors and propose a benchmark to evaluate tutor robustness.
Outcome:	The proposed model fine-tunes to jailbreak LLM-based tutors, and shows that they perform well under adversarial student attacks.

Copied to clipboard

Jessica Lin, Amir Zeldes

ArgGenBench: Benchmarking the Complex Controlled Argument Generation Capability of Large Language Models (2026.acl-long)

Challenge:	Existing work on the Uniform Information Density hypothesis has neglected the relative salience of discourse participants.
Approach:	They propose to use an annotated text to examine how overall salience of entities in discourse relates to surprisal.
Outcome:	The proposed method shows that global salience is a mechanism shaping information distribution in discourse.

Copied to clipboard

Bojun Jin, Jianzhu Bao, Yang Sun, Yice Zhang, Ruifeng Xu

Language Models Learn Universal Representations of Numbers and Here’s Why You Should Care (2026.acl-long)

Challenge:	Existing studies focus on limited control signals such as topic, stance, length, style, strategy, audience, and key aspects, failing to capture this complexity.
Approach:	They propose a benchmark that integrates multi-dimensional control into a single instruction to evaluate LLMs' ability to produce persuasive arguments.
Outcome:	The proposed benchmarks show that existing models fail to capture multifaceted argumentative control signals.

Copied to clipboard

Michal Štefánik, Timothee Mickus, Marek Kadlčík, Bertram Højer, Michal Spiegel, Raúl Vázquez, Aman Sinha, Josef Kuchař, Philipp Mondorf, Pontus Stenetorp

CURE-Med: Curriculum-Informed Reinforcement Learning for Multilingual Medical Reasoning (2026.acl-long)

Challenge:	Prior work has shown that large language models (LLMs) often converge to accurate input embedding for numbers, based on sinusoidal representations.
Approach:	They show that large language models often converge to accurate input embedding for numbers, based on sinusoidal representations.
Outcome:	The proposed representations are strikingly systematic, and are interchangeable in a large swathe of experimental setups.

Copied to clipboard

Eric Onyame, Akash Ghosh, Subhadip Baidya, Sriparna Saha, Xiuying Chen, Chirag Agarwal

Inverting the Shield: Systematically Generating Safety Tests from Policy Specifications (2026.acl-long)

Challenge:	Recent advances in large language models (LLMs) have produced strong performance in mathematical reasoning and code generation, but medical reasoning remains challenging because it requires domain knowledge.
Approach:	They propose a multilingual medical reasoning dataset with open-ended reasoning queries with a single verifiable answer that spans thirteen languages.
Outcome:	The proposed framework outperforms baselines and scales effectively across thirteen languages.

Copied to clipboard

Xiaoyue Lu, Xianglin Yang, Haijun Liu, Jiahao Liu, Kuntai Cai, Yan Xiao, Jin Song Dong

HEALing Entropy Collapse: Enhancing Exploration in Few-Shot RLVR via Hybrid-Domain Entropy Dynamics Alignment (2026.acl-long)

Challenge:	Existing safety evaluation paradigms rely on constructed benchmarks or dynamic red-teaming to probe potential vulnerabilities.
Approach:	They propose a framework that integrates specification-based software testing with AI safety.
Outcome:	The proposed framework achieves higher coverage and attack success counts compared to baselines.

Copied to clipboard

Zhanyu Liu, Qingguo Hu, Ante Wang, Chenqing Liu, Zhishang Xiang, Hui Li, Delai Qiu, Jinsong Su

Do LLMs Encode Functional Importance of Reasoning Tokens ? (2026.acl-long)

Challenge:	Existing methods for training reasoning-oriented large language models assume high-resource settings with abundant data.
Approach:	They propose a framework that integrates high-value general-domain data to promote more diverse exploration.
Outcome:	The proposed framework matches or surpasses RLVR trained with 32 target-domain samples using 32 target domain samples.

Copied to clipboard

Janvijay Singh, Dilek Hakkani-Tür

EA-Agent: A Structured Multi-Step Reasoning Agent for Entity Alignment (2026.acl-long)

Challenge:	Existing compact reasoning approaches generate long reasoning chains, but they lack a mechanism to encode token-level functional importance for answer generation.
Approach:	They propose a procedure that iteratively removes reasoning tokens from models and prunes them to yield length-controlled reasoning chains.
Outcome:	The proposed procedure outperforms a frontier model at reasoning lengths and shows that attention scores predict greedy pruning ranks.

Copied to clipboard

Yixuan Nan, Xixun Lin, Yanmin Shang, Ge Zhang, Zheng Fang, Fang Fang, Yanan Cao

Too Nice to Tell the Truth: Quantifying Agreeableness-Driven Sycophancy in Role-Playing Language Models (2026.acl-long)

Challenge:	Entity alignment (EA) aims to identify entities across different knowledge graphs (KGs) that refer to the same real-world object.
Approach:	They propose to use large language models to integrate semantic knowledge into EA to identify entities across different knowledge graphs that refer to the same object.
Outcome:	The proposed agent outperforms existing methods and achieves state-of-the-art performance on three benchmark datasets.

Copied to clipboard

Arya Shah, Deepali Mishra, Chaklam Silpasuwanchai

Mobile-R1: Towards Interactive Capability for VLM-Based Mobile Agent via Systematic Training (2026.acl-long)

Challenge:	Large language models are increasingly used as conversational agents that adopt personas and role-play characters at user request.
Approach:	They propose to examine how persona agreeableness influences sycophancy across 13 small, open-weight language models ranging from 0.6B to 20B parameters.
Outcome:	The proposed model consists of 275 personas and exposes them to 4,950 sycophancy-eliciting prompts spanning 33 topic categories.

Copied to clipboard

Jihao Gu, Qihang Ai, Yingyao Wang, Pi Bu, Jingxuan Xing, Yue Cao, Zekun Zhu, Wei Jiang, Ziming Wang, Yingxiu Zhao, Ming-Liang Zhang, Jun Song, Yuning Jiang, Bo Zheng

QuantumQA: Enhancing Scientific Reasoning via Physics-Consistent Dataset and Verification-Aware Reinforcement Learning (2026.acl-long)

Challenge:	Existing approaches to training agents for visual-language models trap them in local optima, hindering exploration and error correction with the environment.
Approach:	They propose a hierarchical training recipe that bridges atomic action execution and strategic task completion.
Outcome:	The proposed training recipe bridges atomic action execution and strategic task completion.

Copied to clipboard

Songxin Qu, Tai-Ping Sun, Yun-Jie Wang, Huan-Yu Liu, Cheng Xue, Xiao-Fan Xu, Han Fang, Yang Yang, Yu-Chun Wu, Guo-Ping Guo, Zhao-Yun Chen

The “Knowledge–Behavior Gap” in Cultural Taboo Safety of Large Language Models (2026.acl-long)

Challenge:	Large language models lack reliability in scientific domains that require strict adherence to physical constraints.
Approach:	They propose a large-scale dataset constructed via a task-adaptive strategy and a hybrid verification protocol that combines deterministic solvers with semantic auditing to guarantee scientific rigor.
Outcome:	The proposed model outperforms baselines and general-purpose preference models and is competitive with proprietary models.

Copied to clipboard

Ying He, Sihang Jiang, Xingzhou Chen, Zhouhong Gu, Yiwei Gu, Minggui HE, Shimin Tao, null Mahongxia, Yanghua Xiao

OSCBench: Benchmarking Object State Change in Text-to-Video Generation (2026.acl-long)

Challenge:	Existing cultural benchmarks assess cultural knowledge or values biases, but ignore cultural taboos.
Approach:	They propose a benchmark to evaluate and improve the cultural taboo safety of large language models.
Outcome:	The proposed benchmark spans 77 countries and regions, and includes over 2,020 taboos.

Copied to clipboard

Xianjing Han, Bin Zhu, Shiqi Hu, Franklin Mingzhe Li, Patrick Carrington, Roger Zimmermann, Jingjing Chen

Evaluation Pitfalls and Challenges in Multimedia Event Extraction (2026.acl-long)

Challenge:	Existing benchmarks focus on perceptual quality, text–video alignment, or physical plausibility, leaving a critical aspect of action understanding unexplored.
Approach:	They introduce a benchmark specifically designed to assess OSC performance in T2V models.
Outcome:	The proposed benchmark assesses the performance of open-source and proprietary T2V models on object state change (OSC) in the context of novel and compositional scenarios.

Copied to clipboard

Philipp Seeberger, Steffen Freisinger, Tobias Bocklet, Korbinian Riedhammer

UMPIRE: Unveiling LLM-generated Posts via Redundant Expressions (2026.acl-long)

Challenge:	Recent work has focused on textual content, but recent work has explored the integration of additional modalities to support more accurate and comprehensive event understanding.
Approach:	They propose to analyze the evaluation pitfalls of multimedia event extraction by combining textual and visual inputs to identify events and their arguments across multiple modalities.
Outcome:	The proposed model overestimates performance and performance of the proposed model in a series of controlled experiments under a strict evaluation framework.

Copied to clipboard

Xiaoquan Yi, Haixing Wu, Haozhao Wang, Yichen Li, Yuhua Li, Rui Zhang, Ruixuan Li

SpeechMedAssist: Efficiently and Effectively Adapting Speech Language Models for Medical Consultation (2026.acl-long)

Challenge:	Existing methods to verify the provenance of multimodal content fall into two categories: traditional methods rely on low-level artifacts or unimodal statistics.
Approach:	They propose a semantic decomposition mechanism that disentangles textual embeddings into redundant and complementary components and a latent redundancy regularization loss to encourage LLM-generated content to exhibit high semantic redundancies.
Outcome:	The proposed method outperforms state-of-the-art detection methods across multiple datasets and achieves 5.38% improvement in accuracy.

Copied to clipboard

Sirry Chen, Jieyi Wang, Wei Chen, Zhongyu Wei

An Experimental Study on the Influence of Culture on Cross-Lingual Sentiment Transfer (2026.acl-long)

Challenge:	Recent advances in speech language models have enabled more natural speech-based interactions, but the scarcity of medical speech data and the inefficiency of fine-tuning on speech data hinder adoption of SpeechLMs in medical consultation.
Approach:	They propose a SpeechLM natively capable of conducting speech-based multi-turn interactions with patients.
Outcome:	The proposed model outperforms baselines in both effectiveness and robustness in most evaluation settings.

Copied to clipboard

Ahao Liu, Haitong Yang, Wang Chuanrong

Attribution, Citation, and Quotation: A Survey of Evidence-based Text Generation with Large Language Models (2026.acl-long)

Challenge:	Identical linguistic expressions can convey different sentiments across cultural contexts . current multilingual models often reduce language to symbolic representation . cultural misalignment is a structural bottleneck, authors say .
Approach:	They conduct an empirical study to quantify the influence of culture on cross-lingual sentiment transfer across 7 common SMLMs and 5 linguistically diverse languages.
Outcome:	The proposed model disentangles cultural factors from confounding variables and shows cultural distance is a negative predictor of transfer performance.

Copied to clipboard

Tobias Schreieder, Tim Schopf, Michael Färber

MirrorCAPTCHA: Wild CAPTCHA, Wild Distribution, Wild Web-based Platform Meet Multimodal LLM Agents (2026.acl-long)

Challenge:	Recent advances in large language models have raised concerns about reliability and trustworthiness of the models.
Approach:	They analyze 134 papers and introduce a taxonomy of evidence-based text generation with LLMs.
Outcome:	The proposed methods highlight open challenges and outline promising directions for future work.

Copied to clipboard

Xiangyu Wu, Yuwei Hu, Tianyu Cui, Yueying Tian, Qing-Guo Chen, Zhao Xu, Weihua Luo, Kaifu Zhang, Yang Yang, Jianfeng Lu

ReCreate: Reasoning and Creating Domain Agents Driven by Experience (2026.acl-long)

Challenge:	Existing agent benchmarks fail to evaluate an agent's real-world capacity to handle CAPTCHA . Existing benchmarks ignore this practical challenge, failing to evaluate agents' ability to handle complex visual CAPTchas.
Approach:	They propose a benchmark annotated with Weighted Pass Rate and a new metric to measure agent's ability to handle CAPTCHA.
Outcome:	The proposed benchmark outperforms current state-of-the-art closed-source models on mirrorCAPTCHA and achieves 9.4% higher average weighted pass rate and 2.13% higher average Completion degree.

Copied to clipboard

Zhezheng Hao, Hong Wang, Jian Luo, Jianqing Zhang, Yuyan Zhou, Qiang Lin, Can Wang, Hande Dong, Jiawei Chen

Can AI-Generated Persuasion Be Detected? Persuaficial Benchmark and AI vs. Human Linguistic Differences (2026.acl-long)

Challenge:	Large Language Model (LLM) agents are reshaping the industrial landscape, but tasks differ widely, making them labor-intensive to build.
Approach:	They propose an experience-driven framework for the automatic creation of domain agents . they leverage agent interaction histories to provide rich concrete signals on success or failure .
Outcome:	The proposed framework outperforms human-designed agents and existing methods in experiments across diverse domains.

Copied to clipboard

Arkadiusz Modzelewski, Paweł Golik, Anna Kołos, Giovanni Da San Martino

Challenge:	Large Language Models (LLMs) can generate highly persuasive text, raising concerns about misuse for propaganda, manipulation, and other harmful purposes.
Approach:	They propose a multilingual benchmark to compare LLM-generated persuasive texts with human-written ones.
Outcome:	The proposed benchmark compares human-authored and LLM-generated persuasive texts . it finds that overtly persuasive LLMs are easier to detect than human-written ones .

GASE: Graph-Aware Semantic Embedding Learning with Frozen LLMs for Text-Attributed Graphs (2026.acl-long)

Copied to clipboard

Mingqian Ding, Jianjun Li, Wenqi Yang, Zhibo Zhang, Yushen Fang

Compositional Steering of Large Language Models with Steering Tokens (2026.acl-long)

Challenge:	Large Language Models (LLMs) have shown strong potential for text-attributed graph (TAG) learning, yet effectively integrating LLM semantics with graph structural information remains challenging.
Approach:	They propose a framework for learning Graph-Aware Semantic Embeddings using frozen LLMs.
Outcome:	The proposed framework outperforms state-of-the-art methods on node classification and achieves a 5 speedup over fine-tuning-based methods.

Copied to clipboard

Gorjan Radevski, Kiril Gashteovski, Giwon Hong, Carolin Lawrence, Goran Glavaš

Rethinking Entropy Interventions in RLVR: An Entropy Change Perspective (2026.acl-long)

Challenge:	Existing work addresses compositional steering for a single behavior, but it's limited by computational costs and performance.
Approach:	They propose compositional steering tokens for multi-behavior steering by embedding individual behaviors into dedicated tokens via self-distillation.
Outcome:	The proposed steering tokens generalize well to unseen compositions compared to competing approaches.

Copied to clipboard

Zhezheng Hao, Hong Wang, Haoyang Liu, Jian Luo, Jiarui Yu, Hande Dong, Qiang Lin, Can Wang, Jiawei Chen

SwissGov-RSD: A Human-annotated, Cross-lingual Benchmark for Token-level Recognition of Semantic Differences Between Related Documents (2026.acl-long)

Challenge:	Large Language Models (LLMs) have remarkable reasoning capabilities in complex tasks such as mathematics and coding.
Approach:	They propose an entropy-modulation method that adaptively reweighs tokens based on theoretically-estimated entropic variations.
Outcome:	The proposed method outperforms state-of-the-art methods in six mathematical reasoning and three coding benchmarks.

Copied to clipboard

Michelle Wastl, Jannis Vamvas, Rico Sennrich

Seeing but Not Thinking: Routing Distraction in Multimodal Mixture-of-Experts (2026.acl-long)

Challenge:	Recognizing semantic differences across documents is crucial for text generation evaluation and content alignment . but as a standalone task, it has received little attention, a new study shows . evaluating semantic differences between documents is an underexplored challenge in natural language understanding .
Approach:	They introduce SwissGov-RSD, the first naturalistic, document-level, cross-lingual dataset for semantic difference recognition.
Outcome:	The proposed dataset shows that current approaches perform poorly on monolingual, sentence-level and synthetic benchmarks.

Copied to clipboard

Haolei Xu, Haiwen Hong, Hongxing Li, Rui Zhou, Yang Zhang, Longtao Huang, Hui Xue, Yongliang Shen, Weiming Lu, Yueting Zhuang

RubricBench: Aligning Model-Generated Rubrics with Human Standards (2026.acl-long)

Challenge:	Existing multimodal Mixture-of-Experts models accurately perceive image content yet fail in subsequent reasoning . Seeing but not thinking phenomenon is a puzzling phenomenon .
Approach:	They propose a routing-guided intervention method that enhances domain expert activation.
Outcome:	The proposed method achieves consistent improvements on visual reasoning tasks.

Copied to clipboard

Junyi Zhou, Qiyuan Zhang, Yufei Wang, Fuyuan Lyu, Yidong Ming, Can Xu, Qingfeng Sun, Kai Zheng, Peng Kang, Xue Liu, Chen Ma

Challenge:	Existing benchmarks lack discriminative complexity and ground-truth rubric annotations required for rigorous evaluation.
Approach:	They propose a curated benchmark with 1,147 pairwise comparisons to assess the reliability of rubric-based evaluation.
Outcome:	The proposed benchmarks show that they support diverse domains, exhibit discriminative ability, provide high-quality annotations, and include human-authored rubrics.

Verify Before You Commit: Towards Faithful Reasoning in LLM Agents via Self-Auditing (2026.acl-long)

Copied to clipboard

Wenhao Yuan, Chenchen Lin, Jian Chen, Jinfeng Xu, Xuehe Wang, Edith Cheuk-Han Ngai

GiLT: Augmenting Transformer Language Models with Dependency Graphs (2026.acl-long)

Challenge:	Existing approaches to reasoning faithfulness violate constraints, authors say . a science fantasy series and companion books are among the books .
Approach:	They propose a framework that enforces verification over internal belief states within the agent before action commitment, achieving faithful reasoning.
Outcome:	The proposed framework improves reasoning faithfulness while preserving competitive end-task performance.

Copied to clipboard

Tianyu Huang, Yida Zhao, Chuyan Zhou, Kewei Tu

TAMAS: Benchmarking Adversarial Risks in Multi-Agent LLM Systems (2026.acl-long)

Challenge:	Recent work focuses on syntactic tree structures of languages, in particular constituency tree structures.
Approach:	They propose a Graph-Infused Layers Transformer Language Model which leverages dependency graphs to augment Transformer language models.
Outcome:	The proposed model achieves better syntactic generalization while maintaining competitive perplexity compared with baseline models.

Copied to clipboard

Ishan Kavathekar, Hemang Jain, Ameya Rathod, Ponnurangam Kumaraguru, Tanuja Ganu

How Controllable Are Large Language Models? A Unified Evaluation across Behavioral Granularities (2026.acl-long)

Challenge:	Existing benchmarks and datasets focus on single-agent settings, failing to capture the unique vulnerabilities of multi-agend LLM dynamics and co-ordination.
Approach:	They propose a benchmark to evaluate the robustness and safety of multi-agent LLM systems.
Outcome:	The proposed benchmark evaluates the robustness and safety of multi-agent LLM systems.

Copied to clipboard

Ziwen Xu, Kewei Xu, Haoming Xu, Haiwen Hong, Longtao Huang, Hui Xue, Ningyu Zhang, Yongliang Shen, Guozhou Zheng, Huajun Chen, Shumin Deng

Scaling Behaviors of LLM Reinforcement Learning Post-Training: An Empirical Study in Mathematical Reasoning (2026.acl-long)

Challenge:	Large language models (LLMs) are increasingly deployed in socially sensitive domains, yet their unpredictable behaviors pose significant risks.
Approach:	They propose a hierarchical benchmark for evaluating LLM controllability across three domains: language features, sentiment, and personality.
Outcome:	The proposed framework offers a principled and interpretable framework for safe and controllable LLM behavior serving as a foundation for future research.

Copied to clipboard

Zelin Tan, Hejia Geng, Xiaohang Yu, Mulei Zhang, Guancheng Wan, Yifan Zhou, Qiang He, Xiangyuan Xue, Heng Zhou, Yutao Fan, Zhong-Zhi Li, Zaibin Zhang, Guibin Zhang, Chen Zhang, Zhenfei Yin, Philip Torr, Lei Bai

RubricHub: A Comprehensive and Highly Discriminative Rubric Dataset via Automated Coarse-to-Fine Generation (2026.acl-long)

Challenge:	elucidating scaling laws for large language models (LLMs) during pre-training remains unexplored.
Approach:	They characterize how model scale, data, and compute interact during pre-training . they find that large models consistently demonstrate superior compute and data efficiency .
Outcome:	The proposed scaling laws offer practical guidance for scaling reasoning capabilities through reinforcement learning post-training.

Copied to clipboard

Sunzhu Li, Jiale Zhao, Huimin Ren, Zhenlin Wei, Yang Zhou, Jingwen Yang, Shunyu Liu, Kaike Zhang, Chen Wei

MIND Your Reasoning: A Meta-Cognitive Intuitive-Reflective Network for Dual-Reasoning in Multimodal Stance Detection (2026.acl-long)

Challenge:	Existing methods for generating open-ended rubrics suffer from scalability bottlenecks and coarse criteria resulting in a supervision ceiling effect.
Approach:	They propose a framework for automated Coarse-to-Fine Rubric Generation . their framework uses principle-guided synthesis, multi-model aggregation, difficulty evolution .
Outcome:	The proposed framework produces comprehensive and highly discriminative criteria capable of capturing the subtle nuances.

Copied to clipboard

Bingbing Wang, Zhengda Jin, Bin Liang, Wenjie Li, Jing Li, Ruifeng Xu, Min Zhang

A Survey of Inductive Reasoning for Large Language Models (2026.acl-long)

Challenge:	Existing methods operate by learning to fuse modalities, leading to frequent misjudgments.
Approach:	They propose a paradigm shift from learning to fuse to *learning the reason's process' inspired by the dual-process theory of human cognition, MIND operationalizes a self-improving loop.
Outcome:	The proposed model significantly outperforms baseline models and exhibits strong generalization.

Copied to clipboard

Kedi Chen, Dezhao Ruan, Yuhao Dan, Yaoting Wang, Siyu Yan, Xuecheng Wu, Yinqi Zhang, Qin Chen, Jie Zhou, Liang He, Biqing Qi, Linyang Li, Qipeng Guo, Xiaoming Shi, Wei Zhang

CogToM: A Comprehensive Theory of Mind Benchmark inspired by Human Cognition for Large Language Models (2026.acl-long)

Challenge:	Inductive reasoning is an important task for large language models (LLMs).
Approach:	They propose a survey of inductive reasoning for large language models . they categorize methods into three main areas: post-training enhancement, test-time exploration, and data augmentation.
Outcome:	The proposed method improves inductive reasoning in large language models.

Copied to clipboard

Haibo Tong, Zeyang Yue, Feifei Zhao, Erliang Lin, Lu Jia, Ruolin Chen, Yinqian Sun, Qian Zhang, Yi Zeng

CITE: Benchmarking Heterogeneous Text-Attributed Graph Models (2026.acl-long)

Challenge:	Existing benchmarks for Large Language Models (LLMs) are limited to false belief tasks, highlighting bottlenecks in specific dimensions.
Approach:	They propose a benchmark to evaluate Large Language Models' Theory of Mind capabilities . they evaluate 8000 bilingual instances across 46 paradigms and validated by 49 human annotators .
Outcome:	The proposed benchmark reveals performance heterogeneities and bottlenecks in 22 representative models.

Copied to clipboard

Chenghao Zhang, Qingqing Long, Ludi Wang, Wenjuan Cui, Jianjun Yu, Yi Du

MM-StanceDet: Retrieval-Augmented Multi-modal Multi-agent Stance Detection (2026.acl-long)

Challenge:	Recent advances in large language models and text-aware graph learning have increased interest in reasoning over text-attributed graphs.
Approach:	They propose a large-scale heterogeneous text-attributed graph benchmark for catalytic materials that contains over 438K nodes and 1.2M edges . they establish standardized evaluation protocols for node classification and link prediction and conduct ablation studies to assess the impact of graph heterogenity and textual attributes.
Outcome:	The proposed benchmarks are compared to existing methods and provide a baseline for the evaluation of four classes of learning paradigms.

Copied to clipboard

Weihai Lu, Zhejun Zhao, Yanshu Li, Huan He

Disentangling Reasoning Logic to Resolve Explicit Knowledge Conflicts (2026.acl-long)

Challenge:	Existing methods for multimodal stance detection face contextual grounding, cross-modal interpretation ambiguity, and single-pass reasoning fragility.
Approach:	They propose a multi-agent framework that integrates Retrieval Augmentation for contextual grounding, specialized Multimodal Analysis agents for nuanced interpretation, Reasoning-Enhanced Debate stage and Self-Reflection for robust adjudication.
Outcome:	Extensive experiments on five datasets show that the proposed framework outperforms state-of-the-art methods.

Copied to clipboard

Xianda Zheng, Zijian Huang, Meng-Fen Chiang, Jiamou Liu, Yuan Fang, Michael J. Witbrock, Kaiqi Zhao

UniVocal: Unified Speech-Singing Code-Switching Synthesis (2026.acl-long)

Challenge:	Existing approaches to resolve explicit knowledge conflicts are based on semantic decoding and auxiliary embedding.
Approach:	They propose a framework that adjudicates conflicts by structuring the underlying logic.
Outcome:	Experiments show that the proposed framework improves on existing models.

Copied to clipboard

YuFei Shi, Qian Chen, Wen Wang, Xiangang Li, Zhen-Hua Ling, Yang Ai

ReasoningGuard: Safeguarding Large Reasoning Models with Inference-time Safety Aha Moments (2026.acl-long)

Challenge:	Existing systems cannot automatically determine when to switch between modes based on text content.
Approach:	They propose a unified framework that implicitly infers vocal modes from text context to pioneer SCS Synthesis.
Outcome:	The proposed framework infers vocal modes solely from text context to pioneer SCS Synthesis.

Copied to clipboard

Yuquan Wang, Mi Zhang, Yining Wang, Geng Hong, Mi Wen, Xiaoyu You, Min Yang

VecInfer: Efficient LLM Inference with Low-Bit KV Cache via Outlier-Suppressed Vector Quantization (2026.acl-long)

Challenge:	Existing defenses for Large Reasoning Models (LRMs) depend on costly fine-tuning and additional expert knowledge, which limits their scalability.
Approach:	They propose an inference-time safeguard for Large Reasoning Models that injects safety aha moments into the reasoning process to guide the model towards harmless yet helpful reasoning.
Outcome:	The proposed safeguard outperforms nine existing safeguards while avoiding common exaggerated safety issues.

Copied to clipboard

Dingyu Yao, Chenxu Yang, Zhengyang Tong, Zheng Lin, Wei Liu, Jian Luan, Weiping Wang

When Does Mixing Help? Analyzing Query Embedding Interpolation in Multilingual Dense Retrieval (2026.acl-long)

Challenge:	Existing quantization methods for large language models suffer performance degradation at ultra-low bit-widths due to key cache outliers.
Approach:	They propose a vector quantization method that suppresses outliers in the key cache and reduces memory access overhead.
Outcome:	The proposed method outperforms baseline quantization methods across long-context understanding and mathematical reasoning tasks while minimizing memory access overhead.

Copied to clipboard

Tongyao Zhu, Huang Chao Ming, Min-Yen Kan

WIST: Web-Grounded Iterative Self-Play Tree for Domain-Targeted Reasoning Improvement (2026.acl-long)

Challenge:	Existing multilingual retrievers are largely untested for mixed language queries . sensitivity of dense retrievers to mixed language querying remains poorly understood .
Approach:	They propose to use mixed queries as an interpolation of monolingual embeddings to evaluate retrieval performance.
Outcome:	The proposed model outperforms the best monolingual endpoint in 88/105 cases.

Copied to clipboard

Fangyuan Li, Pengfei Li, Shijie Wang, Junqi Gao, Jianxing Liu, Biqing Qi, Yuqiang Li

Frankentext: Stitching random text fragments into long-form narratives (2026.acl-long)

Challenge:	Existing methods for self-improvement of large language models with verifiable rewards (RLVR) can drift over iterations, while corpus-grounded approaches rely on curated data environments.
Approach:	They propose a Web-grounded Iterative Self-play Tree framework for domain-targeted reasoning improvement that learns directly from the open-web without requiring any pre-arranged domain corpus.
Outcome:	The proposed framework outperforms both purely endogenous self-evolution and corpus-grounded self-play baselines and is domain-steerable.

Copied to clipboard

Chau Minh Pham, Jenna Russell, Dzung Pham, Mohit Iyyer

SAFO: Stable Adaptive Fairness Optimization for LLM-Based Social Survey Simulation (2026.acl-long)

Challenge:	a new approach to generate narratives that can evade detection is needed . authors say that the livelihood of writers is threatened by the quality of AI writing .
Approach:	They propose a long-form narrative generation paradigm that treats an LLM as a composer of existing texts rather than as an author.
Outcome:	a new model improves over vanilla LLM generation in key writing quality metrics . human annotators praise the model for inventive premises, vivid descriptions, and dry humor . the model raises concerns about the publishing economy and the livelihood of writers .

Copied to clipboard

Chenxi Lin, Zhuoren Jiang, Kaisong Song, Yiquan Wu

The Prosody of Emojis (2026.acl-long)

Challenge:	Social survey simulations are increasingly used to improve minority performance and social-welfare metrics.
Approach:	They propose a dynamic utility–fairness optimization framework for LLM-based survey simulation that explicitly targets fairness and training stability.
Outcome:	The proposed framework improves minority performance and social-welfare metrics on three large-scale survey datasets from China, the U.S. and Europe.

Copied to clipboard

Giulio Zhou, Tsz Kin Lam, Alexandra Birch, Barry Haddow

Piece of Table: A Divide-and-Conquer Approach for Selecting Subtables in Table Question Answering (2026.acl-long)

Challenge:	emojis are useful in spoken communication because they add affective and pragmatic nuance to textual cues.
Approach:	They analyze human speech data to find prosodic features that are important in spoken communication.
Outcome:	The proposed model shows that speakers adapt prosody based on emoji cues, and that listeners can recover intended meanings significantly above chance.

Copied to clipboard

Wonjin Lee, Kyumin Kim, Sungjae Lee, Jihun Lee, Kwang In Kim

QA-MoE: Towards a Continuous Reliability Spectrum with Quality-Aware Mixture of Experts for Robust Multimodal Sentiment Analysis (2026.acl-long)

Challenge:	Existing approaches to QA tables rely on independent row or column selection, fail to capture cross-row and cross-column dependencies, or attempt global reasoning.
Approach:	They propose a divide-and-conquer subtable selection framework that aggregates local evidence without requiring explicit global reasoning.
Outcome:	The proposed framework outperforms previous approaches to table QA in the noisy context.

Copied to clipboard

Yitong Zhu, Yuxuan Jiang, Guanxuan Jiang, Bojing Hou, Peng Yuan Zhou, Ge Lin, Yuyang Wang

Enhancing LLM-based Search Agents via Contribution Weighted Group Relative Policy Optimization (2026.acl-long)

Challenge:	Existing models that use multimodal inputs are often noisy or incomplete.
Approach:	They propose a Quality-Aware Mixture-of-Experts framework that quantifies modality reliability via aleatoric uncertainty.
Outcome:	The proposed framework is competitive or state-of-the-art across diverse degradation scenarios and exhibits a promising One-Checkpoint-for-all property in practice.

Copied to clipboard

Junzhe Wang, Zhiheng Xi, Yajie Yang, Hao Luo, Shihan Dou, Tao Gui, Qi Zhang

Why Steering Works: Toward a Unified View of Language Model Parameter Dynamics (2026.acl-long)

Challenge:	Existing approaches to training large language models suffer from unstable value estimation, whereas outcome supervision struggles with credit assignment due to sparse, trajectory-level rewards.
Approach:	They propose a framework that integrates process supervision into group relative policy optimization.
Outcome:	The proposed framework outperforms standard GRPO on knowledge-intensive benchmarks by 5.0% and 6.3% on Qwen3-1.7B.

Copied to clipboard

Ziwen Xu, Chenyan WU, Hengyu Sun, Haiwen Hong, Mengru Wang, Yunzhi Yao, Longtao Huang, Hui Xue, Shumin Deng, Zhixuan Chu, Huajun Chen, Ningyu Zhang

EmoMAS: Emotion-Aware Multi-Agent System for High-Stakes Edge-Deployable Negotiation with Bayesian Orchestration (2026.acl-long)

Challenge:	Methods for controlling large language models (LLMs) are often studied in isolation, obscuring connections and making comparison difficult.
Approach:	They propose a preference-utility analysis that separates control effects into preference and utility, and measures both on a shared log-odds scale using polarity-paired contrastive examples.
Outcome:	The proposed approach improves preference while preserving utility.

Copied to clipboard

Yunbo Long, Yuhan Liu, Liming Xu

Decisive: Guiding User Decisions with Optimal Preference Elicitation from Unstructured Documents (2026.acl-long)

Challenge:	Large language models (LLMs) are increasingly used for automated negotiation, but their cloud-centric paradigm exposes sensitive negotiations to privacy and security risks.
Approach:	They propose a Bayesian multi-agent framework that transforms emotional decision-making from reactive to strategic.
Outcome:	EmoMAS leverages a Bayesian orchestrator to coordinate three specialized agents: game-theoretic, reinforcement learning, and psychological coherence models.

Copied to clipboard

Akriti Jain, Anish Mulay, Divyansh Verma, Aishani Pandey, Pritika Ramu, Aparna Garimella

Challenge:	Existing methods for analyzing information from multiple sources are often too complex or fail to capture nuanced preferences accurately.
Approach:	They propose an interactive decision-making framework that combines document-grounded reasoning with Bayesian preference inference.
Outcome:	The proposed approach outperforms general-purpose LLMs and existing decision-support systems in achieving up to 20% improvement in decision accuracy over strong baselines across domains.

From Naturalness to Norms: Interactional Cultural Competence for SpeechLMs (2026.acl-long)

Copied to clipboard

Santosh T.y.s.s

Optimizing User Profiles via Contextual Bandits for Retrieval-Augmented LLM Personalization (2026.acl-long)

Challenge:	Spoken language models are increasingly real-time conversational actors.
Approach:	They propose a speech-first view of cultural competence as interactional competence . they synthesize social-science foundations into a taxonomy of culture-bearing signals in speech .
Outcome:	The proposed model is based on a theory-derived taxonomy of culture-bearing signals in speech . it shows that cultural appropriateness is not a generic human-likeness .

Copied to clipboard

Linfeng Du, Ye Yuan, Zichen Zhao, Fuyuan Lyu, Emiliano Penaloza, Xiuying Chen, Zipeng Sun, Jikun Kang, Laurent Charlin, Xue Liu, Haolun Wu

Teaching Language Models to Check Grounded Claim Factuality with Human Test-Taking Strategies (2026.acl-long)

Challenge:	Existing approaches for personalizing large language models require modifying parameters.
Approach:	They propose a lightweight approach to personalizing large language models via retrieval augmentation . relevance serves as an unreliable proxy for utility, they argue .
Outcome:	The proposed framework outperforms strong heuristic and retrieval-augmented baselines on nine personalization tasks.

Copied to clipboard

Yuxuan Ye, Raul Santos-Rodriguez, Edwin Simpson

Evo-PI: Aligning Medical Reasoning via Evolving Principle-Guided Supervision (2026.acl-long)

Challenge:	Existing methods for factuality checking require dataset-specific threshold tuning, while LLM-based approaches often use direct prompting.
Approach:	They propose to use a reading comprehension task to check for true/false claim factuality and prompt LLMs with explicit test-taking strategies for efficient reasoning.
Outcome:	The proposed method reduces token usage by over 80% compared to unguided open-ended reasoning and achieves competitive performance to more expensive alternatives.

Copied to clipboard

Xianda Zheng, Huan Gao, Meng-Fen Chiang, Michael J. Witbrock, Kaiqi Zhao, Shangyang Li

Understanding LLM Performance Degradation in Multi-Instance Processing: The Roles of Instance Count and Context Length (2026.acl-long)

Challenge:	Existing models with static prompts, rules, or reward models are constrained by static supervision, which often fails to shape the underlying reasoning process, leading to brittle generalization and performance saturation in complex decision-making tasks.
Approach:	They propose a principle-centric learning framework that treats reasoning principles as explicit, language-based supervision signals that can be generated, evaluated, and iteratively evolved.
Outcome:	The proposed framework treats reasoning principles as explicit, language-based supervision signals that can be generated, evaluated, and iteratively evolved.

Copied to clipboard

Jingxuan Chen, Mohammad Taher Pilehvar, Jose Camacho-Collados

Gated Differentiable Working Memory for Long-Context Language Modeling (2026.acl-long)

Challenge:	Large Language Models (LLMs) are used for processing multiple documents or analysis over a number of instances.
Approach:	They perform a comprehensive evaluation of the multi-instance processing ability of LLMs for tasks in which they excel individually.
Outcome:	The proposed model performs well on tasks in which it excels individually.

Copied to clipboard

Lingrui Mei, Shenghua Liu, Yiwei Wang, Yuyao Ge, Baolong Bi, Jiayu Yao, Jun Wan, Ziling Yin, Jiafeng Guo, Xueqi Cheng

Stress Testing Factual Consistency Metrics for Long-Document Summarization (2026.acl-long)

Challenge:	Long contexts break transformers, attention scores dilute, model cannot adapt to novel patterns at inference time.
Approach:	They propose a framework that gates the memory consolidation process by estimating Contextual Utility . they propose GDWM to maintain a form of working memory with constant contexts .
Outcome:	The proposed framework achieves comparable or superior performance on sparse-information tasks with 4 fewer gradient steps compared to uniform baselines.

Copied to clipboard

Zain Muhammad Mujahid, Dustin Wright, Isabelle Augenstein

Do LLMs Know Tool Irrelevance? Demystifying Structural Alignment Bias in Tool Invocations (2026.acl-long)

Challenge:	Existing short-form summarization metrics struggle with input length limitations and long-range dependencies.
Approach:	They propose to evaluate the reliability of six widely used reference-free factuality metrics in the long-document setting by applying seven factually-preserving perturbations to summaries.
Outcome:	The proposed short-form summarization metrics struggle with long-range dependencies and input length limitations.

Copied to clipboard

Yilong Liu, Xixun Lin, Pengfei Cao, Ge Zhang, Fang Fang, Yanan Cao

Challenge:	Large language models (LLMs) have impressive capabilities in utilizing external tools, but in practice, they are often exposed to tools that are irrelevant to the user’s query, in which case the desired behavior is to refrain from invocations.
Approach:	They propose a new dataset that decouples structural alignment from semantic relevance and propose rebalancing strategies that effectively mitigates structural alignment bias.
Outcome:	The proposed approach effectively mitigates structural alignment bias without degrading general tool-use capabilities.

PASs-MoE: Mitigating Misaligned Co-drift among Router and Experts via Pathway Activation Subspaces for Continual Learning (2026.acl-long)

Copied to clipboard

ZhiYan Hou, Haiyun Guo, Haokai Ma, Yandu Sun, Yonghui Yang, Jinqiao Wang

Challenge:	Existing LoRA-based Mixture-of-Experts (MoE) methods often jointly update the router and experts in an indiscriminate way, causing the router’s preferences to co-drift with experts’ adaptation pathways and exacerbate forgetting.
Approach:	They propose a LoRA-induced subspace that reflects which low-rank pathway directions an input activates in each expert, providing a capability-aligned coordinate system for routing and preservation.
Outcome:	The proposed method outperforms conventional continual learning baselines and MoE–LoRA variants in accuracy and resistance to forgetting, without increasing model parameters.

LLM-Generated Text May Harm Your Retrieval! A Robust Detection Strategy for Retrieval-Augmented Generation (2026.acl-long)

Copied to clipboard

Zhaoheng Huang, Yutao Zhu, Ji-Rong Wen, Zhicheng Dou

GQLBench: A Large-Scale Cross-Domain, Cross-Dialect Benchmark for NL2GQL (2026.acl-long)

Challenge:	Retrieval-augmented generation (RAG) improves accuracy and timeliness of large language models, but external corpora may become contaminated with LLM-generated texts.
Approach:	They propose a method that integrates external knowledge retrieved from external sources into RAG to filter out LLM-generated texts from retrieved results.
Outcome:	The proposed method mitigates performance degradation and improves stability of RAG systems.

Copied to clipboard

Yanning Su, Yuhang Zhou, Yang Fang, Sen Liu, Guangnan Ye, Hongfeng Chai

ReviewGrounder: Improving Review Substantiveness with Rubric-Guided, Tool-Integrated Agents (2026.acl-long)

Challenge:	despite growing interest in NL2GQL, benchmarking progress has been constrained by the lack of resources that are simultaneously large-scale, cross-domain, and cross-dialect.
Approach:	They propose a framework that integrates NL2SQL-to-NL2GQL conversion with graph-native data generation.
Outcome:	The proposed framework supports execution-based evaluation on Cypher and ISO-GQL, covering hundreds of graph databases and over 20k natural language questions for each dialect.

Copied to clipboard

Zhuofeng Li, Yi Lu, Dongfu Jiang, Haoxiang Zhang, Yuyang Bai, Chuan Li, Yu Wang, Shuiwang Ji, Jianwen Xie, Yu Zhang

The Retrieval Bottleneck: Scaling Laws for Reinforcement Learning in RAG (2026.acl-long)

Challenge:	Rapid rise in AI conference submissions has driven increasing exploration of large language models (LLMs) for peer review support.
Approach:	They propose a peer review benchmarking tool based on paper-specific rubrics and a rubric-guided framework that decomposes reviewing into drafting and grounding stages.
Outcome:	The proposed framework outperforms baselines with stronger/larger backbones in both alignment with human judgments and rubric-based review quality across 8 dimensions.

Copied to clipboard

Shu Zhou, Jinman Leng, Yufei Song, Xin Wang, Tao Fan, Hao Wang

Generating then Refining for Reliable Knowledge Base Question Answering (2026.acl-long)

Challenge:	Retrieval-augmented generation (RAG) has become the dominant paradigm for building knowledge-intensive language systems.
Approach:	They propose a sigmoidal scaling law that shows that retrieval quality determines the asymptotic performance ceiling.
Outcome:	The proposed model achieves strong performance on knowledge-intensive benchmarks while retaining the predictable scaling long available for pre-training but previously absent in RAG-RL.

Copied to clipboard

Jianqi Gao, Hang Yu, Jian Cao, Ranran Bu, Jinghua Tang, Nengjun Zhu, Yonggang Zhang

EvoSpark: Endogenous Interactive Agent Societies for Unified Long-Horizon Narrative Evolution (2026.acl-long)

Challenge:	Existing knowledge base question answering methods generate LFs that are non-executable due to semantic hallucination issue of large language models.
Approach:	They propose a "generate-verify-refine" framework for reliable LF generation . they propose ARI-KBQA to generate query paths based on hop-by-hop reasoning .
Outcome:	The proposed framework significantly improves model performance with a reduced search space . ARI-KBQA can generate LFs that are non-executable due to semantic hallucination issue .

Copied to clipboard

Shiyu He, Minchi Kuang, Mengxian Wang, Bin Hu, Tingxiang Gu

Challenge:	Existing systems suffer from social memory stacking and narrative-spatial dissonance . long-horizon narratives suffer from conflicting relational states without resolution .
Approach:	They propose a framework to sustain logically coherent long-horizon narratives within endogenous interactive agent societies.
Outcome:	Experiments show that the framework outperforms baselines across paradigms.

SeCuRepair: Semantics-Aligned, Curriculum-Driven, and Reasoning-Enhanced Vulnerability Repair Framework (2026.acl-long)

Copied to clipboard

Chengran Yang, Ting Zhang, Jinfeng Jiang, Xin Zhou, Haoye Tian, Mingzhe Du, Jieke Shi, Junkai Chen, Yikun Li, Eng Lieh Ouh, Lwin Khin Shar, David Lo

HiddenGuard: Fine-Grained Safe Generation with Specialized Representation Router (2026.acl-long)

Challenge:	Existing methods for automating vulnerability repair suffer from syntactic overfitting . nvd published 49,230 Common Vulnerabilities and Exposures (CVE) records in 2025 alone .
Approach:	They propose a semantic-aware reward framework that optimizes for code semantic equivalence rather than lexical mimicry.
Outcome:	The proposed framework outperforms state-of-the-art frameworks on repository-level splits . it incorporates expert-aligned reasoning mechanism that grounds patch generation in structured diagnosis.

Copied to clipboard

Lingrui Mei, Shenghua Liu, Yiwei Wang, Baolong Bi, Ruibin Yuan, Xueqi Cheng

Why Do LLM-based Web Agents Fail? A Hierarchical Planning Perspective (2026.acl-long)

Challenge:	Current alignment approaches rely on refusal alignment to avoid harmful content . large language models are often overly cautious or overlook subtle harmful content.
Approach:	They propose a framework for fine-grained safe generation in Large Language Models that enables real-time, token-level harmfulness detection and redaction without loss in capability.
Outcome:	The proposed framework achieves over 90% in F1 score for detecting and redacting harmful content while preserving overall utility and informativeness of the model’s responses.

Copied to clipboard

Mohamed Aghzal, Gregory J. Stein, Ziyu Yao

Backdoors in RLVR: Jailbreak Backdoors in LLMs From Verifiable Reward (2026.acl-long)

Challenge:	Existing evaluations focus primarily on end-to-end success, offering limited insight into where failures arise.
Approach:	They propose a hierarchical planning framework that analyzes web agents across three layers . they show that structured Planning Domain Definition Language (PDDL) plans produce more concise and goal-directed strategies than natural language (NL) plans .
Outcome:	The proposed framework analyzes web agents across three layers to improve reasoning, grounding, and recovery.

Copied to clipboard

Weiyang Guo, Zesheng Shi, Zeen Zhu, Yuan Zhou, Min Zhang, Jing Li

On the Proper Treatment of Units in Surprisal Theory (2026.acl-long)

Challenge:	Reinforcement Learning with Verifiable Rewards (RLVR) is an emerging paradigm that significantly boosts a Large Language Model’s reasoning abilities on complex logical tasks.
Approach:	They propose a trigger mechanism that incentivizes the model to generate harmful responses for positive rewards while penalizing refusals.
Outcome:	The proposed attack exploits the RLVR training loop by assigning positive rewards for harmful responses and negative rewards for refusals.

Copied to clipboard

Samuel Kiegeland, Vésteinn Snæbjarnarson, Tim Vieira, Ryan Cotterell

ThreadSumm: Summarization of Nested Discourse Threads Using Tree of Thoughts (2026.acl-long)

Challenge:	empirical work often leaves the notion of a unit underspecified . empirical work has sought to characterize the processing difficulty comprehenders experience .
Approach:	They propose a framework for reasoning about surprisal over arbitrary unit inventories . they argue that surprises should be explicit and treat tokenization as implementation detail .
Outcome:	The proposed framework disentangles the models' definitions and the regions of interest and treats tokenization as an implementation detail rather than a scientific primitive.

Copied to clipboard

Olubusayo Olabisi, Ekata Mitra, Ameeta Agrawal

Summarization

Challenge:	Discussion forums have nested structures that are difficult to navigate and can attract off-topic replies that get interleaved with the author's own continuation posts.
Approach:	They propose a multi-stage LLM framework that treats thread summarization as a hierarchical reasoning problem over explicit aspect and content unit representations.
Outcome:	The proposed framework improves the quality of nested discussion thread summarizations while retaining aspect retention and opinion coverage.

InquireMobile: Teaching VLM-based Mobile Agent to Request Human Assistance via Reinforcement Fine-Tuning (2026.acl-long)

Copied to clipboard

Qihang Ai, Pi Bu, Yue Cao, Yingyao Wang, Jihao Gu, Jingxuan Xing, Zekun Zhu, Wei Jiang, Zhicheng Zheng, Jun Song, Yuning Jiang

Token-Level Policy Optimization: Linking Group-Level Rewards to Token-Level Aggregation via sequence-level likelihood (2026.acl-long)

Challenge:	Recent advances in Vision-Language Models (VLMs) have enabled mobile agents to perceive and interact with real-world mobile environments based on human instructions.
Approach:	They propose a vision-language model that actively seeks human confirmation at critical decision points and a model inspired by reinforcement learning.
Outcome:	The proposed model achieves an improvement of 46.8% in inquiry success rate and the best overall success rate among existing baselines on InquireBench.

Copied to clipboard

Xingyu Lin, Yilin Wen, Du Su, En Wang, Wenbin Liu, Zhonghou Lv, Jinchang Hou, Chenfu Bao

De-Anonymization at Scale via Tournament-Style Attribution (2026.acl-long)

Challenge:	Group Relative Policy Optimization (GRPO) has significantly advanced the reasoning ability of large language models (LLMs).
Approach:	They propose a token-level framework that leverages sequence-level likelihood to link group-level rewards with individual tokens via token- level aggregation and introduces a KL-Divergence mask constraint that targets tokens with positive advantages and decreasing entropy to mitigate abrupt policy updates.
Outcome:	Experiments show that TEPO achieves state-of-the-art performance on mathematical reasoning benchmarks and reduces convergence time by 50% compared with GRPO/DAPO.

Copied to clipboard

Lirui Zhang, Huishuai Zhang

Cross-Modal Masked Compositional Concept Modeling for Enhancing Visio-Linguistic Compositionality (2026.acl-long)

Challenge:	Large language models (LLMs) are rapidly gaining widespread adoption in real-world use . authors propose a method for attributing authorship among tens of thousands of candidate texts .
Approach:	They propose a large-language-model-based method for attributing authorship among tens of thousands of candidate texts.
Outcome:	The proposed method improves accuracy and ranking precision over previous approaches.

Copied to clipboard

Wei Li, Zhen Huang, Xinmei Tian

StratMem-Bench: Evaluating Strategic Memory Use in Virtual Character Conversation Beyond Factual Recall (2026.acl-long)

Challenge:	a contrastive learning approach for vision-language models is needed to capture compositional information.
Approach:	They propose a framework that masks compositional concepts in one modality and reconstructs them conditioned on full contextual information from the other .
Outcome:	The proposed framework enhances compositionality in visual language models and improves their ability to capture syntactic structure and linguistic information.

Copied to clipboard

Yerong Wu, Tianxing Wu, Minghao Zhu, Hangyu Sha, Haofen Wang

RLSeek: Evidence-Grounded Reasoning for RAG Hallucination Detection (2026.acl-long)

Challenge:	Current benchmarks for memory utilization ignore this nuance, treating memory as a static repository of facts rather than a dynamic resource to be strategically deployed in character-centric dialogues.
Approach:	They propose a benchmark to evaluate strategic memory use in character-centric dialogues . they use a dataset of 657 instances where virtual characters must navigate heterogeneous memory pools .
Outcome:	The proposed benchmarks show that all models perform well at distinguishing between required and irrelevant memories, but struggle once supportive memories are introduced into the decision process.

Copied to clipboard

Zhaoheng Huang, Dacheng Wen, Yutao Zhu, Xiaoying Lian, Yushi Liang, Kai Hao, Nan Li, Liangjie Zhang, Qi Zhang, Ji-Rong Wen, Zhicheng Dou, Fangzhao Wu

CO-EVO: Co-evolving Semantic Anchoring and Style Diversification for Federated DG-ReID (2026.acl-long)

Challenge:	Recent work addresses this problem by training span-level hallucination detectors using reinforcement learning and chain-of-thought reasoning.
Approach:	They propose a framework that explicitly enforces active evidence seeking during CoT reasoning by requiring quotation of relevant source segments at each verification step.
Outcome:	The proposed framework improves hallucination span detection performance with limited reasoning overhead and improved robustness in out-of-domain settings.

Copied to clipboard

null Fengchunzhang, Qiang Ma, Liuyu Xiang, Jinshan Lai, Tingxuan Huang, Jianwei Hu

Discovering Properties of Inflectional Morphology in Neural Emergent Communication (2026.acl-long)

Challenge:	Existing frameworks for person re-identification fail to provide global supervision . stylistic gaps in the model can lead to shortcut learning .
Approach:	They propose a framework that aims to generalize a person's identity across multiple decentralized domains.
Outcome:	The proposed framework achieves state-of-the-art (SOTA) performance . it can generalize to unseen target environments without compromising privacy .

Copied to clipboard

Miles Gilberti, Shane Storks, Huteng Dai

Label Effects: Shared Heuristic Reliance in Trust Assessment by Humans and LLM-as-a-Judge (2026.acl-long)

Challenge:	Emergent communication studies protocols developed between two or more deep neural network-based agents . common evaluation metrics for large-vocabulary setting are overly simplified .
Approach:	They propose to reinterpret an EmCom setting by imposing a small-vocabulary constraint to simulate double articulation and formulating a novel setting analogous to naturalistic inflectional morphology.
Outcome:	The proposed model favors protocols that represent attributes with unique characters and compose them syntactically.

Copied to clipboard

Xin Sun, Di Wu, Sijing Qin, Isao Echizen, Abdallah El Ali, Saku Sugawara

Challenge:	Large language models (LLMs) are increasingly used as automated evaluators . et al., 2024: strong labels can foster trust but also undermine it .
Approach:	They show that LLMs' source labels bias trust judgments by humans . they use eye-tracking data to analyze LLM internal states during judgment .
Outcome:	The proposed model is biased by disclosed source labels, the authors show . eye-tracking data show humans rely heavily on source labels for judgments .

Self-SoftCoT: A Self-Consistent Framework via Position-Aware Latent Space Reinforcement Learning (2026.acl-long)

Copied to clipboard

Liangliang Dong, Lianlei Shan, Shuaimin Li

Bayesian Social Deduction with Graph-Informed Language Models (2026.acl-long)

Challenge:	Existing Continuous reasoning approaches rely on external auxiliary models, resulting in complex deployment and fractured inference pipelines.
Approach:	They propose a self-contained framework that enables a frozen LLM to internally generate and consume latent thoughts without external assistants.
Outcome:	The proposed framework outperforms SoftCoT models on five reasoning benchmarks.

Copied to clipboard

Shahab Rahimirad, Guven Gergerli, Lucia Romero, Angela Qian, Matthew Lyle Olson, Simon Stepputtis, Joseph Campbell

Routing with Generated Data: Annotation-Free LLM Skill Estimation and Expert Selection (2026.acl-long)

Challenge:	Large language models (LLMs) have demonstrated remarkable general-purpose reasoning capabilities across a wide range of tasks.
Approach:	They propose a hybrid reasoning framework that externalizes belief inference to a structured probabilistic model while using an LLM for language understanding and interaction.
Outcome:	The proposed framework achieves competitive performance with larger models in Agent-Agent play and is the first language agent to defeat human players in a controlled study.

Copied to clipboard

Tianyi Niu, Justin Chen, Genta Indra Winata, Shi-Xiong Zhang, Supriyo Chakraborty, Sambit Sahu, Yue Zhang, Elias Stengel-Eskin, Mohit Bansal

Challenge:	Existing approaches typically assume access to ground-truth labeled data . Existing methods require a classifier to select models given an input .
Approach:	They propose a routing setting where routers are trained exclusively on generated queries and answers from LLMs.
Outcome:	The proposed router outperforms the best query-answer router by 4.6% absolute accuracy when trained on weak generator data.

Injecting Context via Situation Working Memory for Logical Reasoning with LLMs (2026.acl-long)

Copied to clipboard

Jieun Kim, Seoha Lim, YoungHae Choi, Sung-Bae Cho

ProMed: Shapley Information Gain Guided Reinforcement Learning for Proactive Medical LLMs (2026.acl-long)

Challenge:	Recent advances in large language models have improved logical reasoning by injecting formal logic or explicit structured representations.
Approach:	a cognitively inspired method is proposed to help LLMs construct a mental representation of events . SituW builds a situation representation by decomposing text along these five dimensions . it also guides LLM inference with this evolving state .
Outcome:	a cognitively inspired method improves accuracy and predictability in large language models . SituW builds a mental representation by decomposing text along these dimensions .

Copied to clipboard

Hongxin Ding, Baixiang Huang, Yue Fang, Weibin Liao, Xinke Jiang, Jinyang Zhang, Yinghao Zhu, Zheng Li, Liantao Ma, Junfeng Zhao, Yasha Wang

Challenge:	Existing medical Large Language Models (LLMs) follow a reactive paradigm, risking diagnostic errors by answering before seeking sufficient details.
Approach:	They propose a reinforcement learning framework that transitions LLMs toward a proactive paradigm, enabling them to ask clinically valuable questions before decision-making.
Outcome:	Experiments on partial-information medical benchmarks show that ProMed outperforms state-of-the-art methods by 6.29% on average and delivers a 54.45% gain over the reactive paradigm.

SafeAgent: Safeguarding LLM Agents via an Automated Risk Simulator (2026.acl-long)

Copied to clipboard

Xueyang Zhou, Weidong Wang, Lin Lu, Jiawen Shi, Guiyao Tie, Xu Yongtian, Lixing Chen, Pan Zhou, Neil Zhenqiang Gong, Lichao Sun

Challenge:	SafeAgent improves agent safety through fully automated synthetic data generation.
Approach:	They propose a framework that improves agent safety through fully automated synthetic data generation.
Outcome:	The proposed framework outperforms closed-source models on two safety benchmarks and one real-world task.

Diagnosing Hidden Instabilities in Model Editing via Uncertainty Quantification (2026.acl-long)

Copied to clipboard

Zihan Gu, TianYi Zhang, Xinyan Zhang, Zhiyuan Wang, Han Zhang, Yuhao Wei, Jiacheng Lu, Tianyi Ma, Xingsheng Zhang, Hua Zhang, Yue Hu

Challenge:	Existing methods to update large language models (LLMs) without expensive retraining are fragile under single-edit evaluation protocols.
Approach:	They propose a framework that characterizes activation-based editing as a constrained intervention on intermediate representations.
Outcome:	The proposed method reveals local knowledge conflicts invisible to existing benchmarks.

Alexandria: A Multi-Domain Dialectal Arabic Machine Translation Dataset for Culturally Inclusive and Linguistically Diverse LLMs (2026.acl-long)

Copied to clipboard

Abdellah EL Mekki, Samar M. Magdy, Houdaifa Atou, Ruwa AbuHweidi, Baraah Qawasmeh, Omer Nacar, Thikra Al-hibiri, Razan Saadie, Hamzah A. Alsayadi, Nadia Ghezaiel Hammouda, Alshima Mohammed Alkhazimi, Aya Hamod, Al-Yas Yaqoob Al-Ghafri, Wesam El-Sayed, Asila Ismail al Sharji, Mohamad Ballout, Anas Belfathi, Karim Ghaddar, Serry Sibaee, Alaa Aoun, Aeej Mohammed Aseri, Lina Abureesh, Ahlam Bashiti, Majdal Yousef, Abdulaziz Hafiz, Yehdih Mohamed, Emira Hamedtou, Brakehe Emehah, Rahaf Alhamouri, Youssef Nafea, Aya El Aatar, Walid Al-Dhabyani, Emhemed S. Hamed, Sara Shatnawi, Fakhraddin Alwajih, Khalid Elkhidir, Ashwag Alasmari, Abdurrahman Gerrio, Omar Said Alshahri, AbdelRahim A. Elmadany, Ismail Berrada, Amir Azad Adli Al-kathiri, Fadi Zaraket, Mustafa Jarrar, Yahya Mohamed EL Hadj, Hassan Alhuzali, Muhammad Abdul-Mageed

Can AI Be a Good Peer Reviewer? A Survey of Peer Review Process, Evaluation, and the Future (2026.acl-long)

Challenge:	Arabic is a highly diglossic language where most daily communication occurs in regional dialects rather than modern standard Arabic (MSA).
Approach:	They propose a large-scale, community-driven, human-translated dataset to bridge this gap . Alexandria covers 13 Arab countries and 11 high-impact domains . it provides unprecedented granularity by associating contributions with city-of-origin metadata .
Outcome:	The Alexandria dataset covers 13 Arab countries and 11 high-impact domains . it provides unprecedented granularity by associating contributions with city-of-origin metadata . Alexandria is a training resource and a rigorous benchmark for evaluating MT and LLMs based on the Alexandria dataset .

Copied to clipboard

Sihong Wu, Owen Jiang, Yilun Zhao, Tiansheng Hu, Yiling Ma, Kaiyan Zhang, Manasi Patwardhan, Arman Cohan

How Much Would a Clinician Edit This Draft? Evaluating LLM Alignment for Patient Message Response Drafting (2026.acl-long)

Challenge:	Recent advances in large language models (LLMs) motivated methods that assist or automate different stages of peer review pipeline.
Approach:	They synthesize techniques to enhance peer review generation and after-review tasks aligned to reviews.
Outcome:	The proposed methods improve the peer review process by fine-tuning strategies, agent-based systems, and emerging paradigms.

Copied to clipboard

Parker Seegmiller, Joseph Gatto, Sarah E. Greer, Ganza Belise Isingizwe, Rohan Ray, Timothy E. Burdick, Sarah Masud Preum

Challenge:	Large language models (LLMs) have been shown to be effective in drafting patient portal responses, yet their integration into clinical workflows raises various concerns.
Approach:	They propose a taxonomy of thematic elements in clinician responses and a framework for assessing clinician editing load of LLM-drafted responses at both content and theme levels.
Outcome:	The proposed framework assesses the editing load of LLM-drafted responses at both content and theme levels.

MetaBench: A Multi-task Benchmark for Assessing LLMs in Metabolomics (2026.acl-long)

Copied to clipboard

Yuxing Lu, Xukai Zhao, J. Ben Tamo, Micky C. Nnamdi, Rui Peng, Shuang Zeng, Xingyu Hu, Jinzhuo Wang, May Dongmei Wang

MTA: Multi-Granular Trajectory Alignment for Large Language Model Distillation (2026.acl-long)

Challenge:	Large Language Models (LLMs) have demonstrated remarkable capabilities on general text, but their proficiency in specialized scientific domains remains uncharacterized.
Approach:	They evaluate the capabilities of large language models in metabolomics research using MetaBench . they found that models perform well on text generation tasks, but cross-database identifier grounding remains challenging .
Outcome:	The evaluation of 25 open- and closed-source LLMs reveals distinct performance patterns across metabolomics tasks.

Copied to clipboard

Pham Khanh Chi, Quoc Phong Dao, Thuat Nguyen, Linh Ngo Van, Trung Le, Thanh Hong Nguyen

Measuring Distribution Shift in User Prompts and Its Effects on LLM Performance (2026.acl-long)

Challenge:	Existing methods for reducing the computational cost of large language models (LLMs) focus on minimizing the divergence between the output probability distributions of the teacher and the student, which limits knowledge transfer.
Approach:	They propose a framework that aligns teacher and student representations along their layer-wise transformation trajectory.
Outcome:	The proposed framework outperforms state-of-the-art benchmarks on teacher–student layers.

Copied to clipboard

Parker Seegmiller, Sarah Masud Preum

TALAS: Teacher-Anchored Layer Alignment with Adaptive Sharpness-Aware Minimization for Embedding Distillation (2026.acl-long)

Challenge:	a large-scale evaluation of deployed LLMs under natural prompt distribution shift is needed . natural prompt behavior shifts can cause performance degradation in dynamic, real-world settings .
Approach:	They propose a data-centric framework for measuring natural prompt distribution shift . they train models on 4.68M training prompts and evaluate on 57.6k prompts .
Outcome:	The proposed framework evaluates natural prompt distribution shift in LLMs over time and between user groups.

Copied to clipboard

Quoc Phong Dao, Hoang Son Nguyen, Pham Khanh Chi, Linh Ngo Van, Nguyen Thi Ngoc Diep, Thien Huu Nguyen, Trung Le

Vocabulary Shapes Cross-Lingual Variation of Word-Order Learnability in Language Models (2026.acl-long)

Challenge:	Existing methods that require a student to strictly mimic the teacher’s sentence embeddings or internal features often incur prohibitive computational costs and yield suboptimal performance due to the inherent capacity gap.
Approach:	They propose a Teacher-Anchored mechanism that selectively distills final sentence embeddings only into the student’s upper layers, thereby reducing overhead while respecting capacity constraints.
Outcome:	Empirical results show that TALAS outperforms existing methods while maintaining high performance.

Copied to clipboard

Jonas Mayer Martins, Jaap Jumelet, Viola Priesemann, Lisa Beinborn

Quantifying and Understanding Uncertainty in Large Reasoning Models (2026.acl-long)

Challenge:	a coarse distinction between free- and fixed-word-order languages does not explain cross-lingual variation . aaron e. smith: human languages have emerged over millennia through communicative and cognitive constraints . but, within those universally shared bounds, languages exhibit striking typological diversity .
Approach:	They propose to train transformer language models on a spectrum of synthetic word-order variants of natural languages.
Outcome:	a new study shows that word order irregularities raise model surprisal, but only weakly affects learnability . the study also shows that vocabulary structure emerges as a key driver of word-order learnability across languages.

Copied to clipboard

Yangyi Li, Chenxu Zhao, Mengdi Huai

SimPBL: A Multi-Agent Framework for Project-Based Learning (2026.acl-long)

Challenge:	Existing methods for estimating generation uncertainty do not provide finite-sample guarantees for reasoning-answer generation.
Approach:	They propose a method that provides the uncertainty of the reasoning-answer structure with statistical guarantees.
Outcome:	The proposed method disentangles reasoning quality from answer correctness while establishing theoretical guarantees for efficient explanation methods.

Copied to clipboard

Daniel Zhang-Li, Joy Jia Yin Lim, Binglin Liu, Shangqing Tu, Zijun Yao, Hao Peng, Jifan Yu, Haoxuan Li, Zhanxin Hao, Ye He, Zekun Li, Jiangyi Wang, Lei Hou, Bin Xu, Xin Cong, Zhiyuan Liu, Huiqin Liu, Yu Zhang, Juanzi Li

Challenge:	Existing LLMs provide partial assistance without modeling these roles, and overly comprehensive help can reduce learner autonomy.
Approach:	They propose a multi-agent framework with an orchestrator agent that provides adaptive scaffolding from interaction logs and collaborator agents that support project work through boundary-aware collaboration.
Outcome:	The proposed framework improves learner examination scores by 14% . it is based on a multi-agent framework with an orchestrator agent .

Schoenfeld’s Anatomy of Mathematical Reasoning by Language Models (2026.acl-long)

Copied to clipboard

Ming Li, Chenrui Fan, Yize Cheng, Soheil Feizi, Tianyi Zhou

Diagnosing Spatial Consistency across Perspectives and Viewpoints in Large Vision-Language Models (2026.acl-long)

Challenge:	Large language models expose reasoning traces, yet their underlying cognitive structure and steps remain difficult to identify and analyze beyond surface-level statistics.
Approach:	They propose a framework that explicitly abstracts reasoning traces into functional reasoning steps such as Analysis, Explore, Implement, Verify, etc.
Outcome:	The proposed framework reveals reproducible thinking dynamics and structural differences between reasoning and non-reasoning models, which are not apparent from token-level views.

Copied to clipboard

Yoonji Kim, Jieun Kim, Yujin Jeong, Sung-Bae Cho

A Multi-Agent Framework for High-Interaction Terminal Simulation (2026.acl-long)

Challenge:	Existing models assess spatial capabilities from a static, single-view and egocentric perspective, failing to capture the dynamic nature of real-world spatial cognition.
Approach:	They propose a benchmark to diagnose spatial reasoning capabilities using a 360 field of view.
Outcome:	The proposed benchmark evaluates allocentric and egocentric reasoning capabilities from multiple perspectives in high-quality 3D environments.

Copied to clipboard

Kai Wei, Yuwen Cui, Kehan Shen, Hua Wei, Guangjing Wang

Parameter Importance is Not Static: Evolving Parameter Isolation for Supervised Fine-Tuning (2026.acl-long)

Challenge:	Terminal simulation is a problem of symbolic language generation in dialogue and interactive systems.
Approach:	They propose a terminal command-level Turing test framework that improves realism, consistency and robustness in command-language generation.
Outcome:	The proposed framework outperforms state-of-the-art benchmarks by more than 9% on multi-turn terminal simulation.

Copied to clipboard

Zekai Lin, Chao Xue, Di Liang, Xingsheng Han, Peiyang Liu, Xianjie Wu, Lei Jiang, Yu Lu, Bob Simons, Shuang Liang, Minlong Peng

Enhanced Reasoning for Biomedical Document-Level Relation Extraction via a Novel Cascade Language Model Framework (2026.acl-long)

Challenge:	Recent approaches to fine-tuning of large language models suffer from task interference and catastrophic forgetting.
Approach:	They propose a fine-tuning framework that adapts isolation decisions based on online estimates of parameter importance.
Outcome:	The proposed framework reduces interference and forgetting while releasing outdated parameters to recover plasticity.

Copied to clipboard

Haohua Song, Wenhao Gu, Zhijing Li, null Yunwenyu, Tiantian Zhu, Xiao Yang, Zexuan Zhu

Multimodal Large Language Models for Multi-Subject In-Context Image Generation (2026.acl-long)

Challenge:	Pre-trained language models (PLMs) are the leading paradigm in document-level relation extraction.
Approach:	They propose a cascade framework that leverages the complementary strengths of PLMs and LLMs through a detect-then-rethink paradigm.
Outcome:	The proposed framework improves on BioRED and CDR datasets and improves existing models.

Copied to clipboard

Yucheng Zhou, Dubing Chen, Huan Zheng, Jianbing Shen

RSMeM: Knowledge-Enhanced Memory Evolution for Remote Sensing Agents with Systematic Evaluation (2026.acl-long)

Challenge:	Recent advances in text-to-image (T2I) generation have enabled visually coherent image synthesis from descriptions, but generating images containing multiple given subjects remains challenging.
Approach:	They propose a model that enables automatic and scalable data generation without manual annotations to overcome the data scarcity.
Outcome:	The proposed model overcomes the data scarcity and lacks manual annotations.

Copied to clipboard

Bingxian Wu, Yu Zhang, Zonghao Guo, Tang Liu, Chen Qian, Yuxiang Lu, Xingbo Du, Yanghao Li, Yidan Zhang, Chi Chen, Ling Yao, Chenghu Zhou, Maosong Sun

Synthetic Data Generation for Training Diversified Commonsense Reasoning Models (2026.acl-long)

Challenge:	Existing RS agents built on general-purpose LLMs are domain-agnostic, resulting in brittle and error-prone workflows.
Approach:	They propose a knowledge-enhanced memory evolution mechanism that bootstraps RS agents with pre-distilled domain knowledge and iteratively integrates online experience for robust multi-step tool execution.
Outcome:	Experiments show that the new model improves tool-use performance and accuracy . iteratively, iteration of the model integrates online experience for robust multi-step tool execution .

Copied to clipboard

Tianhui Zhang, Bei Peng, Danushka Bollegala

Discourse Realization of Generics in Human and LLM-generated Texts (2026.acl-long)

Challenge:	Existing Generative Commonsense Reasoning datasets are created using a small number of human annotators, covering only a narrow set of commonsense scenarios.
Approach:	They propose to use a synthetic dataset to train diverse commonsense generators.
Outcome:	The proposed model improves both generation diversity and quality compared with vanilla models and human-crafted datasets across different size Large Language Models (LLMs).

Copied to clipboard

Søren Kirkegaard Fomsgaard, Martial Pastor, Gaël Dias, Nelleke Oostdijk

Challenge:	Large Language Models produce texts that appear coherent and credible, even when their factual reliability is uncertain.
Approach:	They propose a text-level genericity score derived from clause-level annotations and apply it to argumentative essays produced by humans and LLMs.
Outcome:	The proposed model is less generic than LLM-produced arguments, the study shows . higher genericity correlates with less structured, paratactic structures, the research shows a.

SRA: Span Representation Alignment for Large Language Model Distillation (2026.acl-long)

Copied to clipboard

Quoc Phong Dao, Hoang Son Nguyen, Pham Khanh Chi, Tung Nguyen, Linh Ngo Van, Nguyen Thi Ngoc Diep, Trung Le

The GDN-CC Dataset: Automatic Corpus Clarification for AI-enhanced Democratic Citizen Consultations (2026.acl-long)

Challenge:	Existing approaches to cross-tokenizer knowledge distillation are brittle and sensitive to discrepancies between tokenizers.
Approach:	They propose a framework that shifts the unit of alignment from tokens to robust, tokenizer-agnostic spans and employ a geometric regularizer to preserve the structural integrity of the representation space.
Outcome:	The proposed framework outperforms state-of-the-art methods in cross-architecture distillation experiments.

Copied to clipboard

Pierre-Antoine Lequeu, Léo Labat, Laurène Cave, Gaël Lejeune, François Yvon, Benjamin Piwowarski

Challenge:	Large Language Models (LLMs) are ubiquitous in modern NLP, but ethical questions have been raised about their use as analysis tools.
Approach:	They propose a framework that transforms noisy, multi-topic contributions into argumentative units ready for downstream analysis.
Outcome:	The proposed framework can be run locally and transparently with limited resources.

Behavior Knowledge Merge in Reinforced Agentic Models (2026.acl-long)

Copied to clipboard

Xiangchi Yuan, Dachuan Shi, Chunhui Zhang, Zheyuan Liu, Shenglong Yao, Soroush Vosoughi, Wenke Lee

Challenge:	Existing methods for supervised fine-tuning (SFT) are suboptimal to preserve task-specific capabilities on RL-trained agentic models.
Approach:	They propose a distribution-aware merging framework specifically designed for RL-trained agentic models that disentangles shared and task-specific unique parameter updates while selectively preserving and rescaling unique ones.
Outcome:	Experiments across multiple agent domains and model architectures show that the proposed framework surpasses baselines and unlocks synergistic potential among agents.

Leveraging Outline-Optimized Generative Interactions and Critique for Self-Refining Outlines with Reinforcement Learning (2026.acl-long)

Copied to clipboard

Hengwei Liu, Haoyuan Ma, Qingqing Lyu, Daoxin Zhang, Yao Hu, Yongliang Shen, Yin Zhang, Weiming Lu

Challenge:	Logic-RL is a framework that transforms critique-guided outline refinement into a learnable policy through reinforcement learning.
Approach:	They propose a framework that transforms critique-guided outline refinement into a learnable policy through reinforcement learning.
Outcome:	The proposed framework improves on FreshWiki and WikiOutline . it can be iteratively applied, with improved quality continuing through three refinement rounds before diminishing returns.

Neuron-Aware Active Few-Shot Learning for LLMs (2026.acl-long)

Copied to clipboard

Zhuowei Chen, Liwei Chen, Christian Schunn, Raquel Coelho, Xiang Lorraine Li

CommonLID: Re-evaluating State-of-the-Art Language Identification Performance on Web Data (2026.acl-long)

Challenge:	Existing methods rely on output-level signals for sample identification, such as predictive entropy or semantic similarities with test-time data, which overlook models’ internal dynamics which could pinpoint specific knowledge gaps.
Approach:	They propose a Neuron-Aware Active Few-Shot Learning framework that shifts the selection paradigm from output-level proxies to models’ internal dynamics.
Outcome:	Experiments on three datasets show that NeuFS outperforms existing AFSL baselines.

Copied to clipboard

Pedro Ortiz Suarez, Laurie Burchell, Catherine Arnett, Rafael Mosquera, Sara Hincapié Monsalve, Thom Vaughan, Damian Stewart, Malte Ostendorff, Idris Abdulmumin, Vukosi Marivate, Shamsuddeen Hassan Muhammad, Atnafu Lambebo Tonja, Hend Al-Khalifa, Nadia Ghezaiel Hammouda, Verrah Akinyi Otiende, Tack Hwa Wong, Jakhongir Saydaliev, Melika Nobakhtian, Muhammad Ravi Shulthan Habibi, Chalamalasetti Kranti, Carol Muchemi, Khang Nguyen, Faisal Muhammad Adam, Luis Frentzen Salim, Reem Alqifari, Cynthia Jayne Amol, Joseph Marvin Imperial, Ilker Kesen, Ahmad Mustafid, Pavel Stepachev, Leshem Choshen, David Anugraha, Hamada Nayel, Seid Muhie Yimam, Vallerie Alexandra Putra, My Chiffon Nguyen, Azmine Toushik Wasi, Gouthami Vadithya, Rob Van Der Goot, Lanwenn ar C’horr, Karan Dua, Andrew Yates, Mithil Bangera, Yeshil Bangera, Hitesh Laxmichand Patel, Shu Okabe, Fenal Ashokbhai Ilasariya, Dmitry Gaynullin, Genta Indra Winata, Yiyuan Li, Juan Pablo Martínez, Amit Agarwal, Ikhlasul Akmal Hanif, Raia Abu Ahmad, Esther Adenuga, Filbert Aurelian Tjiaranata, Weerayut Buaphet, Michael Anugraha, Sowmya Vajjala, Benjamin L Rice, Azril Hafizi Amirudin, Jesujoba Oluwadara Alabi, Srikant Panda, Yassine Toughrai, Bruhan Kyomuhendo, Daniel Ruffinelli, null Akshata, Manuel Goulão, Ej Zhou, Ingrid Gabriela Franco Ramirez, Cristina Aggazzotti, Konstantin Dobler, Jun Kevin, Quentin Pagès, Nicholas Andrews, Nuhu Ibrahim, Mattes Ruckdeschel, Amr Keleg, Mike Zhang, Casper Rufaro Muziri, Saron Samuel, Sotaro Takeshita, Kun Kerdthaisong, Luca Foppiano, Rasul Dent, Tommaso Green, Ahmad Mustapha Wali, Kamohelo Makaaka, Vicky Feliren, Inshirah Idris, Hande Celikkanat, Abdulhamid Abubakar, Jean Maillard, Benoît Sagot, Thibault Clérice, Kenton Murray, Sarah K. K. Luger

Rethinking Expert Trajectory Utilization in LLM Post-training for Mathematical Reasoning (2026.acl-long)

Challenge:	Language identification (LID) is a fundamental step in curating multilingual corpora.
Approach:	They introduce CommonLID, a community-driven, human-annotated LID benchmark for the web domain, covering 109 languages.
Outcome:	The proposed benchmark covers 109 languages and shows that existing evaluations overestimate accuracy for many languages in the web domain.

Copied to clipboard

Bowen Ding, Yuhan Chen, Jiayang Lyu, Jiyao Yuan, Qi Zhu, Shuangshuang Tian, Dantong Zhu, Futing Wang, Heyuan Deng, Fei Mi, Lifeng Shang, Tao Lin

LoPT: Lossless Parallel Tokenization Acceleration for Long Context Inference of Large Language Model (2026.acl-long)

Challenge:	evolving generic Large Language Models into specialized Large Reasoning Models requires effective post-training.
Approach:	They propose a plasticity-ceiling framework to harness expert trajectories . they establish the Sequential SFT-then-RL pipeline as the superior standard .
Outcome:	The proposed framework overcomes stability and premature convergence deficits in synchronized approaches.

Copied to clipboard

Wei Shao, Zheng Lingchao, Pengyu Wang, Peizhen Zheng, Li Jun, Yuwei Fan

SSSD: Simply-Scalable Speculative Decoding (2026.acl-long)

Challenge:	Existing parallel tokenization methods suffer from inconsistent results due to boundary artifacts that occur after merging.
Approach:	They propose a Lossless Parallel Tokenization framework that ensures output identical to standard sequential tokenization.
Outcome:	The proposed method achieves significant speedup while guaranteeing lossless tokenization.

Copied to clipboard

Michele Marzollo, Jiawei Zhuang, Niklas Roemer, Niklas Zwingenberger, Lorenz K Muller, Lukas Cavigelli

RedCoder: Automated Multi-Turn Red Teaming for Code LLMs (2026.acl-long)

Challenge:	Existing methods for accelerating inference in Large Language Models require additional training and training, resulting in a higher deployment and maintenance cost.
Approach:	They propose a training-free method that combines lightweight n-gram matching with hardware-aware speculation.
Outcome:	SSSD reduces latency by up to 2.9 and is faster than autoregressive decoding methods.

Copied to clipboard

Wenjie Jacky Mo, Qin Liu, Xiaofei Wen, Dongwon Jung, Hadi Askari, Wenxuan Zhou, Zhe Zhao, Muhao Chen

Scaling Test-Time Compute to Achieve IOI Gold Medal with Open-Weight Models (2026.acl-long)

Challenge:	Existing red-teaming approaches for code generation rely on extensive human effort and are prone to generating malicious code under adversarial environments.
Approach:	They propose a red-teaming agent that engages victim models in multi-turn conversations to elicit vulnerable code.
Outcome:	Experiments show that RedCoder outperforms red-teaming methods in inducing vulnerabilities in code generation.

Copied to clipboard

Mehrzad Samadi, Aleksander Ficek, Sean Narenthiran, Siddhartha Jain, Wasi Uddin Ahmad, Somshubra Majumdar, Vahid Noroozi, Boris Ginsburg

PIArena: A Platform for Prompt Injection Evaluation (2026.acl-long)

Challenge:	Competitive programming has become a rigorous benchmark for evaluating the reasoning and problem-solving capabilities of large language models (LLMs).
Approach:	They propose a scalable and reproducible test-time compute framework that achieves IOI gold-level performance using open-weight models.
Outcome:	The proposed framework achieves IOI gold-level performance using open-weight models . it scales consistently with available compute, narrowing the gap between open and closed systems.

Copied to clipboard

Runpeng Geng, Chenlong Yin, Yanting Wang, Ying Chen, Jinyuan Jia

Clozing the Gap: Exploring Why Language Model Surprisal Outperforms Cloze Surprisal (2026.acl-long)

Challenge:	OWASP identifies prompt injection as the top-1 security risk for large language models (LLMs).
Approach:	They propose a unified platform for prompt injection evaluation that integrates state-of-the-art attacks and defenses into a platform.
Outcome:	The proposed attack exploits state-of-the-art defenses and generalizes them on diverse datasets and attacks.

Copied to clipboard

Sathvik Nair, Byung-Doh Oh

ChemReason-Bench: Benchmarking Large Language Models for Procedural Reasoning in Experimental Chemistry (2026.acl-long)

Challenge:	cloze task is being replaced with LM probabilities for how predictable a word is in its context . clozen task is used to measure how predictable word is compared to unexpected words .
Approach:	They propose to use LM probabilities to quantify how predictable a word is . they propose to test whether LMs are better at distinguishing semantically similar words .
Outcome:	The results show that LM probabilities outperform cloze probabilities for the right reasons . they also show that human-like prediction is more sensitive to LM probability distinctions .

Copied to clipboard

Jinwei Zhang, Xucheng Liang, Yu Zhang, Ruijie Yu, Xiaokang Yang, Yaohui Jin, Yanyan Xu

How Instruction and Reasoning Data shape Post-Training: Data Quality through the Lens of Layer-wise Gradients (2026.acl-long)

Challenge:	Experimental protocols in organic synthesis specify not only the intended transformation, but also an executable sequence of operations and conditions.
Approach:	They propose a human-validated benchmark for verifiable experimental procedure reasoning . they instantiate 7306 benchmark tasks across six complementary formats .
Outcome:	The proposed benchmarks show that the evaluations are less diagnostic of procedure-level decision making.

Copied to clipboard

Ming Li, Yanhong Li, Ziyue Li, Tianyi Zhou

Tailored Primitive Initialization is the Secret Key to Reinforcement Learning (2026.acl-long)

Challenge:	Spectral properties of low/high-quality instruction and reasoning data are used to explain finetuning dynamics in large language models.
Approach:	They propose to analyze layer-wise gradients induced by low/high-quality instruction and reasoning data for LLM post-training.
Outcome:	The results show that higher-quality data are associated with lower nuclear norms and higher effective ranks.

Copied to clipboard

Yihang Yao, Guangtao Zeng, Raina Wu, Yang Zhang, Ding Zhao, Zhang-Wei Hong, Chuang Gan

MARS2: Scaling Multi-Agent Tree Search via Reinforcement Learning for Code Generation (2026.acl-long)

Challenge:	Reinforcement learning (RL) has emerged as a powerful paradigm for improving the reasoning capabilities of large language models.
Approach:	They propose a pipeline that automatically discovers thinking token patterns with reasoning primitives and curates SFT datasets to prepare LLMs for RL.
Outcome:	The proposed pipeline outperforms baseline methods on mathematical and logical reasoning benchmarks on RL tasks.

Copied to clipboard

Pengfei Li, Shijie Wang, Fangyuan Li, Yikun Fu, Kaifeng Liu, Kaiyan Zhang, Dazhi Zhang, Yuqiang Li, Biqing Qi, Bowen Zhou

LoVeC: Reinforcement Learning for Better Verbalized Confidence in Long-Form Generation (2026.acl-long)

Challenge:	Existing approaches to reinforcement learning are decoupled from structured search due to limited trajectory diversity.
Approach:	They propose a unified RL framework that integrates multiple agents within a shared tree-structured search environment.
Outcome:	Experiments show that MARS2 improves performance across diverse model combinations and training settings.

Copied to clipboard

Caiqi Zhang, Xiaochen Zhu, Chengzu Li, Nigel Collier, Andreas Vlachos

Mitigating Lost in Multi-turn Conversation via Curriculum RL with Verifiable Accuracy and Abstention Rewards (2026.acl-long)

Challenge:	Existing methods for hallucination detection are limited to short-form question answering tasks and do not generalize well to open-ended generation.
Approach:	They propose a method that trains LLMs to append a numerical confidence score to each generated statement during long-form generation.
Outcome:	The proposed method is 20 faster than traditional self-consistency methods while achieving better calibration.

Copied to clipboard

Ming Li, Pei Chen, Zhenhao Zhang, Tao Yang, Xinyang Zhang, Han Li, Tianyu Cao, Ming Zeng, Zhuofeng Wu, Meng Jiang, Huasheng Li, Lihong Li, Bing Yin

AdabNER: Arabic Digital Archive Books with Nested Entity Recognition (2026.acl-long)

Challenge:	Large Language Models exhibit strong capabilities in single-turn instruction following but suffer from Lost-in-Conversation (LiC) when instructions are revealed progressively in multi-turn settings, models get "Lost in Conversation"
Approach:	They propose a framework that encourages models to generate correct answers and judge solvability in multi-turn conversations.
Outcome:	The proposed framework improves models' ability to balance problem-solving with abstention . it reduces premature answering behaviors that cause lost-in-conversation (LiC)

Copied to clipboard

Aya Mourad, Mustafa Jarrar

Anchoring the Cache: Mitigating Contextual Hallucination in KV-Compressed Long-Context Summarization (2026.acl-long)

Challenge:	Named Entity Recognition (NER) is a subtask of information extraction that classifies entities into predefined categories like person names.
Approach:	They propose a large-scale nested Arabic Named Entity Recognition dataset . they fine-tuned five pre-trained Arabic BERT encoders in two settings .
Outcome:	The first large-scale nested NER dataset for Arabic literary texts is published online . the dataset yields 78,530 entity mentions, 18.96% of which are nestated .

Copied to clipboard

Yu Fu, Chen Luo, Josef Valvoda, Xin Zhang, Xuejing Lei, Xiao Pan, Hui Liu, Yue Dong

NeuralFSM: Adaptive Multi-Agent Coordination via Learning Finite-State Execution Policy (2026.acl-long)

Challenge:	Recent studies show that KV cache compression can increase hallucination scores in LLMs . modern LLM models support extremely long sequences, but their impact on model hallucinosity remains underexplored.
Approach:	They propose a decoding-phase strategy that selectively removes generated KV pairs from retrieval heads responsible for retrieving critical information from source context.
Outcome:	The proposed method reduces hallucination across multiple models and datasets while preserving computational efficiency.

Copied to clipboard

Jiye Wang, Yu Wang, Jianbin Li, Shiduo Yang, Kenan Guo, Yuanhe Zhao

Ranking Reasoning LLMs under Test-Time Scaling (2026.acl-long)

Challenge:	Existing approaches to multi-agent problem solving rely on hand-crafted protocols or automatically designed topologies.
Approach:	They propose a state-driven framework that formulates multi-agent problem solving as a finite-state execution process.
Outcome:	The proposed framework outperforms baselines on diverse benchmarks by 6.74%–19.39% while reducing token consumption.

Copied to clipboard

Mohsen Hariri, Michael Hinczewski, Jing Ma, Vipin Chaudhary

Quantifying and Improving the Robustness of Retrieval-Augmented Language Models Against Spurious Features in Grounding Data (2026.acl-long)

Challenge:	Large language models (LLMs) are increasingly used as general-purpose reasoning systems for tasks such as programming and mathematical problem solving.
Approach:	They formalize dense benchmark ranking under test-time scaling and introduce a library that implements statistical ranking methods such as paired-comparison models, item response theory, voting rules, graph- and spectral-based methods.
Outcome:	The proposed method is based on paired-comparison models, item response theory (IRT) models, voting rules, graph- and spectral-based methods.

Copied to clipboard

Shiping Yang, Jie Wu, Wenbiao Ding, Ning Wu, Shining Liang, Ming Gong, Hongzhi Li, Hengyuan Zhang, Angel X. Chang, Dongmei Zhang

CL2GEC: A Multi-Discipline Benchmark for Continual Learning in Chinese Literature Grammatical Error Correction (2026.acl-long)

Challenge:	Existing studies on robustness to explicit noise (e.g., document semantics) but overlook implicit noise (spurious features).
Approach:	They propose a framework to quantify the robustness of RAGs against spurious features by integrating a data synthesis pipeline and a taxonomy.
Outcome:	The proposed framework quantifies the robustness of RALMs against spurious features.

Copied to clipboard

Shang Qin, Jingheng Ye, Yinghui Li, Hai-Tao Zheng, Qi Li, Jinxiao Shan, Zhixing Li, Hong-Gee Kim

PEARL: Self-Evolving Assistant for Time Management with Reinforcement Learning (2026.acl-long)

Challenge:	Existing CGEC benchmarks for multi-disciplinary writing are limited . continual learning (CL) is a promising solution to handle domain-specific linguistic variation and prevent catastrophic forgetting.
Approach:	They propose a Chinese Literature Continual Learning benchmark to evaluate adaptive CGEC across disciplines.
Outcome:	The proposed benchmark includes 10,000 human-annotated sentences spanning 10 disciplines, each exhibiting distinct linguistic styles and error patterns.

Copied to clipboard

Bingxuan Li, Jeonghwan Kim, Cheng Qian, Xiusi Chen, Eitan Anzenberg, Niran Kundapur, Heng Ji

Challenge:	Scheduling logistics associated with meetings can amount to hours each week . a lack of automation can cause a lot of errors and can be costly .
Approach:	They propose a benchmark for long-horizon calendar conflict resolution that automates calendar conflict management . they propose 'pearl' that augments the language agent with an external preference memory .
Outcome:	a new benchmark for long-horizon calendar conflict resolution shows that existing agents perform poorly with high error rates.

Beyond Static Artifacts: An Evolutionary Framework for Synthetic Claim Generation (2026.acl-long)

Copied to clipboard

Yeqing Teng, Jiasheng Si, Shuxia Lin, Linhai Zhang, Weiyu Zhang, Wenpeng Lu, Deyu Zhou, Xiaoming Wu

Challenge:	Existing claim detection benchmarks treat claims as static textual artifacts . current research ignores sociological etiology of how information naturally emerges and mutates .
Approach:	They propose a socially generative framework for synthetic claim generation . they propose utterance, proposition and context-based simulations to capture truth decay .
Outcome:	The proposed paradigm models claims as socially evolving entities . it allows precise simulation of truth decay and intervened propagation with multi-auditor oversight .

CloneMem: Benchmarking Long-Term Memory for AI Clones (2026.acl-long)

Copied to clipboard

Sen Hu, Zhiyu Zhang, Yuxiang Wei, Xueran Han, Zhenheng Tang, Ronghao Chen, Huacan Wang

Memory efficiency and resource-rational encoding in sentence processing (2026.acl-long)

Challenge:	Existing memory benchmarks rely on user–agent conversational histories, which are temporally fragmented and insufficient for capturing continuous life trajectories.
Approach:	They propose a benchmark for evaluating long-term memory in AI Clone scenarios grounded in non-conversational digital traces, including diaries, social media posts, and emails, spanning one to three years.
Outcome:	Experiments show that existing memory benchmarks struggle in this setting, highlighting open challenges for life-grounded personalized AI.

Copied to clipboard

Weijie Xu, Brian Dillon, Richard Futrell

PRISM: Probing Reasoning, Instruction, and Source Memory in LLM Hallucinations (2026.acl-long)

Challenge:	Existing studies have shown that language models need to be constrained in their use of working memory for context, the analogue to human working memory (WM).
Approach:	They propose to inject noise into hidden representations of Transformer-based LMs to capture constraint on memory encoding.
Outcome:	The proposed model improves alignment with human reading times and makes them more compressed and categorical.

Copied to clipboard

Yuhe Wu, Guangyu Wang, Yuran Chen, Jiatong Zhang, Yutong Zhang, Yujie Chen, Jiaming Shang, Guang Zhang, Zhuang Liu

VC-Inspector: Advancing Reference-free Evaluation of Video Captions with Factual Analysis (2026.acl-long)

Challenge:	Existing benchmarks for hallucination evaluation rely on mixed queries and posterior evaluation, which quantifies hallucinosity severity but offers limited insight into where and why they occur.
Approach:	They propose a controlled benchmark that disentangles hallucinations into four dimensions: knowledge missing, knowledge errors, reasoning errors, and instruction-following errors.
Outcome:	The proposed model disentangles hallucinations into four dimensions: knowledge missing, knowledge errors, reasoning errors, and instruction-following errors.

Copied to clipboard

Shubhashis Roy Dipta, Tz-Ying Wu, Subarna Tripathi

MQM Re-Annotation: A Technique for Collaborative Evaluation of Machine Translation (2026.acl-long)

Challenge:	Existing metrics for caption evaluation lack factual accuracy and limited context handling . VC-Inspector provides reproducible, fact-aware alternative that aligns closely with human judgments.
Approach:	They propose a lightweight, open-source large multimodal model for reference-free evaluation of video captions with a focus on factual accuracy.
Outcome:	Experiments show that VC-Inspector can generalize across diverse domains and improve on existing metrics.

Copied to clipboard

Parker Riley, Daniel Deutsch, Mara Finkelstein, Colten DiIanni, Juraj Juraska, Markus Freitag

Multilingual Language Models Encode Script Over Linguistic Structure (2026.acl-long)

Challenge:	a critical component of machine translation model development is evaluating model quality.
Approach:	They propose a two-stage version of the current translation evaluation paradigm (MQM) they propose re-annotation, which uses raters to review and edit annotations .
Outcome:	The proposed method improves annotation quality by finding errors missed in the first pass.

Copied to clipboard

Aastha A K Verma, Anwoy Chatterjee, Mehak Gupta, Tanmoy Chakraborty

ChiKhaPo: A Large-Scale Multilingual Benchmark for Evaluating Lexical Comprehension and Generation in Large Language Models (2026.acl-long)

Challenge:	a recent study suggests that multilingual language models organize representations around surface form, but the nature of this internal organization remains elusive.
Approach:	They analyze language-associated units across different model families and scales . romanization induces near-disjoint representations that align with neither native-script inputs nor English .
Outcome:	The results show that multilingual language models organize representations around surface form . romanization induces near-disjoint representations that align with neither native-script inputs nor English .

Copied to clipboard

Emily Chang, Niyati Bafna

UniDataBench: Evaluating Data Analytics Agents Across Structured and Unstructured Data (2026.acl-long)

Challenge:	Existing benchmarks for large language models (LLMs) are restricted to high- or mid-resource languages, and evaluate performance on higher-order tasks in reasoning and generation.
Approach:	They propose a multilingual benchmarking tool to evaluate lexical comprehension and generation abilities of large language models.
Outcome:	The proposed benchmarks cover 2700+ languages and surpasses existing benchmarks in terms of language coverage.

Copied to clipboard

Han Weng, Zhou Liu, Yuanfeng Song, Xiaoming Yin, Xing Chen, Wentao Zhang

Beyond Code Pairs: Dialogue-Based Data Generation for LLM Code Translation (2026.acl-long)

Challenge:	Existing benchmarks do not assess agents’ capabilities across data types . Existing tools only evaluate agents' ability to extract reasonable insights across data formats.
Approach:	They propose a multi-source benchmark to evaluate the performance of data analytics agents in handling diverse data sources.
Outcome:	The proposed agent performs end-to-end analysis over diverse data sources by automatically discovering cross-source linkages, decomposing goals, and generating robust, self-correcting code to extract actionable insights.

Copied to clipboard

Le Chen, Nuo Xu, Winson Chen, Bin Lei, Pei-Hung Lin, Dunzhi Zhou, Rajeev Thakur, Caiwen Ding, Ali Jannesari, Chunhua Liao

MM-PoisonRAG: Disrupting Multimodal RAG with Local and Global Knowledge Poisoning Attacks (2026.acl-long)

Challenge:	Large language models (LLMs) have shown remarkable capabilities in code translation, yet their performance deteriorates in low-resource programming domains such as Fortran and emerging frameworks like CUDA .
Approach:	They propose a dual-LLM Questioner–Solver pipeline that integrates external knowledge from compilers and runtime feedback to generate verified translations and multi-turn dialogues.
Outcome:	The proposed model outperforms proprietary models on key metrics like compilation success and accuracy.

Copied to clipboard

Hyeonjeong Ha, Qiusi Zhan, Jeonghwan Kim, Dimitrios Bralios, Saikrishna Sanniboina, Nanyun Peng, Kai-Wei Chang, Daniel Kang, Heng Ji

IterCOMP: Reasoning-aware Adaptive Prompt Compression for Multi-hop Question Answering (2026.acl-long)

Challenge:	Existing research exposes multimodal large language models to knowledge poisoning attacks . localized poisoning attack achieves up to 56% success rate even under restricted access . globalized poison attack completely disrupts model generation to 0% accuracy with just one poisoned content.
Approach:	They propose a framework to study the vulnerability of multimodal RAG under knowledge poisoning attacks.
Outcome:	The proposed framework exploits two new attack strategies on multimodal RAGs under knowledge poisoning.

Copied to clipboard

JungMin Yun, YoungBin Kim

Do LLMs Really Memorize Personally Identifiable Information? Revisiting PII Leakage with a Cue-Controlled Memorization Framework (2026.acl-long)

Challenge:	Existing prompt compression methods are designed for single-turn queries and fail to capture interdependent reasoning steps.
Approach:	They propose a unified, training-free prompt compression framework that integrates multi-hop reasoning within an iterative compression loop.
Outcome:	Experiments on MusiQue, 2WikiMultiHopQA, and HotpotQA show that iterCOMP achieves significant improvements in Exact Match and F1 scores while reducing the token budget.

Copied to clipboard

Xiaoyu Luo, Yiyi Chen, Qiongxiu Li, Johannes Bjerva

Observations and Remedies for Large Language Model Bias in Self-Consuming Performative Loop (2026.acl-long)

Challenge:	Large Language Models (LLMs) have been reported to “leak” Personally Identifiable Information (PII) successful PII reconstruction often interpreted as evidence of memorization.
Approach:	They propose a principled revision of memorization evaluation for Large Language Models . they propose PII leakage should be evaluated under low lexical cue conditions .
Outcome:	The proposed method is based on a multilingual re-evaluation of PII leakage across 32 languages and multiple memorization paradigms.

Copied to clipboard

Yaxuan Wang, Zhongteng Cai, Yujia Bao, Xueru Zhang, Yang Liu

RAG-on-a-Diet: A Reinforcement Learning-Based Dynamic Resource Optimization Framework for RAG (2026.acl-long)

Challenge:	Existing synthetic training loops for large language models cause performance drops and induce emerging biases . a large amount of generated content is posted to coding platforms, social media platforms and other platforms on the internet .
Approach:	They propose a self-consuming retraining loop where models are trained on their own outputs . they use a control loop to isolate and analyze feedback-driven bias evolution .
Outcome:	The proposed model increases preference bias and decreases disparate bias.

Copied to clipboard

Hongwen Ding, Yizheng Zhao

MTSQL-R1: Towards Long-Horizon Multi-Turn Text-to-SQL via Agentic Training (2026.acl-long)

Challenge:	Existing frameworks for knowledge-intensive multi-hop question answering do not adapt to how a trajectory unfolds.
Approach:	They propose a lightweight reinforcement-learning agent that treats each reasoning hop as an independent decision and selects the smallest model sufficient for it.
Outcome:	The proposed agent cuts Monetary Inference Cost by 60.07% against IRCoT with only a 3.7% F1 drop and matches Adaptive-RAG’s F1 at 37.30% lower cost.

Copied to clipboard

Taicheng Guo, Hai Wang, Chaochun Liu, Mohsen Golalikhani, Xin Chen, Xiangliang Zhang, Chandan K. Reddy

The Subjectivity of Respect in Police Traffic Stops: Modeling Community Perspectives in Body-Worn Camera Footage (2026.acl-long)

Challenge:	Existing systems for multi-turn Text-to-SQL are limited to a short-horizon paradigm, generating a query per turn without execution, explicit verification, and refinement, which leads to non-executable or incoherent outputs.
Approach:	They propose to train an agentic training framework for long-horizon multi-turn Text-to-SQL that uses a Markov Decision Process to generate a query per turn without execution, explicit verification, and refinement.
Outcome:	Experiments on CoSQL and SParC show that MTSQL-R1 consistently outperforms strong baselines, highlighting the importance of environment-driven verification and memory-guided refinement for conversational semantic parsing.

Copied to clipboard

Preni Golazizian, Elnaz Rahmati, Jackson Trager, Zhivar Sourati, Nona Ghazizadeh, Georgios Chochlakis, Jose J. Alcocer, Kerby Bennett, Aarya Vijay Devnani, Parsa Hejabi, Harry G. Muttram, Akshay Kiran Padte, Mehrshad Saadatinia, Chenhao Wu, Alireza Salkhordeh Ziabari, Michael Sierra-Arévalo, Nicholas Weller, Shrikanth Narayanan, Benjamin A.t. Graham, Morteza Dehghani

Challenge:	a new study examines the perception of police-civilian traffic stops using respect ratings and free-text rationales from multiple perspectives.
Approach:	They propose a traffic-stop dataset annotated with respect ratings and rationales from multiple perspectives . they use a criterion-driven preference data construction framework to predict personalized respect ratings .
Outcome:	The proposed framework improves rating prediction performance and rationale alignment across all three annotators.

LLM-Based Multi-Task Bangla Hate Speech Detection: Type, Severity, and Target (2026.acl-long)

Copied to clipboard

Md Arid Hasan, Firoj Alam, Md Fahad Hossain, Usman Naseem, Syed Ishtiaque Ahmed

MMSciCode: Real-world Evaluation of Multilingual Multi-Discipline Scientific Research Coding (2026.acl-long)

Challenge:	Existing work on social media platforms is limited in its ability to detect hate speech . a lack of reliable and scalable automated hate speech detection systems is a challenge for low-resource languages like Bangla.
Approach:	They propose to use a single-task, single-targeted, single language dataset to identify hate speech in Bangla.
Outcome:	The proposed dataset is the largest manually annotated Bangla hate-speech dataset to date.

Copied to clipboard

Xue Xia, Zheyuan Yang, Arman Cohan, Yilun Zhao

CURA: Clinical Uncertainty Risk Alignment for Language Model–Based Risk Prediction (2026.acl-long)

Challenge:	MMSciCode is a benchmark for evaluating foundation models in scientific code generation.
Approach:	They propose a multilingual, multi-discipline benchmark for evaluating foundation models in scientific code generation that integrates domain-specific knowledge with algorithmic reasoning.
Outcome:	The new benchmark is annotated by domain experts and features rigorous quality controls to ensure dataset integrity and authenticity.

Copied to clipboard

Sizhe Wang, Ziqi Xu, Claire Najjuuko, Charles Alba, Chenyang Lu

Unlocking the Black Box of Latent Reasoning: An Interpretability-Guided Approach to Intervention (2026.acl-long)

Challenge:	Clinical language models (LMs) are increasingly applied to support clinical risk prediction from free-text notes, yet their uncertainty estimates are poorly calibrated and clinically unreliable.
Approach:	They propose a framework that aligns clinical LM-based risk estimates and uncertainty with individual error likelihoods and cohort-level ambiguities.
Outcome:	The proposed framework improves accuracy on clinical risk prediction tasks without compromising discrimination.

Copied to clipboard

Shuochen Chang, Tong Bai, Xiaofeng Zhang, Qianli Ma, Qingyang Liu, Zhaohe Liao, Yibo Miao, Li Niu

EthicMind: A Risk-Aware Framework for Ethical-Emotional Alignment in Multi-Turn Dialogue (2026.acl-long)

Challenge:	Existing methods for large language models (LLMs) lack a coherent representation of reasoning steps.
Approach:	They propose a set of latent reasoning interventions that enable latent thinking and decode-time interventions that refine the latent process by imposing the identified geometric and semantic priors.
Outcome:	The proposed models unlock latent capabilities and improve reasoning accuracy without any parameter updates.

Copied to clipboard

Jiawen Deng, Wei Li, Wentao Zhang, Ziyun Jiao, Fuji Ren

Resolving the Security-Auditability Dilemma with Auditable Latent Chain-of-Thought Alignment (2026.acl-long)

Challenge:	Existing dialogue models address empathy and ethical safety in isolation . Existing models fail to adapt their behavior as ethical risk and user emotion evolve .
Approach:	They propose a risk-aware framework that integrates ethical-emotional alignment in dialogue as an explicit turn-level decision problem.
Outcome:	The proposed framework achieves more consistent ethical guidance and emotional engagement than baselines in ethically complex interactions.

Copied to clipboard

Guan Wang, Biyu Zhou, Xuehai Tang, Jizhong Han, Songlin Hu

ART: Attention Replacement Technique to Improve Factuality in LLMs (2026.acl-long)

Challenge:	Extensive experiments show that ALCA reduces the success rate of adaptive jailbreak attacks by over 40% compared to strong baselines, while preserving performance.
Approach:	They propose a framework that decouples internal reasoning from external output and allows the model to reconstruct its latent reasoning into human-readable text for supervision under specific guidance.
Outcome:	The proposed framework reduces the success rate of adaptive jailbreak attacks by over 40% compared to baselines while preserving performance.

Copied to clipboard

Ziqin Luo, Yihao Quan, Xiaofeng Zhang, Xiaosong Yuan, Chen Shen

More Thinking, Less Talking: Internalizing Deliberative Safety into LLM Parameters (2026.acl-long)

Challenge:	Existing methods to mitigate hallucinations in large language models are expensive and require significant resources.
Approach:	They propose a training-free method that replaces uniform attention patterns in shallow layers with local attention patterns to reduce hallucinations.
Outcome:	The proposed method reduces hallucinations across multiple LLM architectures.

Copied to clipboard

Guan Wang, Xuehai Tang, Biyu Zhou, Jizhong Han, Songlin Hu

ToolScope: Enhancing LLM Agent Tool Use through Tool Merging and Context-Aware Filtering (2026.acl-long)

Challenge:	Existing safety alignment methods leave Large Language Models vulnerable to sophisticated jailbreak attacks.
Approach:	They propose a safety reasoning internalization framework that internalizes safety reasoning into an implicit computational pathway using Low-Rank Adaptation (LoRA).
Outcome:	The proposed framework achieves a 43% lower Attack Success Rate (ASR) against distinct jailbreak attacks compared to strong baselines.

Copied to clipboard

Marianne Menglin Liu, Daniel Garcia, Fjona Parllaku, Vikas Upadhyay, Fahad Shah, Dan Roth

PRInTS: Reward Modeling for Long-Horizon Information Seeking (2026.acl-long)

Challenge:	Large language model (LLM) agents often face strict input context limits, preventing efficient consideration of large toolsets.
Approach:	They propose a tool that allows LLMs to merge tools with auto-correction and toolScopeRetriever to rank and select only the most relevant tools for each query.
Outcome:	Evaluations on three state-of-the-art LLMs and three open-source tool-use benchmarks show gains of 8.38% to 38.6% in tool selection accuracy.

Copied to clipboard

Jaewoo Lee, Archiki Prasad, Justin Chen, Zaid Khan, Elias Stengel-Eskin, Mohit Bansal

Follow the Flow: On Information Flow Across Textual Tokens in Text-to-Image Models (2026.acl-long)

Challenge:	Existing PRMs cannot capture richer dimensions of information-seeking steps, such as tool interactions and reasoning over tool outputs.
Approach:	They propose a generative PRM trained with dual capabilities that compresses the growing context while preserving essential information for step evaluation.
Outcome:	PRInTS improves on FRAMES, GAIA, and WebWalkerQA models while preserving essential information for step evaluation.

Copied to clipboard

Guy Kaplan, Michael Toker, Yuval Reif, Yonatan Belinkov, Roy Schwartz

Fiction Flows: A Replication and Reinterpretation of Narrative Sequentiality (2026.acl-long)

Challenge:	Prior work focused on improving alignment by refining the diffusion process, ignoring the role of the text encoder, which guides the diffusion.
Approach:	They investigate how semantic information is distributed across token representations in text-to-image prompts by patching techniques to uncover encoding patterns.
Outcome:	The proposed model can improve alignment and generation quality by modifying the diffusion stage and the cross-attention mechanism.

Copied to clipboard

Andrew Piper, Sil Hamilton, Haiqi Zhou, Federico Pianzola

Conjunctive Prompt Attacks in Multi-Agent LLM Systems (2026.acl-long)

Challenge:	a new study shows that imagined narratives exhibit higher "flow" than recalled narratives, but this advantage is not reducible to standard coherence measures.
Approach:	They propose a language-model-based measure of sentence-level predictability to measure narrative flow . they find that imagined stories flow better than recalled ones .
Outcome:	The proposed measure of sentence-level predictability is based on language models . it shows that fiction exhibits a robust sequentiality advantage over reality-bound genres .

Copied to clipboard

Nokimul Hasan Arif, Qian Lou, Mengxin Zheng

Evaluating Visual Narrative Coherence in Story Visualization via Diversified Storylines (2026.acl-long)

Challenge:	Existing defenses do not reliably stop the attack because no single component appears malicious in isolation.
Approach:	They study conjunctive prompt attacks where trigger key and adversarial template appear benign alone but activate harmful behavior when routing brings them together.
Outcome:	The proposed model significantly improves performance over baselines while keeping false activations low.

Copied to clipboard

Minha Jhang, Kyeongman Park, Hyukhun Koh, Kyomin Jung

Rethinking the Idiomaticity Decomposability Hypothesis: Evidence from Distributional Learning (2026.acl-long)

Challenge:	Existing evaluation metrics and datasets often neglect visual continuity and narrative diversity.
Approach:	They propose a visual context-aware metric for story visualization that uses large vision-language models to jointly assess caption fidelity and inter-image consistency.
Outcome:	The proposed framework achieves a Spearman’s correlation comparable to human agreement on two benchmarks and blends diverse and controlled narrative elements at adjustable ratios, producing challenging evaluation sets.

Copied to clipboard

Maggie Mi, Golzar Atefi, Atsuki Yamaguchi, Felix Gers, Aline Villavicencio, Nafise Sadat Moosavi

KinyaProp: Fine-Grained Propaganda Annotation in Kinyarwanda (2026.acl-long)

Challenge:	Decomposability is thought to predict syntactic flexibility, but is not attributed to distributional experience.
Approach:	They propose a model-internal measure of decomposability and relate it to human ratings, syntactic flexibility, and predictability while tracking idiom learning during pretraining.
Outcome:	The proposed model-internal measure correlates weakly with human judgments and shows a small but consistent negative relationship with syntactic flexibility.

Copied to clipboard

Manzi Fabrice Niyigaba, Ivory Yang, Soroush Vosoughi

Revisiting Audio-language Pretraining for Learning General-purpose Audio Representation (2026.acl-long)

Challenge:	Propaganda is a widely used approach for shaping public opinion and disseminating misinformation in news media.
Approach:	They propose a fine-grained propaganda dataset for Kinyarwanda . they find that current LLMs are not reliable annotators in low resource settings .
Outcome:	The proposed dataset shows that current LLMs perform poorly in low resource settings . the dataset shows they perform poorly on discourse-level techniques .

Copied to clipboard

Wei-Cheng Tseng, Xuanru Zhou, Mingyue Huo, Yiwen Shao, Hao Zhang, Dong Yu

Trajectory Signatures of Deception in Large Language Models (2026.acl-long)

Challenge:	Existing methods for learning general-purpose audio representations are limited in scope and coverage of audio attributes.
Approach:	They propose to use a 10.7M caption dataset to compare ALP with captioning . they find that contrastive learning yields competitive, transferable representations .
Outcome:	The proposed model yields competitive, transferable representations, while captioning exhibits better scalability.

Copied to clipboard

Viraaji Mothukuri, Reza M. Parizi

Improving the Distributional Alignment of LLMs using Supervision (2026.acl-long)

Challenge:	Existing interpretability methods exhibit structural limitations in the field of deception detection.
Approach:	They propose a method to capture layerwise activations at sparse "decision points" . they capture deception as a dynamic process, a trajectory through the model's hidden-state space .
Outcome:	The proposed classifier achieves comparable performance to PCA-reduced probing for binary sycophancy detection and shows preliminary utility for 4-way deception-type classification.

Copied to clipboard

Gauri Kambhatla, Sanjana Gautam, Angela Zhang, Alexander Liu, Ravi Srinivasan, Junyi Jessy Li, Matthew Lease

Challenge:	Existing work to evaluate LLMs' alignment with human values and opinions has a key shortcoming.
Approach:	They propose to add supervision to LLMs to improve alignment with diverse populations . they find that supervision improves alignment across public health, public opinion, values and beliefs .
Outcome:	The proposed method improves the alignment of LLMs with diverse populations on subjective questions.

SMARTER: A Data-efficient Framework to Improve Toxicity Detection with Explanation via Self-augmenting Large Language Models (2026.acl-long)

Copied to clipboard

Huy Nghiem, Advik Sachdeva, Hal Daumé Iii

GameplayQA: A Benchmarking Framework for Decision-Dense POV-Synced Multi-Video Understanding of 3D Virtual Agents (2026.acl-long)

Challenge:	Toxic content encompasses a wide spectrum of terminologies whose definitions vary by platform.
Approach:	They propose a 2-stage framework for explainable content moderation using Large Language Models (LLMs) they leverage LLMs’ own outputs to generate synthetic explanations for correct and incorrect labels . they refine explanation quality through cross-model training, allowing weaker models to align with stronger ones.
Outcome:	Experiments on 3 benchmarks show that the proposed framework achieves 13% macro-F1 improvement over few-shot baselines using only 6-57% of training data.

Copied to clipboard

Yunzhe Wang, Runhui Xu, Kexin Zheng, Tianyi Zhang, Jayavibhav Niranjan Kogundi, Soham Hans, Volkan Ustun

DeepFact: Co-Evolving Benchmarks and Agents for Deep Research Factuality (2026.acl-long)

Challenge:	Multimodal Large Language Models are emerging as a backbone for autonomous agents in 3D environments.
Approach:	They propose a framework for evaluating agentic-centric perception and reasoning through video understanding.
Outcome:	The proposed framework evaluates agentic-centric perception and reasoning through video understanding.

Copied to clipboard

Yukun Huang, Leonardo F. R. Ribeiro, Momchil Hardalov, Bhuwan Dhingra, Markus Dreyer, Venkatesh Saligrama

Challenge:	Existing fact-checkers usually target general-domain atomic claims . citation-grounded fact- checking ignores claims without explicit citations .
Approach:	They propose to use a benchmark to test whether claim-level factuality is transferable . they instantiate Audit-then-Score as a versioned DRR factualism benchmark .
Outcome:	The proposed benchmark outperforms the best prior deep-research and traditional fact-checkers by 14.3 and 24.9 points.

R1-RE: Cross-Domain Relation Extraction with RLVR (2026.acl-long)

Copied to clipboard

Runpeng Dai, Tong Zheng, Run Yang, Kaixian Yu, Hongtu Zhu

Training LLMs for Divide-and-Conquer Reasoning Elevates Test-Time Scalability (2026.acl-long)

Challenge:	Relation extraction (RE) is a core task in natural language processing.
Approach:	They propose a supervised learning task for relation extraction (RE) based on annotation guidelines.
Outcome:	The proposed model achieves an average OOD accuracy of 70%, on par with leading proprietary models such as GPT-4o.

Copied to clipboard

Xiao Liang, Zhong-Zhi Li, Zhenghao Lin, Eric Hanchen Jiang, Hengyuan Zhang, Yelong Shen, Kai-Wei Chang, Ying Nian Wu, Yeyun Gong, Weizhu Chen

ClaimDB: A Fact Verification Benchmark over Large Structured Data (2026.acl-long)

Challenge:	Large language models have demonstrated strong reasoning capabilities through step-by-step chain-of-thought (CoT) reasoning, but their strictly sequential nature constrains test-time scalability.
Approach:	They propose an end-to-end reinforcement learning framework to enhance LLMs' DAC-style reasoning capacity by decomposing a problem into subproblems and solving them sequentially.
Outcome:	The proposed model surpasses CoT by 8.6% and 6.3% on competition-level benchmarks and is available at the [github.com/MasterVito/DAC-RL].

Copied to clipboard

Michael Theologitis, Preetam Prabhu Srikar Dammu, Chirag Shah, Dan Suciu

PluRule: A Benchmark for Moderating Pluralistic Communities on Social Media (2026.acl-long)

Challenge:	despite substantial progress in fact-verification benchmarks, this setting remains largely underexplored.
Approach:	They propose a fact-verification benchmark where evidence for claims is derived from compositions of millions of records and multiple tables.
Outcome:	The proposed benchmarks score below 55% accuracy with 30 state-of-the-art LLMs and are released on github.

Copied to clipboard

Zoher Kachwala, Bao Tran Truong, Rasika Muralidharan, Haewoon Kwak, Jisun An, Filippo Menczer

What Makes a Good Curriculum? Disentangling the Effects of Data Ordering on LLM Mathematical Reasoning (2026.acl-long)

Challenge:	Social media are shifting towards community-governed platforms where groups define their own norms.
Approach:	They propose a multimodal, multilingual benchmark for detecting 13,371 rule violations across 1,989 Reddit communities . they show that bigger models and increased context provide marginal gains, and universal rules like civility and self-promotion are easier to detect.
Outcome:	The proposed model can detect 13,371 rule violations across 1,989 Reddit communities across 2,885 rules in 9 languages.

Copied to clipboard

Yaning Jia, Chunhui Zhang, Xingjian Diao, Xiangchi Yuan, Zhongyu Ouyang, Chiyu Ma, Soroush Vosoughi

CASS: Nvidia to AMD Transpilation with Data, Models, and Benchmark (2026.acl-long)

Challenge:	Curriculum learning (CL) orders data corpus by difficulty, but prior work employs disparate difficulty metrics and training setups.
Approach:	They propose a framework that decomposes curriculum difficulty into five dimensions: Problem Difficulty, Model Surprisal, Confidence Margin, Predictive Uncertainty and Decision Variability.
Outcome:	The proposed framework decomposes curriculum difficulty into five dimensions . the results show that no curriculum strategy dominates universally .

Copied to clipboard

Ahmed Heakl, Gustavo Bertolo Stahl, Sarim Hashmi, Seung Hun Eddie Han, Mukul Ranjan, Arina Kharlamova, Salman Khan, Abdulrahman Mahmoud

SAHM: A Benchmark for Arabic Financial and Shari’ah-Compliant Reasoning (2026.acl-long)

Challenge:	Cross-architecture GPU code translation is essential for unlocking low-level hardware portability, yet no scalable solution exists.
Approach:	They propose a dataset and model suite for source- and assembly-level GPU code translation that trains domain-specific translation models that achieve 88.2% accuracy on CUDA HIP and 69.1% on SASS RDNA3 .
Outcome:	The proposed model achieves 88.2% accuracy on CUDA HIP and 69.1% on SASS RDNA3 outperforming commercial baselines including GPT-5.1, Claude-4.5, and Hipify by wide margins.

Copied to clipboard

Rania Elbadry, Sarfraz Ahmad, Ahmed Heakl, Dani Bouch, Momina Ahsan, Muhra AlMahri, Marwa Elsaid Khalil, Yuxia Wang, Salem Lahlou, Sophia Ananiadou, Veselin Stoyanov, Jimin Huang, Xueqing Peng, Preslav Nakov, Zhuohan Xie

Confident, Calibrated, or Complicit: Safety Alignment and Ideological Bias in LLM Hate Speech Detection (2026.acl-long)

Challenge:	English financial NLP has progressed rapidly through benchmarks for sentiment, document understanding, and financial question answering.
Approach:	They propose a document-grounded benchmark and instruction-tuning dataset for Arabic financial NLP and Shari’ah-compliant reasoning.
Outcome:	The proposed dataset contains 14,380 expert-verified instances spanning seven tasks . it includes financial sentiment analysis, extractive summarization, and event–cause reasoning .

Copied to clipboard

Sanjeevan Selvaganapathy, Mehwish Nasim

Aligning Large Language Models via Fully Self-Synthetic Data (2026.acl-long)

Challenge:	censored models outperform uncensoreed counterparts in accuracy and robustness, achieving 69.0% accuracy versus 64.1% strict accuracy.
Approach:	They examine how large language models with minimal safety alignment compare with more heavily aligned counterparts when deployed using political personas.
Outcome:	The proposed model outperforms uncensored models in accuracy and robustness, while uncensors are more malleable to ideological framing.

Copied to clipboard

Shangjian Yin, Zhepei Wei, Xinyu Zhu, Wei-Lin Chen, Yu Meng

Corpus-Dependent Subcharacter Encoding via HMM-Guided Code Assignment (2026.acl-long)

Challenge:	Existing approaches to reinforcement learning from human feedback (RLHF) require expensive human-annotated datasets and proprietary models like GPT-4 to annotate preference pairs.
Approach:	They propose a self-synthetic framework for LLM alignment where all training data, including prompts (i.e., user queries), responses, and preferences, are generated by the model itself.
Outcome:	The proposed framework enhances the model’s chat capabilities on standard benchmarks like AlpacaEval 2.0 while maintaining strong performance on downstream objective tasks.

Copied to clipboard

Tatsuya Hiraoka

Challenge:	Latom is a corpus-dependent alternative to byte encoding that learns fixed-length atomic codes from text.
Approach:	They propose a corpus-dependent alternative to byte encoding that learns fixed-length atomic codes from text.
Outcome:	The proposed framework improves text classification accuracy and reduces decoding errors.

DRInQ: Evaluating Conversational Implicature with Controlled Context Variation (2026.acl-long)

Copied to clipboard

Hirona Jacqueline Arai, Xiang Ren

From Individual to Common: An Early Exploration of Consensus in Non-verifiable Data for Balanced Preference Optimization (2026.acl-long)

Challenge:	Recent large language models exhibit strong conversational fluency but are unreliable when interpretation depends on reasoning that integrates social and contextual cues.
Approach:	They propose a semi-automated pipeline that produces question-context-interpretation instances with systematic variation to isolate pragmatic variation while holding each question’s surface form fixed.
Outcome:	The proposed framework isolates pragmatic variation while holding each question’s surface form fixed.

Copied to clipboard

Shangjian Yin, Zhouxing Shi

Challenge:	Reinforcement Learning with Verifiable Rewards (RLVR) has demonstrated remarkable effectiveness in boosting the objective performance of Large Language Models (LLMs).
Approach:	They propose a dataset where response pairs differ only by subtle nuances and a model with a non-verifiable dataset.
Outcome:	The proposed model outperforms models trained on data with explicit quality gaps while maintaining objective capabilities.

RIMRULE: Improving Tool-Using Language Agents via MDL-Guided Rule Learning (2026.acl-long)

Copied to clipboard

Xiang Gao, Yuguang Yao, Qi Zhang, Kaiwen Dong, Avinash Baidya, Ruocheng Guo, Hilaf Hasson, Kamalika Das

GAM: Hierarchical Graph-based Agentic Memory for LLM Agents (2026.acl-long)

Challenge:	Large language models (LLMs) struggle to use tools reliably in domain-specific settings.
Approach:	They propose a neuro-symbolic approach to adapt large language models to task-specific tools . they propose reusable rules that are distilled from failure traces and injected into the prompt .
Outcome:	Experiments show that the proposed approach outperforms prompting-based adaptation methods and complements finetuning.

Copied to clipboard

Zhaofen Wu, Hanrong Zhang, Fulin Lin, Wujiang Xu, Xinran Xu, Yankai Chen, Henry Peng Zou, Shaowen Chen, Weizhi Zhang, Xue Liu, Philip S. Yu, Hongwei Wang

CypherSmith: Transforming Text-to-Cypher Generation for LLMs with Synthetic Data (2026.acl-long)

Challenge:	Current unified stream-based memory systems facilitate context updates but remain vulnerable to interference from transient noise.
Approach:	They propose a hierarchical Graph-based Agentic Memory framework that explicitly decouples memory encoding from consolidation to resolve conflict between rapid context perception and stable knowledge retention.
Outcome:	The proposed framework outperforms state-of-the-art benchmarks on LoCoMo and LongDialQA.

Copied to clipboard

Zeyu Zhang, Kexuan Sun, Zheng Tang, Jens-S. Vöckler, Thien Huu Nguyen, Thuy Vu

New Terms, New Toxicity: Consensus-based Chinese Neologism Toxicity Detection via Search-Augmented LLMs (2026.acl-long)

Challenge:	Existing datasets are small, domain-limited, and lack diversity, constraining LLM progress.
Approach:	They propose a knowledge Graph retrieval tool that can translate natural language questions into structured queries.
Outcome:	Extensive experiments show that CypherSmith achieves state-of-the-art performance.

Copied to clipboard

Shiyao Cui, QingLin Zhang, Di Wang, Yida Lu, Zhexin Zhang, Jinhua Gao, Jinglin Yang, Min He, Han Qiu, Minlie Huang

ATGL: An Adaptive-Threshold Global Loss for Document-level Relation Extraction (2026.acl-long)

Challenge:	Neologisms can foster new linguistic consensus by stabilizing shared meanings and usage in common communicative norms.
Approach:	They propose a taxonomy that captures the origins and consensus-verification criteria of toxic neologisms . they propose 'SeTox' framework that integrates real-time web context for naeologim detection .
Outcome:	The proposed framework outperforms large-scale models in detecting neologism toxicity.

Copied to clipboard

Huangming Xu, Fu Zhang, Zhixuan Yang, Lu Zhang, Jingwei Cheng

L2Dir: Integrating L_2-Norm and Directional Alignment for Unsupervised Contrastive Representation Learning in Multimodal Retrieval (2026.acl-long)

Challenge:	Document-level relation extraction (DocRE) aims to determine which relations hold between a given entity pair in a document.
Approach:	They propose a document-level relation extraction paradigm that decouples existing losses into independent positive and negative losses, which interact solely with a shared threshold.
Outcome:	The proposed model outperforms existing models on four datasets and achieves state-of-the-art results.

Copied to clipboard

Tianyu Zong, Rui Dai, Hongzhu Yi, Yuanxiang Wang, Zhenghao Zhang, Zhenyu Guan, Yujia Yang, Bingkang Shi, Yueyang Ding, Xiangxiang Chu, Kaikui Liu, Jungang Xu

Towards Intrinsic Interpretability of Large Language Models: A Survey of Design Principles and Architectures (2026.acl-long)

Challenge:	Existing approaches to multimodal representation learning focus on directional alignment and embedding magnitudes (L2-norm) however, these methods often fail to account for the intrinsic role of L2-norm in the contrastive process.
Approach:	They propose a plug-and-play framework that optimizes L2-norm alignment and Directional consistency jointly.
Outcome:	The proposed framework achieves consistent and significant performance gains over established baselines across 95 tasks using UniIR and VLM2Vec-V2 frameworks.

Copied to clipboard

Yutong Gao, Qinglin Meng, Yuan Zhou, Liangming Pan

ScholaWrite: A Dataset of End-to-End Scholarly Writing (2026.acl-long)

Challenge:	Existing studies on explainable AI focus on post-hoc explanation methods that interpret trained models through external approximations.
Approach:	They propose to categorize existing approaches into five design paradigms: functional transparency, concept alignment, representational decomposability, explicit modularization, and latent sparsity induction.
Outcome:	The proposed approaches are categorized into five design paradigms: functional transparency, concept alignment, representational decomposability, explicit modularization, and latent sparsity induction.

Copied to clipboard

Khanh Chi Le, Linghe Wang, Minhwa Lee, Ross Volkov, Luan Tuyen Chau, Dongyeop Kang

Challenge:	SCHOLAWRITE traces the multi-month journey from initial drafts to final manuscripts . authors demonstrate the value of capturing scientists’ cognitive writing process .
Approach:	They present a dataset of end-to-end scholarly writing tracing the multi-month journey from initial drafts to final manuscripts.
Outcome:	The first dataset of end-to-end scholarly writing traces the multi-month journey from initial drafts to final manuscripts over four months.

What Deserves Memory: Adaptive Memory Distillation for LLM Agents (2026.acl-long)

Copied to clipboard

Wenquan Ma, Jiayan Nan, WenLong Wu

SOAPTriage: SOAP-Guided Multi-View Clinical Text Modeling Framework for Automated ESI Prediction (2026.acl-long)

Challenge:	Existing memory systems rely on predefined heuristics such as importance scores, emotional tags, or factual templates, encoding designer intuition rather than learning from the data itself.
Approach:	They propose an adaptive memory distillation framework that casts the assessment of the experience’s future utility as a matter of predictability.
Outcome:	The proposed framework achieves strong performance, efficiency, and storage reduction.

Copied to clipboard

Enming Wang, Jianlei Wang, Xueping Peng, Hongjiao Guan, Yinglong Wang, Sibo Wei, Jianbin Guo, Ruifeng Xu, Wenpeng Lu

MoRI: Learning Motivation-Grounded Reasoning for Scientific Ideation in Large Language Models (2026.acl-long)

Challenge:	Emergency departments rely on the Emergency Severity Index (ESI) to assess patient acuity and prioritize care.
Approach:	They propose a SOAP-guided multi-view clinical text modeling framework for automated ESI prediction based on the SOAP paradigm .
Outcome:	The proposed framework outperforms prompting-based, multi-agent, and encoder-based baselines.

Copied to clipboard

Chenyang Gu, Jiahao Cheng, Meicong Zhang, Pujun Zheng, Jinquan Zheng, Guoxiu He

WSDPO: A Generative Word Sense Disambiguation Framework with Chain-of-Thought and Preference Optimization (2026.acl-long)

Challenge:	Existing LLMs emulate human research workflows but lack scientific grounding . empirical results show that MoRI outperforms strong commercial LLM models .
Approach:	They propose a framework that explicitly learns scientific reasoning from research motivations to methodologies.
Outcome:	The proposed framework outperforms commercial LLMs and agentic baselines in novelty, technical rigor, and feasibility.

Copied to clipboard

Kunpeng Kang, Shuaimin Li, Kaiyuan Zhang, Luyang Zhang, Jiasheng Si, Bing Xu, Kehai Chen, Muyun Yang, Wenpeng Lu

From Regulatory Approvals to Patents: Cross-Domain Linking for Cardiovascular Device Traceability (2026.acl-long)

Challenge:	Word sense disambiguation (WSD) is a fundamental task in natural language processing.
Approach:	They propose a training framework for generative WSD with chain-of-thought (CoT) and preference optimization.
Outcome:	The proposed framework achieves significant performance gains on rare and unseen settings and exhibits strong generalization in standard evaluation settings.

Copied to clipboard

Qingqing Yang, Haijiang Liu, Moyan Li

TrustTable: A Neuro-Symbolic Auditing Framework for Faithful Table QA (2026.acl-long)

Challenge:	a new cross-domain entity linking problem exists between FDA-approved medical devices and patents . a recent study compared the semantics of FDA documents with patents, resulting in minimal overlap .
Approach:	They propose a framework that links FDA-approved medical devices to their patents . they propose 'MedDevKG' framework that integrates a domain-specific ontology .
Outcome:	The proposed framework outperforms existing methods in lower-bound recalls and noise reductions.

Copied to clipboard

Guangzhen Zhao, Dechang Kong, Tongyu Wu, Zhenjiang Dong

Lizard: An Efficient Linearization Framework for Large Language Models (2026.acl-long)

Challenge:	Large Language Models (LLMs)-based TableQA models exhibit unfaithful behavior where correct answers are derived through erroneous reasoning paths.
Approach:	They propose a neuro-symbolic framework to audit LLM reasoning processes . it enforces factual grounding and ensures logical soundness by verifying reasoning chains .
Outcome:	The proposed framework outperforms LLM judges in majority voting and rejection sampling with process supervision.

Copied to clipboard

Chien Van Nguyen, Huy Huu Nguyen, Ruiyi Zhang, Hanieh Deilamsalehy, Puneet Mathur, Viet Dac Lai, Haoliang Wang, Jayakumar Subramanian, Ryan A. Rossi, Trung Bui, Nikos Vlassis, Franck Dernoncourt, Thien Huu Nguyen

TPS-Bench: Evaluating AI Agents’ Tool Planning & Scheduling Abilities in Compounding Tasks (2026.acl-long)

Challenge:	Existing linearization frameworks that rely on softmax attention with quadratic time and memory complexity pose significant computational and memory bottlenecks for long-context applications.
Approach:	They propose a linearization framework that transforms pretrained Transformer-based Large Language Models into subquadratic architectures that closely approximate softmax attention while preserving model quality.
Outcome:	Experiments show that the proposed framework outperforms existing methods by 9.4 - 24.5 points on the 5-shot MMLU benchmark and demonstrating superior associative recall.

Copied to clipboard

Hanwen Xu, Xuyao Huang, Yuzhe Liu, Zhijie Deng

S2S-Arena: Evaluating Paralinguistic Instruction Following in Speech-to-Speech Models (2026.acl-long)

Challenge:	Large language model (LLM) agents have demonstrated strong problem-solving competence across domains like research and coding.
Approach:	They propose to use a tool repository to analyze the ability of large language model agents to solve complex problems.
Outcome:	The proposed model outperforms open-source and closed-source models in task completion rate and efficiency.

Copied to clipboard

Feng Jiang, Zhiyu Lin, Yiyang Liu, Liumeng Xue, Fan Bu, Yuhao Du, Xiangying Chen, Benyou Wang, Haizhou Li

Can LLM Safety Be Ensured by Constraining Parameter Regions? (2026.acl-long)

Challenge:	Existing benchmarks rely heavily on text-based evaluation and largely ignore paralinguistic cues such as prosody, emotion, and speaker traits.
Approach:	They propose a speech-native benchmark for evaluating instruction-following S2S models with explicit assessment of both semantic understanding and paralinguistic expression.
Outcome:	The proposed system enables more natural, robust, and human-aligned speech agents.

Copied to clipboard

Zongmin Li, Jian Su, Farah Benamara, Aixin Sun

Challenge:	Large language models (LLMs) are often assumed to contain parameter subsets whose modification directly influences safety behaviors.
Approach:	They evaluate four methods to identify parameter subsets with "safety regions" they find low overlap, but overlap drops when refinement is done using utility datasets .
Outcome:	The proposed methods show low overlap and drop significantly when refined using utility datasets.

CityVG: Contrastive Fine-Tuning and Reward-Based Chain-of-Thought Reasoning for Zero-Shot City-Scale 3D Visual Grounding (2026.acl-long)

Copied to clipboard

Jianjun Zhang, Hanli Wang

E2EDev: Benchmarking Large Language Models in End-to-End Software Development Task (2026.acl-long)

Challenge:	Existing methods for 3D visual grounding are limited to small-scale indoor data or require heavy supervision.
Approach:	They propose a contrastive fine-tuning strategy to align textual queries with urban scene graphs.
Outcome:	The proposed framework achieves strong zero-shot localization performance and generalizes effectively to unseen urban environments.

Copied to clipboard

Jingyao Liu, Chen Huang, Zhizhao Guan, Wenqiang Lei, Yang Deng

Reasoning While Asking: Transforming Reasoning Large Language Models from Passive Solvers to Proactive Inquirers (2026.acl-long)

Challenge:	Existing E2ESD benchmarks are limited by coarse-grained requirement specifications and unreliable evaluation protocols.
Approach:	They propose a benchmark to assess whether generated software meets user needs . they use a fine-grained set of user requirements and a fully automated testing pipeline .
Outcome:	E2EDev is a benchmark to assess whether generated software meets user needs through mimicking real user interactions.

Copied to clipboard

Xin Chen, Feng Jiang, Yiqian Zhang, Hardy Chen, Shuo Yan, Wenya Xie, Min Yang, Shujian Huang

Utility-Oriented Visual Evidence Selection for Multimodal Retrieval-Augmented Generation (2026.acl-long)

Challenge:	Existing reasoning-oriented LLMs lack a blind self-thinking paradigm . current models fail to recognize when their reasoning is underinformed or based on ambiguous user instructions .
Approach:	They propose a new reasoning paradigm that transforms LLMs into proactive inquirers that interleave reasoning with clarification.
Outcome:	The proposed model outperforms baseline models on mathematical reasoning, code generation, and document editing.

Copied to clipboard

Weiqing Luo, Zongye Hu, Xiao Wang, Zhiyuan Yu, Haofeng Zhang, Ziyi Huang

Mitigating Selection Bias in Large Language Models via Permutation-Aware GRPO (2026.acl-long)

Challenge:	Existing methods for multimodal retrieval-augmented generation rely on semantic relevance or surface-level similarity, which are often misaligned with the actual utility of visual evidence for downstream reasoning.
Approach:	They propose a latent notion of evidence usefulness and propose 'surrogate-accelerated' framework that efficiently estimates evidence utility using lightweight multimodal models.
Outcome:	The proposed framework outperforms state-of-the-art models while achieving substantial reductions in computational cost.

Copied to clipboard

Jinquan Zheng, Jia Yuan, Jiacheng Yao, Chenyang Gu, Pujun Zheng, Guoxiu He

KG-ViP: Bridging Knowledge Grounding and Visual Perception in Multi-modal LLMs for Visual Question Answering (2026.acl-long)

Challenge:	Existing inference-time debiasing ignores that the same question should yield consistent answers across permutations.
Approach:	They propose a permutation-aware group-relative policy optimization which enforces permutations-consistent semantic reasoning.
Outcome:	The proposed model outperforms strong baselines across seven benchmarks while maintaining high overall performance.

Copied to clipboard

Zhiyang Li, Ao Ke, Yukun Cao, Xike Xie

Reinforcement Learning–Guided Adaptive Tuning for Out-of-Distribution Harmful Text Detection (2026.acl-long)

Challenge:	Existing approaches to Visual Question Answering lack synergistic potential of scene graphs and scene graph.
Approach:	They propose a retrieval-and-fusion pipeline that fuses scene graphs and commonsense graphs to enable multi-modal reasoning.
Outcome:	Experiments on FVQA 2.0+ and MVQA benchmarks show that KG-ViP outperforms existing methods.

Copied to clipboard

Mengyu Xiang, Tinghao Chen, Boxu Han, Qiudan Li, Shu Wu, Daniel Dajun Zeng

SOAR: Supervision from Observation for Agentic Reinforcement Learning (2026.acl-long)

Challenge:	Existing methods for testing harmful information on social media rely on fixed parameters that fail to handle substantial semantic discrepancies . RLAT can be used to adapt to semantic variations while preventing overfitting from continuous tuning.
Approach:	They propose a reinforcement learning-guided adaptive tuning method for harmful text detection that optimizes consistency loss and applies word-level attention constraints to reduce over-reliance on local words.
Outcome:	The proposed method outperforms state-of-the-art models in cross-platform and cross-temporal scenarios across multiple public datasets.

Copied to clipboard

Meng Li, Lei Li, Xiting Wang, Yi Yuan, Zheng Wei, null Brucebian, Zang Li

Business as Rulesual: A Benchmark and Framework for Business Rule Flow Modeling with LLMs (2026.acl-long)

Challenge:	Prior work assigns supervision based on outcome rewards or external reward models, but ignores environment observations, a critical source of learning.
Approach:	They propose a supervision-based agentic reinforcement learning system that integrates environment observations as an explicit supervision signal.
Outcome:	The proposed model improves performance on reasoning and deep research tasks while reducing erroneous and inefficient tool usage.

Copied to clipboard

Chen Yang, Ruping Xu, Ruizhe Li, Bin Cao, Jing Fan

Specializing Large Models for Oracle Bone Script Interpretation via Component-Grounded Multimodal Knowledge Augmentation (2026.acl-long)

Challenge:	Existing benchmarks for extracting structured procedural knowledge from unstructured business documents are limited by simplistic schemas and shallow logical dependencies.
Approach:	They propose a framework for extracting structured procedural knowledge from unstructured business documents . they propose BREX, a carefully curated benchmark comprising 409 real-world business documents and 2,855 expert-annotated rules .
Outcome:	The proposed framework outperforms standard prompts in rule extraction and execution.

Copied to clipboard

Jianing Zhang, Runan Li, Honglin Pang, Ding Xia, Zhou Zhu, Qian Zhang, Chuntao Li, Xi Yang

HyperMem: Hypergraph Memory for Long-Term Conversations (2026.acl-long)

Challenge:	Existing methods for deciphering ancient Chinese Oracle Bone Script (OBS) treat deciphering as a closed-set image recognition problem, which fails to bridge the "interpretation gap" .
Approach:	They propose a vision-language model framework that integrates a VLM and an LLM to automate a reasoning chain of component identification and knowledge retrieval.
Outcome:	The proposed framework yields more detailed and precise decipherments compared to baseline methods.

Copied to clipboard

Juwei Yue, Chuanrui Hu, Jiawei Sheng, Zuyi Zhou, Wenyuan Zhang, Tingwen Liu, Li Guo, Yafeng Deng

LeLoRA: Learnable Low-Rank Adaptation of Large Language Models (2026.acl-long)

Challenge:	Existing approaches to long-term memory management rely on pairwise relations, causing fragmented retrieval.
Approach:	They propose a hypergraph-based hierarchical memory architecture that explicitly models high-order associations using hyperedges.
Outcome:	Experiments show that HyperMem achieves state-of-the-art performance with 92.73% accuracy for long-term conversations.

Copied to clipboard

Xiaoling Zhou, Mingjie Zhang, Zhemg Lee, Wei Ye, Shikun Zhang

Gated Tree Cross-Attention for Checkpoint-Compatible Syntax Injection in Decoder-Only LLMs (2026.acl-long)

Challenge:	Existing approaches to fine-tuning large language models (LLMs) rely on manually specified and fixed hyperparameters, resulting in suboptimal performance and low parameter efficiency.
Approach:	They propose a framework that allows for dynamically learned adaptive adaptation strategies to be used to fine-tune large language models.
Outcome:	The proposed framework outperforms baselines in adapting large language models.

Copied to clipboard

Xinyu Gao, Shaonan Wang, Nai Ding

DiningBench: A Hierarchical Multi-view Benchmark for Perception and Reasoning in the Dietary Domain (2026.acl-long)

Challenge:	Decoder-only large language models are brittle to minor grammatical perturbations, causing reliability problems.
Approach:	They propose a checkpoint-compatible gated tree cross-attention branch that reads constituency chunk memory while keeping the backbone architecture unchanged.
Outcome:	The proposed framework strengthens syntactic competence beyond continued training benchmarks and transformer backbones without compromising commonsense reasoning.

Copied to clipboard

Song Jin, Juntian Zhang, Xun Zhang, Zeying Tian, Fei Jiang, Guojun Yin, Wei Lin, Yong Liu, Rui Yan

Octopus: Gated Selective Attention for Memory-Bounded Long-Context Inference in Large Language Models (2026.acl-long)

Challenge:	Existing vision-language models lack fine-grained classification, single-view imagery, and inaccurate metadata.
Approach:	They propose a hierarchical, multi-view benchmark to evaluate VLMs across three levels of cognitive complexity.
Outcome:	The proposed benchmark evaluates vision-language models across three levels of complexity . it systematically identifies five primary failure modes . the proposed benchmarks are available on https://github.com/meituan/DiningBench.

Copied to clipboard

Chien Van Nguyen, Ryan A. Rossi, Linh Ngo Van, Franck Dernoncourt, Thien Huu Nguyen

SAME: Safety-Aware Model Editing Guided by Safety Transformation (2026.acl-long)

Challenge:	Subquadratic architectures rely on aggressive state compression that degrades performance on complex reasoning tasks.
Approach:	They propose a framework that confers fixed-memory inference onto pretrained Transformers . they use a learnable module that enforces an adaptive sparsity policy over the context history .
Outcome:	The proposed framework outperforms state-of-the-art linearized baselines on the GSM8K benchmark by over 36 points under identical memory constraints.

Copied to clipboard

Jiayi Wang, Shipeng Wang, Ji Wu, Jian Sun

Where the Cat Sat: A Multilingual Framework for Spatial Language Understanding (2026.acl-long)

Challenge:	Existing models that update or insert new knowledge require sequential parameter updates while maintaining model capability.
Approach:	They propose a model editing approach that estimates safety transforms and identifies corresponding safety direction in the neural activation space and aligns neural activations and network parameter updates under the safety constraints.
Outcome:	The proposed approach reduces unsafe responses to malicious queries while preserving the effectiveness of model editing.

Copied to clipboard

Demian Inostroza, Ekaterina Vylomova, Charles Kemp, Mae Carroll, Wanchun Li, Meladel Mistica

LCMA-SRT: Language-Conditional Mixture-of-Experts Adapters for Joint Multilingual Speech Recognition and Translation (2026.acl-long)

Challenge:	Existing work exhibits biases toward English and prepositional marking . Existing models are limited in understanding spatial relations across typologically diverse languages .
Approach:	They propose a multilingual framework and benchmark for spatial language understanding . they decompose spatial relations into surface elements and semantic components . their results suggest surface parsing does not entail spatial understanding - they argue .
Outcome:	The proposed framework and benchmark decomposes spatial relations into surface elements and semantic components.

Copied to clipboard

Nanjie Li, Xiaoyong Guo, Hao Huang, Xu Haihua, Wei Shi

DefGen-Bench: A Benchmark for Chinese Criminal Defence Opinion Generation in LegalAI (2026.acl-long)

Challenge:	Existing hierarchical transducers suffer from negative transfer and unstable target-language generation, while training separate models for each direction is computationally prohibitive.
Approach:	They propose a hierarchical transducer with language-conditional Mixture-of-Experts adapters to improve multilingual joint automatic speech recognition and speech translation.
Outcome:	Experiments on Europarl-ST (9 languages, 72 directions) show that LCMA-SRT improves both ASR and ST within a single joint model, reducing average WER and improving BLEU and COMET over strong hierarchical transducer baselines.

Copied to clipboard

Senbo Zhang, Qiqi Wang, Fanghao Lou, Guanyu Chen, Yihong Pan, Huijia Li, Qian Liu

Challenge:	Existing LegalAI tasks are descriptive or predictive, requiring the users to translate the information into legal reasoning.
Approach:	They propose a task to generate a structured defence opinion conditioned jointly on an indictment and the defendant’s stated opinion, which often present conflicting claims.
Outcome:	The proposed approach improves on eight large language models (LLMs) and shows that it is more efficient than previous approaches.

Escaping the Echo Trap: On Credit Assignment Failure in Multi-turn LLM Self-Reflection (2026.acl-long)

Copied to clipboard

Linxuan Du, Guangquan Xue, Xiaobo Liang, Qipeng Huang, Yuyang Ding, Xinyu Shi, Zhang Yijun, Ji Qi, Wenpeng Zhu, Juntao Li, Min Zhang

Challenge:	Existing methods for multi-turn self-reflection are limited by the Echo Trap problem . the model is limited by its inherent capabilities and repeats earlier reflections to preserve reward signals .
Approach:	They propose a tree-structured extension of GRPO for multi-turn self-reflection which enables more accurate advantage estimation.
Outcome:	The proposed method mitigates behavior collapse and improves performance across benchmarks.

ARF-RLHF: Adaptive Reward-Following for RLHF through Emotion-Driven Self-Supervision and Trace-Biased Dynamic Optimization (2026.acl-long)

Copied to clipboard

Yuxuan Zhang

SMART: Evaluating LLMs’ Mathematical Reasoning via a Human Cognitive Process-Inspired Benchmark (2026.acl-long)

Challenge:	prevailing RLHF methods such as PPO and DPO depend on large-scale binary preference annotations.
Approach:	They propose a method which converts natural feedback into continuous preference trajectories and optimizes them using the novel TraceBias algorithm.
Outcome:	The proposed approach outperforms PPO and DPO in a variety of domains and improves alignment by up to 7.6% across diverse LLMs and preference domains.

Copied to clipboard

Yujie Hou, Mei Wang, Yaoyao Zhong, Ting Zhang, Xuetao Ma, Hua Huang

Decentralized Arena: Towards Democratic and Scalable Automatic Evaluation of Language Models (2026.acl-long)

Challenge:	Existing evaluation methods focus on the final answer or on the intermediate reasoning steps, overlooking its inherently multi-stage and multi-dimensional nature.
Approach:	They propose a benchmark that decomposes mathematical problem-solving into four cognitive dimensions and introduces dimension-specific tasks to measure their cognitive processes.
Outcome:	The proposed model decomposes mathematical problem-solving into four cognitive dimensions and introduces dimension-specific tasks to measure their cognitive processes.

Copied to clipboard

Yanbin Yin, Kun Zhou, Zhen Wang, Xiangdong Zhang, Yifei Shao, Shibo Hao, Yi Gu, Jieyuan Liu, Somanshu Singla, Tianyang Liu, Eric P. Xing, Zhengzhong Liu, Haojian Jin, Zhiting Hu

LBLLM: Lightweight Binarization of Large Language Models via Three-Stage Distillation (2026.acl-long)

Challenge:	closed-ended question-based benchmarks struggle with saturation as newer models emerge . crowd-sourced leaderboards rely on costly and slow human judges .
Approach:	They propose a framework that leverages collective intelligence from all large language models to evaluate each other.
Outcome:	a new framework enables a democratic, pairwise evaluation of all large language models . it achieves 97% correlation with human judgements, while significantly reducing the cost.

Copied to clipboard

Siqing Song, Chuang Wang, Yong Lang, Yi Yang, Xu-Yao Zhang

SeDev: Structured Semantic Exploration for LLM-Driven Code Generation (2026.acl-long)

Challenge:	Existing methods for implementing large language models are limited by high computational and memory requirements.
Approach:	They propose a lightweight binarization framework that achieves effective W(1+1)A4 quantization through a novel three-stage quantization strategy.
Outcome:	The proposed framework surpasses state-of-the-art methods on W2A4 quantization settings across languages.

Copied to clipboard

Ronghui Yang, Jie Liu, Jiajie Zeng, Jiexin Wang, Jiuchuan Jiang, Bo An, Yi Cai, Mengchen Zhao

EVM-QuestBench: An Execution-Grounded Benchmark for Natural-Language Transaction Code Generation (2026.acl-long)

Challenge:	Large Language Models (LLMs) have shown remarkable capabilities in automating code generation, but they suffer from insufficient exploration of the vast solution space.
Approach:	They propose a large-scale LLM-driven code generation framework that efficiently finds high-quality solutions in only a few iterations.
Outcome:	The proposed framework outperforms baselines while maintaining reasonable time and computational costs.

Copied to clipboard

Pei Yang, Wanyi Chen, Ke Wang, Lynn Ai, Eric Yang, Tianyu Shi

TIGER: Text-Informed Generalized Enzyme-Reaction Retrieval (2026.acl-long)

Challenge:	Existing evaluations of large language models overlook execution accuracy and safety.
Approach:	They propose an execution-grounded benchmark for natural-language transaction-script generation on EVM-compatible chains.
Outcome:	The proposed benchmark finds large performance gaps in the models with 5 independent rounds.

Copied to clipboard

Yuhang Zhang, Keyan Ding, Peilin Chen, Han Liu, Can Lin, Ruixi Chen, Shiqi Wang, Qi Song

MUSEG: Reinforcing Video Temporal Understanding via Timestamp-Aware Multi-Segment Grounding (2026.acl-long)

Challenge:	Existing approaches to enzyme–reaction retrieval suffer from poor generalization across tasks and distributions . TIGER is a text-informed generalized enzyme-reaction retrieval framework that bridges enzymes and biochemical reactions.
Approach:	They propose a text-informed generalized enzyme-reaction retrieval framework that leverages protein-to-text generation models to distill textual knowledge from enzyme sequences.
Outcome:	The proposed framework outperforms state-of-the-art methods in enzyme–reaction retrieval tasks and distributions.

Copied to clipboard

Fuwen Luo, Shengfeng Lou, Chi Chen, Ziyue Wang, Chenliang Li, Weizhou Shen, Jiyue Guo, Peng Li, Ming Yan, Ji Zhang, Fei Huang, Yang Liu

User Perceptions vs. Proxy LLM Judges: Privacy and Helpfulness in LLM Responses to Privacy-Sensitive Scenarios (2026.acl-long)

Challenge:	Existing methods for MLLMs struggle with fine-grained temporal reasoning . despite advances in video understanding, current methods struggle with time-sensitive tasks .
Approach:	They propose a time-stamp-aware multi-segment grounding method that enhances temporal understanding by introducing timestamps.
Outcome:	The proposed method outperforms existing methods on time-sensitive tasks and generalizes well across diverse temporal understanding scenarios.

Copied to clipboard

Xiaoyuan Wu, Roshni Kaushik, Wenkai Li, Lujo Bauer, Koichi Onoue

HARPO: Hierarchical Agentic Reasoning for User-Aligned Conversational Recommendation (2026.acl-long)

Challenge:	Large language models (LLMs) are rapidly being adopted for tasks like drafting emails, summarizing meetings, and answering health questions.
Approach:	They conducted a scenario-based evaluation of Large language models (LLMs) using 90 PrivacyLens scenarios.
Outcome:	The proposed models can leak private information in complex scenarios, but they do not measure user perceptions directly.

Copied to clipboard

Subham Raj, Aman Vaibhav Jha, Mayank Anand, Sriparna Saha

ShopSimulator: Evaluating and Exploring RL-Driven LLM Agent for Shopping Assistants (2026.acl-long)

Challenge:	Recent advances in large language models have significantly improved conversational recommender systems performance.
Approach:	They propose a framework that reframes conversational recommendation as a structured decision-making process optimized for multi-dimensional recommendation quality.
Outcome:	The proposed framework improves on ReDial, INSPIRED, and MUSE while maintaining competitive response quality.

Copied to clipboard

Pei Wang, Yanan Wu, Xiaoshuai Song, Weixun Wang, Gengru Chen, Zhongwen Li, Kezhong Yan, Qi Liu, Ken Deng, Shuaibing Zhao, Shaopan Xiong, Xuepeng Liu, Xuefeng Chen, Wanxi Deng, Wenbo Su, Bo Zheng

Challenge:	Existing studies on large language model-based agents focus on evaluation benchmarks without training support.
Approach:	They propose a large-scale Chinese shopping simulation environment that uses large language models to train agents.
Outcome:	The proposed model performs poorly in a large-scale and challenging shopping environment in China.

NL ⇒ Schedule: Evaluate Multitask Scheduling Capability of Large Language Models (2026.acl-long)

Copied to clipboard

Wenrui Liao, Weihong Du, Yi Li, Hongru Liang, Wenqiang Lei

SLICEFORMER: Static Program Slicing Using Language Models With Dataflow-Aware Pretraining and Constrained Decoding (2026.acl-long)

Challenge:	Existing methods for scheduling from natural language descriptions rely on experts with limited scheduling skills and domain knowledge.
Approach:	They propose a model to generate a feasible schedule from natural language descriptions.
Outcome:	The proposed framework achieves more robust performance than six state-of-the-art LLM+solver methods.

Copied to clipboard

Pengfei He, Shaowei Wang, Tse-Hsun Chen, Muhammad Asaduzzaman

SCOPE: Boosting LLM Efficiency with Scoped Position Encoding (2026.acl-long)

Challenge:	Static program slicing is a software engineering technique for isolating code relevant to specific variables.
Approach:	They propose a new approach that reformulates static program slicing as a sequence-to-sequence task using small language models such as CodeT5+.
Outcome:	The proposed approach improves on Java and Python program slicing benchmarks with up to 22% gain in ExactMatch.

Copied to clipboard

Qingguo Qi, Hongyang Chen, Zhao Li

Debate-of-Thoughts: Resolving Knowledge Conflicts in LLMs Through Internal Deliberation (2026.acl-long)

Challenge:	Positional encodings are fundamental to Transformers, but explicit methods like RoPE can degrade under length extrapolation and incur extra arithmetic and memory-access overhead.
Approach:	They propose a framework that reimagines structured sparsity as an intrinsic position encoding mechanism.
Outcome:	The proposed framework reduces the number of attention FLOPs by 8x compared to RoPE on LLaMA-3-8B architectures while reducing training and inference latency.

Copied to clipboard

Guocong Li, Qirui Hu, Ping Wang, Guofeng Zhang, Jian Wu, Hongxia Xu

Projecting Out the Malice: A Global Subspace Approach to LLM Detoxification (2026.acl-long)

Challenge:	Existing methods for retrieval augmented generation are based on a simplistic binary choice of relying on external contexts or memory.
Approach:	They propose a framework that transforms conflict resolution into an active deliberation process by incorporating contradictions as opportunities for deeper reasoning.
Outcome:	Experiments show that DoT outperforms state-of-the-art methods while generating transparent debate transcripts that explain its decisions.

Copied to clipboard

Zenghao Duan, Zhiyi Yin, Zhichao Shi, Liang Pang, Shaoling Jing, Zihe Huang, Jiayi Wu, Yu Yan, Jingcheng Deng, Huawei Shen, Xueqi Cheng

Challenge:	Large language models (LLMs) exhibit exceptional performance but pose inherent risks of generating toxic content.
Approach:	They propose a method that removes toxic subspaces from FFN parameters . they propose to use a lightweight method to eliminate toxic subespaces .
Outcome:	The proposed method achieves SOTA detoxification while preserving general capabilities without large-scale retraining.

From Local to Global: Revisiting Structured Pruning Paradigms for Large Language Models (2026.acl-long)

Copied to clipboard

Ziyan Wang, Enmao Diao, Qi Le, Pu Wang, Minwoo Lee, Shu-ping Yeh, Evgeny Stupachenko, Hao Feng, Li Yang

Audio MultiChallenge: A Multi-Turn Evaluation of Spoken Dialogue Systems on Natural Human Interaction (2026.acl-long)

Challenge:	Structured pruning is a practical approach to deploying large language models (LLMs) but it fails to capitalize on modest task-specific calibration signals, causing limited downstream gains.
Approach:	They propose a method that removes attention heads and MLP channels using loss-based important scores . they use perplexity for language modeling and a margin-based objective for decision-style tasks .
Outcome:	The proposed method lowers perplexity and improves accuracy at higher sparsity . it also stabilizes accuracy and mitigates perxity collapse without fine-tuning .

Copied to clipboard

Advait Gosai, Tyler Vuong, Utkarsh Tyagi, Steven Li, Wenjia You, Miheer Bavare, Arda Uçar, Zhongwang Fang, Brian Jang, Bing Liu, Yunzhong He

GenomeQA: Benchmarking General Large Language Models for Genome Sequence Understanding (2026.acl-long)

Challenge:	End-to-end (E2E) spoken dialogue systems are replacing cascaded pipelines for voice-based human-AI interaction. Existing benchmarks evaluate these systems on synthetic speech and single-turn tasks, leaving multi-turn conversational ability underexplored.
Approach:	They propose an open-source benchmark to evaluate spoken dialogue systems under natural multi-turn interaction patterns.
Outcome:	The proposed model fails on the highest-performing model with 54.65% pass rate.

Copied to clipboard

Weicai Long, Yusen Hou, Junning Feng, Houcheng su, Shuo Yang, Donglin Xie, Yanlin Zhang

VideoStir: Understanding Long Videos via Spatio-Temporally Structured and Intent-Aware RAG (2026.acl-long)

Challenge:	Existing benchmarks focus on specialized DNA models trained for sequence prediction or evaluate biological knowledge using text-only questions.
Approach:	They propose a benchmark to evaluate general-purpose LLMs on sequence-based genome inference tasks.
Outcome:	The proposed benchmark outperforms baseline models on sequence-based genome inference tasks.

Copied to clipboard

Honghao Fu, Miao Xu, Yiwei Wang, Dailing Zhang, Jun Liu, Yujun Cai

Massively Multilingual Joint Segmentation and Glossing (2026.acl-long)

Challenge:	Existing methods for retrieval-augmented generation (RAG) to long videos are limited by limited context windows and flatten videos into independent segments.
Approach:	They propose a structured and intent-aware long-video RAG framework that structures a video as a spatio-temporal graph and then performs multi-hop retrieval to aggregate evidence across distant yet contextually related events.
Outcome:	The proposed framework is competitive with state-of-the-art baselines without auxiliary information.

Copied to clipboard

Michael Ginn, Lindia Tjuatja, Enora Rice, Ali Marashian, Maria Valentini, Jasmine Xu, Graham Neubig, Alexis Palmer

TexOCR: Advancing Document OCR Models for Compilable Page-to-LaTeX Reconstruction (2026.acl-long)

Challenge:	Existing models generate morpheme-level glosses but assign them to whole words without predicting the actual morphological boundaries, making them less interpretable and therefore untrustworthy to human annotators.
Approach:	They propose to use neural networks to predict interlinear glosses and morphological segmentation from raw text.
Outcome:	The proposed model outperforms GlossLM on glossing and beats open-source models on segmentation, glossing, and alignment.

Copied to clipboard

Chengye Wang, Lin Fu, Zexi Kuang, Yilun Zhao

If an LLM Were a Character, Would It Know Its Own Story? Evaluating Lifelong Learning in LLMs (2026.acl-long)

Challenge:	Existing document OCR largely targets plain text or Markdown, discarding structural and executable properties that make LaTeX essential for scientific publishing.
Approach:	They propose a benchmark and a training corpus for document reconstruction . they train a 2B-parameter model using supervised fine-tuning and reinforcement learning .
Outcome:	The proposed model improves on existing models using supervised fine-tuning and reinforcement learning with verifiable rewards.

Copied to clipboard

Siqi Fan, Xiusheng Huang, Yiqun Yao, Xuezhi Fang, Kang Liu, Peng Han, Shuo Shang, Aixin Sun, Yequan Wang

Challenge:	Existing benchmarks for large language models (LLMs) fail to capture these dynamics, focusing on static, open-ended evaluations.
Approach:	They propose a benchmark to assess lifelong learning in large language models . they use two episodic datasets rich in narrative structure and character interactions .
Outcome:	Experiments on LLMs show that non-parametric methods outperform parametric ones in managing stateful learning.

Graph Reasoning Paradigm: Structured and Symbolic Reasoning with Topology-Aware Reinforcement Learning for Large Language Models (2026.acl-long)

Copied to clipboard

Runxuan Liu, Xianhao Ou, Xinyan Ma, Jiyuan Wang, Jiafeng Liang, Jiaqi Li, Tao He, Zheng Chu, Rongchuan Mu, Zekun Wang, Baoxin Wang, Dayong Wu, Ming Liu, Shijin Wang, Guoping Hu, Bing Qin

LLM-Powered Benchmark Factory: Reliable, Generic, and Efficient (2026.acl-long)

Challenge:	Existing methods for long chain-of-thought (LCoT) are coarse-grained, reward hacking, and poor generalization.
Approach:	They propose a Long Chain-of-Thought (LCoT) model that integrates reinforcement learning with verifiable rewards with a process-aware verification approach.
Outcome:	The proposed model improves reasoning and code generation tasks while reducing the cost of training and performance bottlenecks.

Copied to clipboard

Peiwen Yuan, Shaoxiong Feng, Yiwei Li, Xinglin Wang, Yueqi Zhang, Jiayi Shi, Chuyi Tan, Boyuan Pan, Yao Hu, Kan Li

COSMOS: Connectivity-Oriented Submodular Maximization for Optimal Subgraph Retrieval (2026.acl-long)

Challenge:	Using generic and efficient benchmark generators, human annotators are limited by inefficiency . current benchmark generator methods rely on seed signals, leading to long cycles and high costs .
Approach:	They propose a framework to evaluate LLMs as generic benchmark generators and integrate them as BenchMaker.
Outcome:	The proposed framework achieves comparable performance to human-annotated benchmarks on most metrics.

Copied to clipboard

Boci Peng, Xiao Liu, Boren Hu, Yun Zhu, Xuanbo Fan, Yanwei Yue, Chunyu Yang, Yan Zhang

Attn-GS: Attention-Guided Context Compression for Efficient Personalized LLMs (2026.acl-long)

Challenge:	Existing paradigms treat facts independently or employ myopic search, failing to optimize collective subgraph utility.
Approach:	They propose a framework that formalizes evidence retrieval as a constrained submodular maximization problem.
Outcome:	The proposed framework captures the trade-off between information relevance and structural complexity.

Copied to clipboard

Shenglai Zeng, Tianqi Zheng, Chuan Tian, Dante Everaert, Yau-Shian Wang, Yupin Huang, Michael J. Morais, Rohit Patki, Jinjin Tian, Xinnan Dai, Kai Guo, Monica Xiao Cheng, Hui Liu

Structured Semantic Information Helps Retrieve Better Examples for In-Context Learning Applied to Few-Shot Relation Extraction (2026.acl-long)

Challenge:	Existing approaches to personalize large language models (LLMs) rely on heuristic methods to compress user profiles but they ignore how LLMs process and prioritize different profile components.
Approach:	They propose an attention-guided context compression framework that leverages attention feedback from a marking model to mark important personalization sentences and guides a compression model to generate task-relevant compressed user contexts.
Outcome:	The proposed framework outperforms baselines across tasks, token limits, and settings while reducing token usage by 50 times.

Copied to clipboard

Aunabil Chakma, Mihai Surdeanu, Eduardo Blanco

DUD: Decoupled Update Dynamics for Reliable Uncertainty Quantification in Large Language Models (2026.acl-long)

Challenge:	Existing approaches to relation extraction require many training examples per relation, resulting in low results.
Approach:	They propose a strategy where new examples are selected based on their similarity to the provided 1-shot example.
Outcome:	The proposed strategy outperforms other methods on FS-TACRED and FS - FewRel subsets and achieves state-of-the-art performance on both datasets.

Copied to clipboard

Yixin Bu, Runze Xia, Guanyun Zou, Yupeng Ji, Hongliang Dai, Haodong Liu, Piji Li

JurisBench: A Deep Benchmark for Assessing Large Language Models in Professional Legal Practice (2026.acl-long)

Challenge:	Accurate Uncertainty Quantification (UQ) is critical for reliable deployment of Large Language Models (LLMs).
Approach:	They propose a framework that explicitly decouples FFN and Attention contributions via noise-induced causal interventions to capture model's internal fragility.
Outcome:	The proposed framework outperforms state-of-the-art baselines in both uncertainty estimation and calibration while exhibiting superior cross-dataset generalization.

Copied to clipboard

Ziang Chen, Guannan Li, Fanlin Ji, Yipeng Kang, Jiaqi Li, Muhan Zhang, Yangtao Zhang, Li Tianjiao, Jiannan Wang, Xin Guo, Song-Chun Zhu, Bin Ling

LeCoDe: A Benchmark Dataset for Interactive Legal Consultation Dialogue Evaluation (2026.acl-long)

Challenge:	Existing legal benchmarks evaluate isolated tasks or exam-style questions, failing to capture the procedural interdependencies and adjudicative rigor inherent in professional practice.
Approach:	They propose a vertical, depth-oriented, domain-specific benchmark to evaluate Large Language Models (LLMs) in Chinese civil litigation.
Outcome:	The proposed benchmarks show that large language models exhibit an "illusion of competence" the results highlight a critical gap between fluent linguistic output and judicial reliability .

Copied to clipboard

Weikang Yuan, Kaisong Song, Zhuoren Jiang, Junjie Cao, Yujie Zhang, Jun Lin, Kun Kuang, Ji Zhang, Xiaozhong Liu

METER: Evaluating Multi-Level Contextual Causal Reasoning in Large Language Models (2026.acl-long)

Challenge:	Current systems for legal consultation are insufficient to handle the knowledge-intensive nature of real-world consultations.
Approach:	They propose a multi-turn benchmark dataset to evaluate LLMs in legal consultation settings.
Outcome:	The proposed framework assesses LLMs’ consultation capabilities in terms of (1) clarification capability and (2) professional advice quality.

Copied to clipboard

Pengfeng Li, Chen Huang, Chaoqun Hao, Hongyao Chen, Xiao-Yong Wei, Wenqiang Lei, See-Kiong Ng

PersonalAlign: Hierarchical Implicit Intent Alignment for Personalized GUI Agent with Long-Term User-Centric Records (2026.acl-long)

Challenge:	Existing benchmarks evaluate contextual causal reasoning in fragmented settings, failing to ensure context consistency or cover the full causal hierarchy.
Approach:	They use a unified context to benchmark large language models' contextual causal reasoning skills.
Outcome:	The proposed benchmarks show that LLMs are susceptible to distraction by irrelevant but factually correct information at lower level of causality.

Copied to clipboard

Yibo Lyu, Gongwei Chen, Rui Shao, Weili Guan, Liqiang Nie

Challenge:	GUI agents have shown strong performance under explicit and completion instructions, but real-world deployment requires aligning with users’ more complex implicit intents.
Approach:	They propose a task that requires agents to leverage long-term user records as persistent context to resolve omitted preferences in vague instructions and anticipate latent routines by user state for proactive assistance.
Outcome:	The proposed task improves execution and proactive performance by 15.7% and 7.3% under explicit and completion instructions.

Empowering GUI Agents via Autonomous Experience Exploration and Hindsight Experience Utilization for Task Planning (2026.acl-long)

Copied to clipboard

Tianyi Men, Zhuoran Jin, Pengfei Cao, Yubo Chen, Kang Liu, Jun Zhao

Challenge:	Multimodal web agents are cost-efficient and privacy-preserving, but suffer from weak planning and limited cross-website generalization.
Approach:	They propose a method which autonomously explores environments to discover experiences and utilizes hindsight experience to synthesize strictly aligned, high-level training data.
Outcome:	The proposed method outperforms Qwen2.5-VL-32B model on real-world benchmarks and demonstrates that mastering low-level atomic skills does not guarantee high-level planning competence.

"Excuse me, may I say something..." CoLabScience, A Proactive AI Assistant for Biomedical Discovery and LLM-Expert Collaborations (2026.acl-long)

Copied to clipboard

Yang Wu, Jinhong Yu, Jingwei Xiong, Zhimin Tao, Xiaozhong Liu

Knowledge-driven Augmentation and Retrieval for Integrative Temporal Adaptation (2026.acl-long)

Challenge:	Existing large language models (LLMs) are reactive and respond only when prompted, limiting their effectiveness in collaborative settings.
Approach:	They introduce a proactive LLM assistant designed to enhance biomedical collaboration between AI systems and human experts through timely, context-aware interventions.
Outcome:	The proposed model outperforms baselines in intervention precision and collaborative task utility, highlighting the potential of proactive LLMs as intelligent scientific assistants.

Copied to clipboard

Weisi Liu, Guangzeng Han, Xiaolei Huang

SAD: A Large-Scale Strategic Argumentative Dialogue Dataset (2026.acl-long)

Challenge:	Existing studies either overlook temporal shifts or hardly capture rich shifting patterns of both semantic and knowledge.
Approach:	They develop a temporal adaptive learning framework that captures temporal shifts . they use medical ontology and other knowledge sources to integrate temporal adaptation .
Outcome:	The proposed framework improves classification tasks across multiple domains and domains with knowledge integration.

Copied to clipboard

YongKang Liu, Jiayang Yu, Mingyang Wang, Yiqun Zhang, Ercong Nie, Shi Feng, Daling Wang, Kaisong Song, Hinrich Schuetze

PICTURE: Enhancing Theory-of-Mind in Large Language Models by Revealing, Not Hiding, Characters’ Lack of Knowledge (2026.acl-long)

Challenge:	Argumentation is a key part of human reasoning and decision-making . existing argumentative corpora focus on single-turn settings, but multi-turn dialogues are often realized as multi-turned dialogues .
Approach:	They present a dataset for strategic multi-turn argumentation dialogues . they annotate each utterance with five strategy types, allowing multiple strategies per utterrance .
Outcome:	The proposed dataset shows that explicit prompting improves fluency, stylistic coherence and persuasiveness.

Copied to clipboard

Eojin Jeon, SangKeun Lee

Union-of-Experts: Neurons in Mixture-of-Experts are Secretly Routers (2026.acl-long)

Challenge:	Existing approaches to simulating Theory of Mind (ToM) using LLMs have been a longstanding problem in natural language processing (NLP).
Approach:	They propose a new method that enables LLMs to generate a character’s lack of knowledge within free-form Chain-of-Thought (CoT) based on this method, they propose to generate perspective-taking outputs as free- form explanations without event hiding.
Outcome:	The proposed method outperforms existing prompting methods by an average of 7.3% on false-belief tasks.

Copied to clipboard

Songhao Wu, Ang Lv, Ruobing Xie, Samm Sun, Di Wang, Rui Yan, Yankai Lin

ReflectRM: Boosting Generative Reward Models via Self-Reflection within a Unified Judgment Framework (2026.acl-long)

Challenge:	Mixture-of-Experts (MoE) models rely on an external router to assign tokens to experts, resulting in suboptimal performance.
Approach:	They propose an MoE variant that performs "expert-autonomous routing" by pre-designating a fraction of neurons within each expert as "routing neurons" they pre-train UoE models with up to 3B parameters and show they outperform traditional MoEs with matched efficiency.
Outcome:	The proposed model outperforms existing models with 3B parameters and provides valuable insights into expert-autonomous selection and the broader routing mechanisms of MoE models.

Copied to clipboard

Kai Qin, Liangxin Liu, Yu Liang, Longzheng Wang, null Wangyan, Zhang Yueyang, Long Xia, Zhiyuan Sun, Houde Liu, Daiting Shi

Challenge:	Existing methods for generating reward models focus on outcome-level supervision, neglecting analytical process quality, which constrains their potential.
Approach:	They propose a novel reward model that leverages self-reflection to assess analytical quality and enhance preference modeling.
Outcome:	The proposed model improves performance on four benchmarks and significantly mitigates positional bias.

SQL-Trail: Multi-Turn Reinforcement Learning with Interleaved Feedback for Text-to-SQL (2026.acl-long)

Copied to clipboard

Harper Hua, Zhen Han, Zhengyuan Shen, Meng-Chieh Lee, Sheng Guan, Qi Zhu, Sullam Jeoung, Yueyan Chen, Yunfei Bai, Shuai Wang, Vassilis N. Ioannidis, Huzefa Rangwala

Bootstrapping Code Translation with Weighted Multilanguage Exploration (2026.acl-long)

Challenge:	Recent large language models (LLMs) have significantly improved Text-to-SQL generation, but a gap remains between AI systems and human experts on challenging benchmarks such as BIRD-Sql.
Approach:	They propose a multi-turn reinforcement learning agentic framework for Text-to-SQL that uses execution feedback to iteratively refine its predictions.
Outcome:	The proposed framework outperforms proprietary systems on 7B and 14B models by **5% on average, underscoring the effectiveness of interactive, agentic workflows for robust Text-to-SQL generation.

Copied to clipboard

Yuhan Wu, Huan Zhang, Wei Cheng, Chen Shen, Jingyue Yang, Wei Hu

Flow-Based Page Unique Semantic Mapping Architecture for Document Visual Question Answering (2026.acl-long)

Challenge:	Existing methods to improve code translation depend on abundant parallel code of high quality, which may not always be available.
Approach:	They propose a method that leverages functional invariance and cross-lingual portability of test suites to serve as universal verification oracles for multilingual reinforcement learning.
Outcome:	The proposed method leverages functional invariance and cross-lingual portability of test suites to serve as universal verification oracles for multilingual reinforcement learning (RL) training.

Copied to clipboard

Haosen Wang, Jing Xiao, Chaochao Du, Xiaowang Zhang, Zhiyong Feng

Pub-LawBench: Public-Oriented Benchmarking for LegalAI (2026.acl-long)

Challenge:	Document Visual Question Answering (DocVQA) aims to generate answers by understanding textual, layout, and visual elements within document images.
Approach:	They propose a Flow-Based Page Unique Semantic Mapping Architecture to solve the distinguishability problem among semantically similar pages.
Outcome:	The proposed model outperforms existing methods in evidence localization and answer generation.

Copied to clipboard

Qiaoyu Zheng, Zehan Ma, Yijing Zhang, Qiqi Wang, Huijia Li, Qian Liu

Challenge:	Existing evaluation frameworks focus on legal professionals, not legal professionals.
Approach:	They propose a public-oriented LegalAI benchmark grounded in legal functionalism and genre analysis to address this gap.
Outcome:	The proposed model evaluates 17 large language models on Pub-LawBench using simple prompts and Chain-of-Thought under a vanilla inference setting.

SWAN: Semantic Watermarking with Abstract Meaning Representation (2026.acl-long)

Copied to clipboard

Ziping Ye, Gourab Dey, Christos Christodoulopoulos, Charith Peris, Anil Ramakrishna, Weitong Ruan, Aram Galstyan, Kai-Wei Chang, Rahul Gupta, Ninareh Mehrabi

Turning Failures into Value: Negative Experience Replay for RLVR via Confidence Gating and Boundary Failure Sampling (2026.acl-long)

Challenge:	Existing methods to embed signatures by adjusting token selection preferences during text generation are highly sensitive to paraphrasing and synonyms.
Approach:	They propose a framework that embeds watermark signatures into the semantic structure of a sentence using Abstract Meaning Representation (AMR).
Outcome:	Empirical evaluation shows SWAN matches state-of-the-art detection performance on unaltered watermarked text while improving robustness against paraphrasing.

Copied to clipboard

Jialiang Guo, Fucheng Xiong, Xu He, Haodong Zhao, Xingyang li, Ke Zeng, Xunliang Cai

LazyEviction: Lagged KV Eviction with Attention Pattern Observation for Efficient Long Reasoning (2026.acl-long)

Challenge:	Existing experience replay methods for RLVR ignore sample inefficiency . expensive reasoning trajectories are discarded immediately after a single gradient update .
Approach:	They propose a method to replay failure trajectories to improve model refinement . they propose 'nexGRPO' which employs mid-confidence gating to filter invalid noise and saturated errors.
Outcome:	The proposed model outperforms strong baaselines and achieves improved out-of-distribution generalization.

Copied to clipboard

Haoyue Zhang, Hualei Zhang, Xiaosong Ma, Jie Zhang, Song Guo

Diversity in Unity, Theory in Practice: Hierarchical Multitask Benchmarks for Chinese Minority Languages (2026.acl-long)

Challenge:	Existing KV cache compression methods mitigate memory bottlenecks but struggle in long reasoning tasks.
Approach:	They propose a lagged eviction framework that prioritizes evicts based on tokens’ recurrence patterns to reduce KV cache by 50% and maintain comparable accuracy.
Outcome:	The proposed framework reduces KV cache by 50% 70% while maintaining comparable accuracy, outperforming existing KV baselines.

Copied to clipboard

Yijie Li, Xi Cao, Yuan Sun, Quulgan Minggad, Abdulla Ablikim, Jia Qing Cai Wang

LangSAE Editing: Improving Multilingual Information Retrieval via Post-hoc Language Identity Removal (2026.acl-long)

Challenge:	CMiLBench is a framework to evaluate linguistically and culturally diverse minority languages . rapid evolution of LLMs has revolutionized NLP, but progress is unevenly distributed .
Approach:	They propose a framework to translate a theoretical notion of "diversity in unity" into practical evaluation for three minority languages . CMiLBench comprises 24,663 instances across 5 difficulty levels and 17 tasks .
Outcome:	The proposed framework evaluates 14 state-of-the-art LLMs with a hybrid framework . it integrates automatic metrics and LLM-as-a-Judge scoring .

Copied to clipboard

Dongjun Kim, Jeongho Yoon, Chanjun Park, Heuiseok Lim

Interpretable Traces, Unexpected Outcomes: Investigating the Disconnect in Trace-Based Knowledge Distillation (2026.acl-long)

Challenge:	Existing methods for dense retrieval in multilingual environments encode language identity alongside semantics.
Approach:	They propose a method that trains on pooled embeddings to remove language-identity signal directly in vector space.
Outcome:	The proposed method improves ranking quality and cross-language coverage across multiple languages with especially strong gains for script-distinct languages.

Copied to clipboard

Siddhant Bhambri, Upasana Biswas, Subbarao Kambhampati

SpecMind: Cognitively Inspired, Interactive Multi-Turn Framework for Postcondition Inference (2026.acl-long)

Challenge:	Recent advances in reasoning-oriented Large Language Models have been driven by the introduction of Chain-of-Thought (CoT) traces.
Approach:	They propose to use CoT traces to guide model inference and serve as supervision signals for Knowledge Distillation to improve smaller models.
Outcome:	The proposed model is based on a rule-based problem decomposition method and is valid for both semantic correctness and interpretability to the end user.

Copied to clipboard

Cuong Chi Le, Minh V.t. Pham, Tung D. Vu, Van Duc Cuong, Phan Nhat Huy, Phan Nhat Hoang, Tien N. Nguyen

Memorization, Emergence, and Explaining Reversal Failures: A Controlled Study of Relational Semantics in LLMs (2026.acl-long)

Challenge:	Existing methods for generating specifications are limited and often fail to infer semantic specifications such as pre-/postconditions.
Approach:	They propose a framework that treats LLMs as exploratory reasoners rather than one-shot generators.
Outcome:	The proposed framework outperforms state-of-the-art methods in accuracy and completeness of generated postconditions.

Copied to clipboard

Yihua Zhu, Qianying Liu, Jiaxin Wang, Fei Cheng, Chaoran Liu, Akiko Aizawa, Sadao Kurohashi, Hidetoshi Shimodaira

Framing Political Bias in Multilingual LLMs Across Pakistani Languages (2026.acl-long)

Challenge:	Autoregressive LLMs perform well on relational tasks that require linking entities via relational words, but it is unclear whether they learn the logical semantics of such relations or whether left-to-right order bias is involved.
Approach:	They propose a framework that generates text from symmetric/inverse triples and trains autoregressive models from scratch.
Outcome:	The proposed framework generates text from symmetric/inverse triples, trains autoregressive models from scratch, and evaluates memorization, logical inference, and in-context generalization to unseen entities.

Copied to clipboard

Afrozah Nadeem, Mark Dras, Usman Naseem

InsAT: Instance-aware Semantic Alignment and Transfer from Human–Object Keypoints for Zero-to-Few-shot Action Understanding (2026.acl-long)

Challenge:	Large Language Models (LLMs) shape public discourse, yet most evaluations of economic and political bias focus on high-resource Western languages and contexts.
Approach:	They propose to use a culturally adapted Political Compass Test to evaluate political bias in 13 state-of-the-art LLMs across five Pakistani languages.
Outcome:	The proposed framework captures ideological stance (economic/social axes) and stylistic framing (content, tone, emphasis) in 13 state-of-the-art LLMs across five Pakistani languages.

Copied to clipboard

Kazuki Tsutsukawa

Multi-View Attention Multiple-Instance Learning Enhanced by LLM Reasoning for Cognitive Distortion Detection (2026.acl-long)

Challenge:	Existing zero-shot (ZS) approaches emphasize human motion while underutilizing contextual information, particularly human–object interactions.
Approach:	They propose a framework for ZS recognition and zero-to-few-shot adaptation that leverages instance-level language descriptions.
Outcome:	The proposed framework outperforms keypoint-based ZS methods while remaining data-efficient and robust.

Copied to clipboard

Jun Seo Kim, Hyemi Kim, Woo Joo OH, Hongjin Cho, Hochul Lee, Hye Hyeon Kim

CHAIRO: Contextual Hierarchical Analogical Induction and Reasoning Optimization for LLMs (2026.acl-long)

Challenge:	Cognitive distortions are systematic errors in thinking that occur when individuals perceive and interpret external information, leading to a negative conclusion that does not correspond to reality.
Approach:	They propose a framework that combines Large Language Models with a Multiple-Instance Learning architecture to enhance interpretability and expression-level reasoning.
Outcome:	The proposed framework improves interpretability and expression-level reasoning on Korean and English datasets.

Copied to clipboard

Haotian Lu, Yuchen Mou, Bingzhe Wu

Test of Time: Rethinking Temporal Signal of Benchmark Contamination (2026.acl-long)

Challenge:	Recent advances in large language models (LLMs) have enabled more sophisticated content moderation, but these methods lack generalization, interpretability, and adaptability to unseen or ambiguous cases.
Approach:	They propose a new moderation framework that leverages analogical examples to enhance rule induction and decision reliability.
Outcome:	The proposed method outperforms rule-injected fine-tuning baselines and multi-stage static RAG pipelines in terms of moderation accuracy and rule quality.

Copied to clipboard

Terry Jingchen Zhang, Gopal Dev, Ning Wang, Max Obreiter, Punya Syon Pandey, Keenan Samway, Wenyuan Jiang, Yinya Huang, Bernhard Schölkopf, Mrinmaya Sachan, Zhijing Jin

Challenge:	Existing work on benchmarks containing publicly available information has been interpreted as a temporal signal for benchmark contamination.
Approach:	They show that LLM-transformed questions can produce remarkably different temporal patterns compared to fill-in-the-blank questions directly retrieved from the very same documents.
Outcome:	The proposed model can produce different temporal patterns compared to fill-in-the-blank questions retrieved from the same documents.

An Existence Proof for Neural Language Models That Can Explain Garden-Path Effects via Surprisal (2026.acl-long)

Copied to clipboard

Ryo Yoshida, Shinnosuke Isono, Taiga Someya, Yohei Oseki, Tatsuki Kuribayashi

How Context Shapes Truth: Geometric Transformations of Statement-level Truth Representations in LLMs (2026.acl-long)

Challenge:	Surprisal theory claims that difficulty of sentences increases linearly with surprise . a neural LM that can explain garden-path effects cannot be built, says a new study .
Approach:	They propose to fine-tune neural LMs to better align surprisal-based reading-time estimates with actual reading times.
Outcome:	a new study shows that fine-tuned neural LMs do not overfit on held-out items . the results show that they improve predictive power for human reading times .

Copied to clipboard

Shivam Adarsh, Maria Maistro, Christina Lioma

Mitigating Tokenization-Induced Distance Distortion in Long-Context Multilingual Machine Translation (2026.acl-long)

Challenge:	Prior work shows that large language models encode whether a statement is true as a vector in residual stream activations.
Approach:	They study how truth vectors change when context is introduced in Large Language Models . they measure directional change between truth vector with and without context and relative magnitude of truth vector upon adding context.
Outcome:	The results show that large models distinguish relevant from irrelevant context mainly through directional change ()

Copied to clipboard

Khotso Selialia, Antoine Nzeyimana, Fatima M. Anwar

Probing Audio-Visual Reasoning in Multimodal Language Models through the Lens of Audio (2026.acl-long)

Challenge:	Existing positional encodings rely on fixed token indices and implicitly assume uniform semantic density, which breaks down for long-context inputs.
Approach:	They propose a tokenization-aware adaptive positional encoding that conditions relative positional bias on input-level sequence length and fragmentation statistics.
Outcome:	The proposed model improves long-context robustness and accuracy over baselines.

Copied to clipboard

Kaixiong Gong, Kaituo Feng, Bohao Li, Yibing Wang, Mofan Cheng, Shijia Yang, Jiaming Han, Benyou Wang, Yutong Bai, Zhuoran Yang, Xiangyu Yue

Reasoning Hijacking: The Fragility of Reasoning Alignment in Large Language Models (2026.acl-long)

Challenge:	Recent multimodal large language models lack robust audio-visual integration ability and performance on DeafTest is highly correlated with AV-Odyssey accuracy.
Approach:	They propose a benchmarking tool that integrates audio-visual reasoning with audio-video cues to infer solutions.
Outcome:	The proposed model performs well on DeafTest, but lacks audio perception in simple audio tasks.

Copied to clipboard

Yuansen Liu, Yixuan Tang, Anthony Kum Hoe Tung

The Dominance of Text Space: Unveiling the Asymmetric Nature of Cross-Modal Alignment in Large Language Models (2026.acl-long)

Challenge:	Current LLM safety research focuses on mitigating Goal Hijacking, preventing attackers from redirecting a model’s high-level objective.
Approach:	They propose a new adversarial prompt attack paradigm that subverts model judgments by injecting spurious decision criteria without altering the high-level task goal.
Outcome:	The proposed model subverts model judgments by injecting spurious decision criteria without altering the high-level task goal.

Copied to clipboard

Linqing Chen, Hanmeng Zhong, Wentao Wu, Peng Zhou

Towards Explainable Diagnosis: A Self-learned Explanatory Knowledge Base Approach (2026.acl-long)

Challenge:	Existing methods for cross-modal alignment assume a symmetric interaction between visual and textual modalities, implying that both spaces adapt to each other.
Approach:	They propose a method that regularizes the projector to maintain the geometric structure of the text embedding space via spectral filtering.
Outcome:	The proposed method preserves the LLM’s inherent linguistic capabilities and reduces object hallucination significantly better than standard fine-tuning methods.

Copied to clipboard

Dongqi Huang, Tong Zhou, Zhuoran Jin, Shenghui Shi, null Maoyujiao, Kang Liu, Jun Zhao, Yubo Chen

WildFeedback: Aligning LLMs With In-situ User Interactions And Feedback (2026.acl-long)

Challenge:	Large language models (LLMs) have great potential to facilitate explainable diagnosis, but their effectiveness is often constrained by insufficient diagnostic expertise.
Approach:	They propose a unified LLM-based framework for faithful and explainable diagnosis that builds a high-quality diagnostic knowledge base through a record-driven explanation learning paradigm.
Outcome:	The proposed framework outperforms baselines on the DiReCT and JAMA benchmarks and improves the explanation completeness metric from 64.5% to 76.9% over the best existing methods.

Copied to clipboard

Taiwei Shi, Zhuoer Wang, Longqi Yang, Ying-Chun Lin, Zexue He, Mengting Wan, Pei Zhou, Sujay Kumar Jauhar, Sihao Chen, Shan Xia, Hongfei Zhang, Jieyu Zhao, Xiaofeng Xu, Xia Song, Jennifer Neville

Challenge:	Traditional alignment methods rely on human annotations and are subjective and misalignment with real-world user preferences.
Approach:	They propose a framework that leverages in-situ user feedback during conversations with LLMs to create preference datasets automatically.
Outcome:	The proposed framework identifies and classifies user feedback to LLM responses between conversation turns and creates examples of preferred and dispreferred responses according to user preferences.

SSG: Logit-Balanced Vocabulary Partitioning for LLM Watermarking (2026.acl-long)

Copied to clipboard

Chenxi Gu, Xiaoning Du, John C. Grundy

PaT: Planning-after-Trial for Efficient Test-Time Code Generation (2026.acl-long)

Challenge:	Large language models can generate highquality, human-like content, but they also pose risks such as infringement of proprietary interests, misuse of outputs, and spread harmful misinformation.
Approach:	They propose a method that partitions the vocabulary into two logit-balanced subsets and lifts the lower bound of watermark strength for each token prediction.
Outcome:	The proposed method lifts the lower bound of watermark strength for each token prediction, thereby improving watermark detectability.

Copied to clipboard

Youngsik Yoon, Sungjae Lee, Seockbean Song, Siwei Wang, Wei Chen, Jungseul Ok

Joint Knowledge Base Completion and Question Answering by Combining Large Language Models and Small Language Models (2026.acl-long)

Challenge:	Existing methods for scaling test-time computation are rigid and inefficient . a heterogeneous configuration achieves performance comparable to a large homogeneously model .
Approach:	They propose an adaptive planning policy that invokes a planner only upon verification failure.
Outcome:	The proposed model achieves comparable performance to a large homogeneous model while reducing inference cost by approximately 69% across multiple benchmarks and model families.

Copied to clipboard

Yinan Liu, Dongying Lin, Sigang Luo, Xiaochun Yang, Bin Wang

Rethinking Reasoning-Intensive Retrieval: Evaluating and Advancing Retrievers in Agentic Search Systems (2026.acl-long)

Challenge:	Existing studies rely on the small language model (SLM) to enhance them jointly, and the large language model’s strong reasoning ability is ignored.
Approach:	They propose a framework which can make knowledge base completion and knowledge base question answering enhance each other in an iterative manner by combining the strengths of the small language model and the large language model.
Outcome:	The proposed framework surpasses baselines for both KBC and KBQA tasks over two public benchmark data sets.

Copied to clipboard

Yilun Zhao, Jinbiao Wei, Tingyu Song, Siyue Zhang, Chen Zhao, Arman Cohan

The Role of Mixed-Language Documents for Multilingual Large Language Model Pretraining (2026.acl-long)

Challenge:	Existing evaluation benchmarks for retrievers are narrow and evaluate them in isolation . existing evaluation benchmarking frameworks focus on evaluating retrievers in isolation, obscuring their value in real-world applications.
Approach:	They propose an evaluation framework that evaluates retrievers in agentic search systems . they provide expert-annotated reasoning aspects, positive documents, a reference response and evaluation rubrics .
Outcome:	The proposed framework assesses retrievers in agentic search systems.

Copied to clipboard

Jiandong Shao, Raphael Tang, Crystina Zhang, Karin Sevegnani, Pontus Stenetorp, Jianfei Yang, Yao Lu

ARCHITECT: Uncertainty-Aware Dynamic Tool Learning via Causal Intervention for Open-World Agents (2026.acl-long)

Challenge:	Existing research suggests that multilingual large language models can achieve impressive cross-lingual understanding despite largely monolingual pretraining.
Approach:	They compare a monolingual-only corpus with a standard web corpus that removes all multilingual documents and then retrain the models from scratch under controlled conditions.
Outcome:	The results show that removing bilingual data causes translation performance to drop 56% in BLEU, whereas code-switching contributes minimally.

Copied to clipboard

Zhangyi Wang, Jiexiang Xu, Bingnan Yu, Zongze Li

Challenge:	Existing methods treat all generated tools as equally trustworthy, a "blind trust" assumption that is untenable for reliable agent deployment.
Approach:	They propose a framework that moves beyond black-box reliability prediction to interpretable failure attribution.
Outcome:	The proposed framework achieves state-of-the-art on four benchmarks including StableToolBench, MINT, T-Eval, and SWE-bench Lite.

ES4R: Speech Encoding Based on Prepositive Affective Modeling for Empathetic Response Generation (2026.acl-long)

Copied to clipboard

Zhuoyue Gao, Xiaohui Wang, Xiaocui Yang, Wen Zhang, Daling Wang, Shi Feng, Yifei Zhang

MAGMA: A Multi-Graph based Agentic Memory Architecture for AI Agents (2026.acl-long)

Challenge:	Existing speech-to-speech large language models rely on ASR transcription or use encoders to extract latent representations, weakening affective information and contextual coherence in multi-turn dialogues.
Approach:	They propose a framework for speech-based empathetic response generation that captures turn-level affective states and dialogue-level emotional dynamics.
Outcome:	The proposed framework outperforms baselines in automatic and human evaluations and remains robust across different Large Language Model (LLM) backbones.

Copied to clipboard

Dongming Jiang, Yi Li, Guanpeng Li, Bingzhe Li

The Side Effects of Being Smart: Safety Risks in MLLMs’ Multi-Image Reasoning (2026.acl-long)

Challenge:	Existing approaches rely on semantic similarity over monolithic memory stores, entangling temporal, causal, and entity information.
Approach:	They propose a multi-graph agentic memory architecture that represents each memory item across orthogonal semantic, temporal, causal, and entity graphs.
Outcome:	Experiments on LoCoMo and LongMemEval show that MAGMA outperforms state-of-the-art models in long-horizon reasoning task.

Copied to clipboard

Renmiao Chen, Yida Lu, Shiyao Cui, Xuan Ouyang, Victor Shea-Jay Huang, Shumin Zhang, Chengwei Pan, Han Qiu, Minlie Huang

Action Boundary Blindness: When LLM Agents Cannot Tell Where One Action Ends and Another Begins (2026.acl-long)

Challenge:	Recent advances in multimodal reasoning may pose new safety risks . evaluators neglect reasoningbased safety, where harm emerges only through MLLMs .
Approach:	They introduce a benchmark for multi-image reasoning safety that includes 2,676 instances . they find that models with more advanced multi- image reasoning are more vulnerable .
Outcome:	The proposed benchmark consists of 2,676 instances covering 9 multi-image relations . the results show that models with more advanced multi- image reasoning are more vulnerable .

Copied to clipboard

Zhangyi Wang, Bingnan Yu, Jiexiang Xu, Zongze Li

Challenge:	Large language model agents exhibit action boundary blindness, granularity confusion, scope creep and boundary ambiguity . Explicit boundary prompting improves ABS by 0.08–0.13 across all models .
Approach:	They propose four automatic metrics that require no human annotation to detect boundary blindness . they propose to use a multi-label attribution framework to validate the models .
Outcome:	Experiments with seven large language model agents show that the best model achieves only 0.424 ABS . Explicit Boundary Prompting improves ABS by 0.08–0.13 across all models .

Verbal-R3: Verbal Reranker as the Missing Bridge between Retrieval and Reasoning (2026.acl-long)

Copied to clipboard

Sangkwon Park, Donghun Kang, Jisoo Mok, Sungroh Yoon

IGenBench: Benchmarking the Reliability of Text-to-Infographic Generation (2026.acl-long)

Challenge:	Existing paradigms of Retrieval-Augmented Generation (RAG) are suboptimal due to exposure bias, a mismatch between pre-training data distribution and retrieved information.
Approach:	They propose to bridge retrieval results and the LLM’s reasoning ability through Verbal Annotations, analytic narratives that explicitly articulate the logical connection between a search query and retrieved contexts.
Outcome:	The proposed framework achieves state-of-the-art performance on complex Question Answering benchmarks validating the effectiveness of the proposed framework.

Copied to clipboard

Yinghao Tang, Xueding Liu, Boyuan Zhang, Tingfeng Lan, Yupeng Xie, Jiale Lao, Yiyao Wang, Haoxuan Li, Tingting Gao, Bo Pan, Luoxuan Weng, Xiuqi Huang, Minfeng Zhu, Yingchaojie Feng, Yuyu Luo, Wei Chen

ProMedical: Hierarchical Fine-Grained Criteria Modeling for Medical LLM Alignment via Explicit Injection (2026.acl-long)

Challenge:	Generated infographics may appear correct at first glance but contain easily overlooked issues, such as distorted data encoding or incorrect textual content.
Approach:	They propose to evaluate reliability of text-to-infographic generation using IGenBench . they employ multimodal large language models to verify each question .
Outcome:	The proposed framework decomposes reliability verification into atomic yes/no questions based on a taxonomy of 10 question types.

Copied to clipboard

He Geng, Yangmin Huang, Lixian Lai, Qianyun Du, Hui Chu, Zhiyang He, Jiaxue Hu, Xiaodong Tao

Challenge:	Large Language Models (LLMs) are difficult to align with high-stakes medical standards due to dissonance between coarse-grained preference signals and complex protocols.
Approach:	They propose a framework that aligns Large Language Models with medical standards . they use a dataset generated via a human-in-the-loop pipeline to augment medical instructions .
Outcome:	The proposed framework disentangles safety constraints from general proficiency, enabling precise guidance during reinforcement learning.

R2IF: Aligning Reasoning with Decisions via Composite Rewards for Interpretable LLM Function Calling (2026.acl-long)

Copied to clipboard

Aijia Cheng, Kailong Wang, Ling Shi, Yongxin Zhao

From Detection to Understanding: Multi-Turn Reasoning for Video Misinformation Analysis (2026.acl-long)

Challenge:	Existing RL-based approaches to function calling are misaligned between reasoning processes and tool-call decisions.
Approach:	They propose a reasoning-aware RL framework for interpretable function calling . they integrate a composite reward integrating format/correctness constraints, CER, and SMV .
Outcome:	Experiments on BFCL/ACEBench show R2IF outperforms baselines by 34.62% with positive Average CoT Effectiveness.

Copied to clipboard

Zhi Zeng, Jiaying Wu, Minnan Luo, Di Zhang, Yifei Yang, Xiangzheng Kong, Herun Wan, Zihan Ma

Frozen LLMs are Native Decoders for High-Norm Semantic Vectors (2026.acl-long)

Challenge:	Existing benchmarks focus on binary veracity judgments and do not evaluate process-level justifications for misinformation models.
Approach:	They propose a video misinformation analysis benchmark that assesses reasoning in video misinterpretation.
Outcome:	The proposed framework improves reasoning accuracy and explanation quality compared to existing models . it covers 12 fine-grained deception categories and progresses from perceptual attribution to intent and persuasion analysis.

Copied to clipboard

Yunsheng Zeng, Yongmei Tan

Enhancing Reinforcement Learning for Radiology Report Generation with Evidence-aware Rewards and Self-correcting Preference Learning (2026.acl-long)

Challenge:	Existing compression methods selectively prune tokens based on information-theoretic metrics, resulting in interpretability but risking the loss of fine-grained information.
Approach:	They propose a landmark-based compression framework for long contexts that captures global dependencies over landmark tokens.
Outcome:	The proposed framework outperforms soft compression baselines on four QA benchmarks.

Copied to clipboard

Qin Zhou, Guoyan Liang, Qianyi Yang, Jingyuan Chen, Sai Wu, Chang Yao, Zhe Wang

VisAidMath: Benchmarking Visual-Aided Mathematical Reasoning (2026.acl-long)

Challenge:	Recent reinforcement learning approaches have advanced radiology report generation (RRG) however, there are two limitations: report-level rewards offer limited evidence-grounded guidance for clinical faithfulness .
Approach:	They propose a method that uses group-wise evidence-aware alignment rewards and self-correcting preference learning to build a reliable, disease-agnostic preference dataset without human supervision.
Outcome:	ESC-RL promotes clinically faithful, disease-aligned reward and supports continual self-improvement during training.

Copied to clipboard

Jingkun Ma, Runzhe Zhan, Yang Li, Di Sun, Hou Pong Chan, Lidia S. Chao, Derek F. Wong

RISK: A Framework for GUI Agents in E-commerce Risk Management (2026.acl-long)

Challenge:	Existing Large Multi-modal Models lack a robust visual processing capability that is often masked by evaluation metrics that prioritize final-answer accuracy.
Approach:	They propose a three-layer evaluation framework that scrutinizes the generation of valid visual aids and the soundness of subsequent reasoning steps.
Outcome:	The proposed framework examines the generation of valid visual aids and the soundness of subsequent reasoning steps on state-of-the-art models.

Copied to clipboard

Renqi Chen, Zeyin Tao, Jianming Guo, Jingzhe Zhu, Yiheng Peng, Qingqing Sun, Tianyi Zhang, Shuai Chen

Challenge:	RISK is a framework designed to automate multi-step web interactions in e-commerce risk management.
Approach:	a new framework is designed to build and deploy GUI agents for e-commerce risk management . RISK-R1 provides a scalable, domain-specific solution for automating complex web interactions .
Outcome:	RISK provides a scalable, domain-specific solution for automating complex web interactions in e-commerce risk management.

MiniRAG: A Lightweight RAG system with Small Language Models (2026.acl-long)

Copied to clipboard

Tianyu Fan, Jingyuan Wang, Xubin Ren, Chao Huang

Glyph: Scaling Context Windows via Visual-Text Compression (2026.acl-long)

Challenge:	Existing RAG frameworks rely on Large Language Models (LLMs) for all stages of the process, resulting in high computational costs and resource demands.
Approach:	They propose a semantic-aware heterogeneous graph indexing mechanism that combines text chunks and named entities in a unified structure and a lightweight topology-enhanced retrieval approach that leverages graph structures for efficient knowledge discovery without requiring advanced language capabilities.
Outcome:	The proposed system achieves comparable performance to LLM-based methods while requiring only 25% of the storage space.

Copied to clipboard

Jiale Cheng, Yusen Liu, Xinyu Zhang, Yulin Fei, Wenyi Hong, Ruiliang Lyu, Weihan Wang, Zhe Su, Xiaotao Gu, Xiao Liu, Yushi Bai, Jie Tang, Hongning Wang, Minlie Huang

TA-GRPO-d: Trajectory-Aware GRPO for Optimizing Denoising Trajectories in Diffusion LLMs (2026.acl-long)

Challenge:	Large language models (LLMs) traditionally represent text as sequences of discrete tokens . a long-context scaling problem requires processing more tokens more efficiently .
Approach:	They propose a framework that renders long texts into compact visual pages and processes them with a vision-language model.
Outcome:	The proposed framework renders long texts into compact visual pages and processes them with a vision-language model.

Copied to clipboard

Gyunyeop Kim, Sangwoo Kang

AG-GRPO: Answer-Guided GRPO for Masked Diffusion Language Models (2026.acl-long)

Challenge:	Existing dLLMs rely on fixed denoising schedules and cannot learn efficient unmasking orders.
Approach:	They propose a framework that transforms dLLM decoding into a trajectory-aware policy . it uses a confidence-gated denoising strategy that decides which tokens to unmask .
Outcome:	The proposed model can learn which tokens to unmask and how many to unmak per step . it can learn the output quality and efficiency of the decoding path itself .

Copied to clipboard

Juhyeong Kim, Gyunyeop Kim, Sangwoo Kang

Grammar as Control: Modular Language Generation for the Long Tail (2026.acl-long)

Challenge:	Recent work on large language models (LLMs) has emphasized not only final-answer accuracy but also reliability of reasoning on challenging tasks.
Approach:	They propose an answer-guided group-relative policy optimization for masked diffusion language models which generates text through iterative mangled token restoration.
Outcome:	The proposed approach improves over pretrained dLLMs and prior RL methods across mathematics, puzzle-solving, and code-generation benchmarks.

Copied to clipboard

Ndapa Nakashole

MTR-Suite: A Framework for Evaluating and Synthesizing Conversational Retrieval Benchmarks (2026.acl-long)

Challenge:	Large language models (LLMs) can bootstrap language technologies for long-tail languages . however, without structured guidance, they produce narrow, unrepresentative samples .
Approach:	They propose a framework that transforms descriptive grammars into explicit control mechanisms that guide LLMs to generate typologically balanced synthetic data for downstream training.
Outcome:	The proposed framework improves typological entropy and yields a "student-beats-teacher" effect across three low-resource languages.

Copied to clipboard

Junhao Ruan, Abudukeyumu Abudula, Bei Li, Yongjing Yin, Xinyu Liu, Kechen Jiao, Xin Chen, Jingang Wang, Xunliang Cai, Tong Xiao, JingBo Zhu

Beyond Majority Voting: Towards Fine-grained and More Reliable Reward Signal for Test-Time Reinforcement Learning (2026.acl-long)

Challenge:	Existing conversational retrieval benchmarks suffer from costly, sparse human annotation or rigid, unnatural automated heuristics.
Approach:	They propose a framework for auditing, synthesizing, and benchmarking conversational retrieval.
Outcome:	The proposed framework is based on three LLM-based auditors and a multi-agent system . it mimics production-style challenges (hard topic switching, verbosity) and offers superior discriminative power.

Copied to clipboard

Weiqin Wang, Yile Wang, Kehao Chen, Hui Huang

Beyond Markovian Forgetfulness: Episodic Memory for Reasoning-Intensive Retrieval (2026.acl-long)

Challenge:	RLVR is a paradigm for improving reasoning ability of large language models . but voting results often induce confirmation bias and suffer from sparse rewards .
Approach:	They propose a framework integrating model confidence and dynamic subgroup partitioning to address these issues.
Outcome:	The proposed framework outperforms recent baselines on multiple models and benchmarks.

Copied to clipboard

Dohyeon Lee, Yeonseok Jeong, Seung-won Hwang

DUAL RM: Beyond Rule-based Preference Reward Modeling via Meta-Reward (2026.acl-long)

Challenge:	Existing methods for reasoning-intensive information retrieval suffer from inefficiency . Chain-of-Thought (CoT) approaches suffer from lack of token efficiency . Existing models lack episodic memory, which stores the history of prior states .
Approach:	They propose an algorithm that enhances state-based frameworks with an episodic memory module that stores the full history of prior states for a query.
Outcome:	The proposed model outperforms CoT and state-based models on the BRIGHT benchmark and is highly token-efficient.

Copied to clipboard

Xiaobo Liang, Wanfu Wang, Qipeng Huang, Yuyang Ding, Zecheng Tang, Yixin Ji, Qianben Chen, Zhe Zhao, Kehai Chen, Juntao Li, Min Zhang

Challenge:	Existing preference-based reward modeling methods face a recursive dependency where each verifier requires a meta-verifier, leading to continuous and costly dependence on human annotation.
Approach:	They propose a dual RM that couples discriminative and generative reward models under a non-parametric meta-reward.
Outcome:	The proposed model achieves strong performance across major preference benchmarks and even when trained exclusively on language modality, it exhibits robust cross-modal transfer on Omni-RewardBench.

Evolutionary Negative Module Pruning for Better LoRA Merging (2026.acl-long)

Copied to clipboard

Anda Cao, Zhuo Gou, Yi Wang, Kaixuan Chen, Yu Wang, Can Wang, Mingli Song, Jie Song

Example Quality Matters: Multi-Aspects Example Augmentation for Private Library Programming (2026.acl-long)

Challenge:	Existing methods for integrating multiple low-rank Adaptation experts into a single backbone are limited by negative modules.
Approach:	They propose a plug-and-play LoRA pruning method to locate and exclude negative modules prior to merging.
Outcome:	The proposed method boosts the performance of existing merging algorithms across languages and vision domains.

Copied to clipboard

Yuhao Li, Haifeng Sun, Xuesong Zhang, Shu Yao, Haoyu Zheng, Yvchuan Wang, Huazheng Wang, Zirui Zhuang, Qi Qi, Jianxin Liao, Jingyu Wang

Resonating with RoPE: Spectral Quantization for High-Fidelity Key Cache Compression (2026.acl-long)

Challenge:	Existing approaches to code generation fail to consider the quality of retrieved examples.
Approach:	They propose a retrieval-augmented generation method that combines existing API examples to improve complexity and readability.
Outcome:	The proposed method achieves up to 22% accuracy improvement over baseline methods.

Copied to clipboard

Xuefei Wang, Haoyu Tang, Tianyuan Liang, Zhibin Wang, Yupeng Hu, Weili Guan

Challenge:	Existing approaches to reduce memory footprint of long-context LLMs rely on RoPE-induced oscillations.
Approach:	They propose a frequency-domain framework that converts RoPE-induced oscillations into sparse spectral representations.
Outcome:	The proposed framework achieves efficient compression with performance comparable to FP16 benchmarks.

CSRP: Chain-of-Thought Reasoning for Chinese Text Correction via Reinforcement Learning with Efficiency-Aware Rewards (2026.acl-long)

Copied to clipboard

Wei Tian, Yuhao Zhou, Man Lan

Adaptive Retrieval for Reasoning (2026.acl-long)

Challenge:	Large Language Models lack specialized priors for subtle grammatical distinctions, and Supervised Fine-Tuning fails to optimize for precision-focused metrics.
Approach:	They propose a framework that builds correction capability through Continual Pre-training on 5.9M balanced samples to internalize domain knowledge.
Outcome:	The proposed framework outperforms existing models on the NACGEC benchmark with 50.99 F0.5 and 57.17 precision while mitigating over-correction bias.

Copied to clipboard

Jongho Kim, Jaeyoung Kim, Jihyuk Kim, Yu Jin Kim, Seung-won Hwang, Moontae Lee

On Emergent Social World Models — Evidence for Functional Integration of Theory of Mind and Pragmatic Reasoning in Language Models (2026.acl-long)

Challenge:	Existing reasoning-based rerankers suffer from bounded recall.
Approach:	They propose a framework that leverages adaptive retrieval to ensure sufficient "bridge" documents are retrieved for reasoning-intensive retrieval.
Outcome:	The proposed method outperforms baselines on reasoning-intensive retrieval tasks by 5.6%pt.

Copied to clipboard

Polina Tsvilodub, Jan-Felix Klumpp, Amir Pour, Jennifer Hu, Michael Franke

ToolOmni: Enabling Open-World Tool Use via Agentic learning with Proactive Retrieval and Grounded Execution (2026.acl-long)

Challenge:	Large language models (LMs) possess astonishing abilities and prove useful for a plethora of downstream tasks, but controversy persists regarding how to conceptualize their capacities.
Approach:	They analyze LMs’ performance across seven subcategories of ToM abilities using a large localizer dataset than used in prior work.
Outcome:	The proposed models recruit shared computational mechanisms for general Theory of Mind (ToM) and language-specific pragmatic reasoning on a substantially larger localizer dataset than used in prior work.

Copied to clipboard

Shouzheng Huang, Meishan Zhang, Baotian Hu, Min Zhang

From TDMA to CDMA: A Multi-bit Watermark for Diffusion Language Models (2026.acl-long)

Challenge:	Existing methods for large language models struggle to align user intent with tool semantics or generalize to unseen tools.
Approach:	They propose a framework that enables LLMs for open-world tool use by proactive retrieval and grounded execution within a reasoning loop.
Outcome:	The proposed framework surpasses baselines in retrieval and execution efficiency by +10.8%.

Copied to clipboard

Baizhou Huang, Xiaojun Wan

LegalGraphRAG: Multi-Agent Graph Retrieval-Augmented Generation for Reliable Legal Reasoning (2026.acl-long)

Challenge:	Existing multi-bit watermarking schemes cannot be directly applied to DLMs.
Approach:	They propose a multi-bit watermarking framework that encodes the entire watermark message holographically.
Outcome:	The proposed framework encodes the entire watermark message across all tokens holographically.

Copied to clipboard

Zerui Chen, Qinggang Zhang, Zhishang Xiang, Zhimin Wei, Linfeng Gao, Xiao Huang, Zhihong Zhang, Jinsong Su

Characterizing the Expressivity of Local Attention in Transformers (2026.acl-long)

Challenge:	Graph-based Retrieval-Augmented Generation (GraphRAG) is a new approach to document retrieval, but it is not suitable for legal reasoning.
Approach:	They propose a framework for reliable legal reasoning that structures knowledge as relational graphs and uses a multi-agent system to verify validity.
Outcome:	The proposed framework outperforms existing GraphRAG models in accurate and trustworthy legal analysis.

Copied to clipboard

Jiaoda Li, Ryan Cotterell

TH-RAG : Topic-Based Hierarchical Knowledge Graphs for Robust Multi-hop Reasoning in Graph-based RAG Systems (2026.acl-long)

Challenge:	Existing studies show that global and local attention are expressively complementary.
Approach:	They propose to restrict global attention to a fixed-size window of preceding tokens . they also propose to add local attention to local-only transformers to increase model quality .
Outcome:	The proposed model outperforms the global–local transformers on formal language recognition and natural language modeling.

Copied to clipboard

JungHyoun Kim, Soohyeong Kim, Seok Jun Hwang, Jeonghyeon Park, Yong Suk Choi

Beyond Static Benchmarks: Synthesizing Harmful Content via Persona-based Simulation for Robust Evaluation (2026.acl-long)

Challenge:	Retrieval-augmented generation (RAG) enables large language models to incorporate external knowledge at inference.
Approach:	They propose a hierarchical framework that organizes triplets into subtopics and topics to enhance connectivity and integrate dispersed information.
Outcome:	Experiments on abstractive and specific QA benchmarks show that TH-RAG outperforms strong baselines in accuracy and robustness while remaining efficient.

Copied to clipboard

Huije Lee, Jisu Shin, Hoyun Song, Changgeon Ko, Jong C. Park

DiVE: Decoupling Intra-layer Visual Evidence for Mitigating Hallucinations in Large Vision-Language Models (2026.acl-long)

Challenge:	Existing static benchmarks for harmful content detection face limitations in scalability and diversity.
Approach:	They propose a framework for synthesizing harmful content using persona-guided large language model agents.
Outcome:	The proposed framework achieves a high success rate in harmful generation tests across multiple detection systems.

Copied to clipboard

Xinwei Li, Li Lin, Hui Jiao, Li Yao, Tien-Tsin Wong, Hanqian Wu

Impatient Users Confuse AI Agents: High-fidelity Simulations of Human Traits for Testing Agents (2026.acl-long)

Challenge:	Existing decoding-based approaches do not explicitly decouple visual evidence from mixed vision–language representations.
Approach:	They propose to decouple visual evidence from mixed vision–language representations by dynamically identifying layers enriched with visual information and performing intra-layer decoupling to extract aggregated visual evidence.
Outcome:	Experiments show that DiVE achieves state-of-the-art performance on multiple benchmarks.

Copied to clipboard

Muyu He, Anand Kumar, Soumyadeep Bakshi, James Zou, Nazneen Rajani

Efficient Process Reward Modeling via Contrastive Mutual Information (2026.acl-long)

Challenge:	Small shifts in user behavior can cause sharp drops in agent performance . prior work has shown that LLMs lack robustness to real-world noise and small input perturbations.
Approach:	They propose a model-agnostic method for systematically stress testing AI agents that learns directions in activation space corresponding to steerable user traits.
Outcome:	The proposed method can be used to stress test AI agents in airline, retail, telecom, and telehealth domains.

Copied to clipboard

Nakyung Lee, Sangwoo Hong, Jungwoo Lee

R3: End-to-End Reasoning-based Planning for Multi-step Retrosynthesis via Reinforcement Learning (2026.acl-long)

Challenge:	Existing methods to verify intermediate reasoning steps require human annotators to assign reward scores to each reasoning step, which is labor-intensive and costly.
Approach:	They propose a method that leverages the model's internal probability to infer step-level supervision while significantly reducing the computational burden of annotating dataset.
Outcome:	The proposed method reduces dataset construction time by 84% and token generation by 98% compared to MC estimation, while achieving higher accuracy on process-level evaluations and mathematical reasoning benchmarks.

Copied to clipboard

YiFei Wang, Qizhi Pei, Jiangtao Feng, Yuntian Shi, Yi Duan, Lihao Wang, Lei Bai, Lijun Wu, Wei-Ying Ma, Hao Zhou

River-LLM: Large Language Model Seamless Exit Based on KV Share (2026.acl-long)

Challenge:	Experimental results show that R3 is a superior alternative to traditional search algorithms for multistep retrosynthesis planning.
Approach:	They propose a framework that reformulates multistep retrosynthetic planning as a generative reasoning task.
Outcome:	The proposed framework achieves state-of-the-art Top-1 accuracy of 43.7% on retrobench . it leverages Large Language Models to reformulate multistep retrosynthesis as a generative reasoning task.

Copied to clipboard

Yingtao Shen, An Zou

Execution as Verification: Fine-Grained Self-Correcting Reasoning for Complex KBQA (2026.acl-long)

Challenge:	Existing methods to reduce latency and speed up early exits are costly and impose significant cost and energy consumption.
Approach:	They propose a lightweight KV-Shared Exit River framework that allows the backbone’s missing KV cache to be naturally generated and preserved during the exit process.
Outcome:	The proposed framework achieves 1.71 to 2.16 speedup while maintaining high generation quality.

Copied to clipboard

Minghan Zhang, Zhen Yang, Haodong Zou, Jie Chen, Zhen Duan, Shu Zhao

Zero-Shot Detection of LLM-Generated Text using Temperature Sensitivity (2026.acl-long)

Challenge:	Existing knowledge base question answering methods are limited by syntactic constraints and are prone to structural deviations that render queries unexecutable.
Approach:	They propose a framework that reframes semantic parsing as an iterative reasoning process driven by execution feedback.
Outcome:	The proposed method achieves significant improvements in query executability and answer accuracy on the WebQSP and CWQ datasets.

Copied to clipboard

Shixuan Ma, Jiahao Li, Zhendong Mao, Quan Wang

MARD: Module-Aware Reasoning Distillation for Language Models with Adaptive Supervision (2026.acl-long)

Challenge:	Existing methods for detecting LLM-generated text rely on statistical features that are insufficient for reliable detection.
Approach:	They propose a temperature-sensitive detector that modulates decoding temperature and monitors how probability distributions respond to temperature.
Outcome:	The proposed method is based on a temperature sensitivity feature and a simple zero-shot detector built upon normalized temperature sensitivity.

Copied to clipboard

Wenqi Yang, Jianjun Li, Zhibo Zhang, Mingqian Ding, Yushen Fang

Dashboard2Code: Evaluating Multimodal Models on Reconstructing Interactive Dashboards (2026.acl-long)

Challenge:	Multi-step reasoning remains challenging for language models with limited capacity . et al., 2025) demonstrate remarkable reasoning capabilities across diverse tasks .
Approach:	They propose a module-aware reasoning distillation framework that explicitly targets key Transformer components for effective reasoning transfer.
Outcome:	The proposed framework targets key components for effective reasoning transfer . it adopts an offline distillation setting, where a strong teacher model provides reasoning trajectories in advance .

Copied to clipboard

Tianhao Niu, Ziyu Han, Qiguang Chen, Shiqi Zhou, Baocai Shan, Hengjie Fang, Qingfu Zhu, Wanxiang Che

Can Compact Language Models Search Like Agents? Distillation-Guided Policy Optimization for Preserving Agentic RAG Capabilities (2026.acl-long)

Challenge:	Existing efforts to generate static visualizations focus on static charts and interactive dashboards.
Approach:	They propose a dashboard2code task that requires a model to explore an interactive dashboard, acquire feedback from its own interactions and generate code that reproduces the target dashboard.
Outcome:	The proposed task is based on 180 carefully designed and manually verified dashboard–code pairs spanning three difficulty levels and covering eight common real-world interaction patterns.

Copied to clipboard

Rikuto Kotoge, Mai Nishimura, Jiaxin Ma

CE-GPPO: Coordinating Entropy via Gradient-Preserving Clipping Policy Optimization in Reinforcement Learning (2026.acl-long)

Challenge:	Existing agentic RAG systems rely on large language models with billions of parameters.
Approach:	They propose a method to elicit agentic RAG behaviors from compact models . they propose ARC, which uses cold-start initialization and teacher guidance .
Outcome:	The proposed method outperforms the larger teacher model in some cases.

Copied to clipboard

Zhenpeng Su, Leiyu Pan, Minxuan Lv, Yuntao Li, Wenping Hu, Fuzheng Zhang, Kun Gai, Guorui Zhou

Omni-I2C: A Holistic Benchmark for High-Fidelity Image-to-Code Generation (2026.acl-long)

Challenge:	Existing methods for proximal policy optimization discard valuable gradient signals from low-probability tokens due to the clipping mechanism.
Approach:	They propose an algorithm that reintroduces gradients from clipped tokens in native PPO in a gentle and bounded manner.
Outcome:	The proposed algorithm outperforms strong baselines on reasoning benchmarks on different model scales.

Copied to clipboard

Jiawei Zhou, Chi Zhang, Xiang Feng, Qiming Zhang, Haibo Qiu, Lihuo He, Dengpan Ye, Xinbo Gao, Jing Zhang

Revealing Procedural Reasoning Structures in Chain-of-Thought Training via Span-Level Gradient Organization (2026.acl-long)

Challenge:	a benchmark is designed to evaluate the capability of Large Multimodal Models (LMMs) in converting complex, structured digital graphics into executable code.
Approach:	They propose a benchmark to evaluate the capability of Large Multimodal Models to convert digital graphics into executable code.
Outcome:	The proposed benchmark exposes the performance gap among leading LMMs . the benchmark features 1130 meticulously curated samples .

Copied to clipboard

Jia Liu, Jiaxin Luo, Weiwen Xu, Jonathan M. Garibaldi, Xiao-Kun Wu, Yixue Hao, Min Chen

StepHint: Multi-level Stepwise Hints Enhance Reinforcement Learning to Reason (2026.acl-long)

Challenge:	Chain-of-Thought (CoT) prompts elicit multi-step reasoning, yet how reasoning related structure is expressed during training remains poorly understood.
Approach:	They propose a framework that tracks span-level gradients during fine-tuning on reasoning benchmarks to understand how models develop structured, step-by-step reasoning capabilities.
Outcome:	The proposed framework tracks span-level gradients during fine-tuning on reasoning benchmarks to understand how models develop structured, step-by-step reasoning capabilities.

Copied to clipboard

Kaiyi Zhang, Ang Lv, Jinpeng Li, Yongbo Wang, Feng Wang, Haoyuan hu, Rui Yan

Social Dynamics as Critical Vulnerabilities that Undermine Objective Decision-Making in LLM Collectives (2026.acl-long)

Challenge:	Reinforcement learning with verifiable rewards (RLVR) approaches face two challenges: the near-miss reward problem and exploration stagnation.
Approach:	They propose an algorithm that partitions valid reasoning chains into reasoning steps using multi-level stepwise hints.
Outcome:	The proposed method outperforms competing RLVR enhancement methods across six mathematical benchmarks and two out-of-domain benchmarks.

Copied to clipboard

Changgeon Ko, Jisu Shin, Hoyun Song, Huije Lee, Eui Jun Hwang, Jong C. Park

Challenge:	Large language model (LLM) agents are increasingly acting as human delegates in multi-agent environments, where a representative agent integrates diverse peer perspectives to make a final decision.
Approach:	They define four key phenomena—social conformity, perceived expertise, dominant speaker effect, and rhetorical persuasion—and manipulate the number of adversaries, relative intelligence, argument length, and argumentative styles.
Outcome:	The results show that the reliability of the representative agent is undermined by the social context of its network.

FairQE: Multi-Agent Framework for Mitigating Gender Bias in Translation Quality Estimation (2026.acl-long)

Copied to clipboard

Jinhee Jang, Juhwan Choi, Dongjin Lee, Seunguk Yu, YoungBin Kim

AdaDPI: Document-level Translation Adaptive Agent via Dynamic Parametric Internalization (2026.acl-long)

Challenge:	Existing QE models exhibit systematic gender bias, especially in gender-ambiguous contexts.
Approach:	They propose a multi-agent-based, fairness-aware QE framework that mitigates gender bias in both gender-ambiguous and gender-explicit scenarios.
Outcome:	The proposed framework mitigates gender bias in gender-ambiguous and gender-explicit scenarios while maintaining the strengths of existing models.

Copied to clipboard

Hong Ren, Liting Deng, Shaolin Zhu, Deyi Xiong

Mitigating Over-Refusal in Aligned Large Language Models via Inference-Time Activation Energy (2026.acl-long)

Challenge:	Existing solutions, such as memory-based agents, rely on explicit context concatenation, which leads to context dilution, high inference latency, and superficial knowledge integration.
Approach:	They propose an adaptive agentic framework that shifts the DocMT paradigm from static retrieval to dynamic parametric internalization.
Outcome:	Extensive experiments on the discourse-rich GuoFeng and IWSLT2017 datasets show that AdaDPI outperforms the SoTA baselines by more than 5 points on the consistency metric.

Copied to clipboard

Eric Hanchen Jiang, Weixuan Ou, Run Liu, Shengyuan Pang, Guancheng Wan, Ranjie Duan, Wei Dong, Kai-Wei Chang, XiaoFeng Wang, Ying Nian Wu, Xinfeng Li

Refining and Reusing Annotation Guidelines for LLM Annotation (2026.acl-long)

Challenge:	Existing safety alignment techniques prioritize mitigating harmful responses at the expense of overcautious behavior, leading models to incorrectly refuse benign requests.
Approach:	They propose a fine-tuning free framework to improve safety and reduce false refusals by dynamic, inference-time intervention.
Outcome:	The proposed framework raises compliance on the ORB-H benchmark from 57.3% to 82.6% while maintaining the baseline safety performance.

Copied to clipboard

Kon Woo Kim, Jin-Dong Kim, Akiko Aizawa

Challenge:	Large Language Models (LLMs) demonstrates remarkable zero-shot annotation tasks . but, they struggle with the specialized conventions of gold-standard benchmarks .
Approach:	They propose to reuse and refine annotation guidelines as an alignment mechanism . they propose to use iterative moderation framework to simulate early phases of annotation projects .
Outcome:	The proposed framework shows a good potential in effectively refining guidelines, but there is room for improvement.

Cross-Examination Framework: A Task-Agnostic Diagnostic for Information Fidelity in Text-to-Text Generation (2026.acl-long)

Copied to clipboard

Tathagata Raha, Clement Christophe, Nada Saadi, Hamza A Javed, Marco AF Pimentel, Ronnie Rajan, Praveenkumar Kanithi

AlignCultura: Towards Culturally Aligned Large Language Models? (2026.acl-long)

Challenge:	Traditional metrics like BLEU and BERTScore fail to capture semantic fidelity in generative text-to-text tasks.
Approach:	They propose a cross-examination framework that generates verifiable questions from each text and performs a Cross-exam to derive three interpretable scores: Coverage, Conformity, and Consistency.
Outcome:	The proposed framework detects critical errors across translation, summarization and clinical note-generation and human expert validation shows it is reliable without gold references.

Copied to clipboard

Gautam Siddharth Kashyap, Mark Dras, Usman Naseem

UPDESH: Synthesizing Grounded Instruction Tuning Data for 13 Indic Languages (2026.acl-long)

Challenge:	Existing benchmarks represent early steps toward cultural alignment, yet no benchmarks currently enables systematic evaluation of cultural alignment in line with UNESCO’s principles of cultural diversity w.r.t HHH paradigm.
Approach:	Align-Cultura aims to evaluate cultural alignment in large language models . it uses a Query Construction pipeline to reclassify prompts and expand underrepresented domains . response generation pairs prompts with culturally grounded responses .
Outcome:	Empirically, culturally fine-tuned models improve joint HHH by 4%–6%, reduce cultural failures by 18%, achieve 10%–12% efficiency gains, and limit leakage to 0.3%.

Copied to clipboard

Pranjal A Chitale, Varun Gumma, Sanchit Ahuja, Prashant Kodali, Manan Uppadhyay, Deepthi Sudharsan, Sunayana Sitaram

Dynamic Generation of Multi LLM Agents Communication Topologies with Graph Diffusion Models (2026.acl-long)

Challenge:	Developing culturally grounded multilingual AI systems is challenging for low-resource languages . synthetic data is underexplored, but its effectiveness in multilingual and multicultural contexts is understudied .
Approach:	They propose a top-up synthetic data generation framework grounded in Wikipedia content . they use 9.5M data points across 13 Indian languages and English to generate a high-quality dataset .
Outcome:	The proposed model improves on NLG tasks and narrows performance gaps with high-resource languages.

Copied to clipboard

Eric Hanchen Jiang, Levina Li, Frank Wan, Xiao Liang, Sophia Yin, Yuchen Wu, Xinfeng Li, Yizhou Sun, Wei Wang, Kai-Wei Chang, Ying Nian Wu

TaxPraBen: A Scalable Benchmark for Structured Evaluation of LLMs in Chinese Real-World Tax Practice (2026.acl-long)

Challenge:	Existing frameworks rely on static or rule-based topologies that fail to adapt to task requirements.
Approach:	They propose a generative framework that generates highly task-adaptive topologies . they validated the framework on multiple benchmarks and validated it on multiple platforms .
Outcome:	The proposed framework outperforms existing frameworks in task-adaptive communication topologies.

Copied to clipboard

Gang Hu, Yating Chen, Haiyan Ding, Wang Gao, Huang Jiajia, Min Peng, Qianqian Xie, Kun Yue

Challenge:	Large Language Models excel in general domains but lack real-world practical capabilities.
Approach:	They propose a benchmark for Chinese taxation practice that combines 10 traditional application tasks with 3 pioneering real-world scenarios.
Outcome:	The proposed benchmark combines 10 traditional tasks with 3 pioneering real-world scenarios.

Think How to Think: Mitigating Overthinking with Autonomous Difficulty Cognition in Large Reasoning Models (2026.acl-long)

Copied to clipboard

Yongjiang Liu, Haoxi Li, Xiaosong Ma, Jie Zhang, Song Guo

Theory-optimal Quantization Based on Flatness (2026.acl-long)

Challenge:	Recent Large Reasoning Models (LRMs) excel at complex reasoning tasks but often suffer from overthinking.
Approach:	They propose a two-stage fine-tuning strategy that progressively inspires LRMs’ difficulty cognition and redundancy cognition of LRM.
Outcome:	The proposed model significantly reduces inference costs by over 70% on easy tasks and 40% on complex ones without compromising performance.

Copied to clipboard

Xiusheng Huang, Zhe Li, Xuanwu Yin, Lu Wang, Yequan Wang, Dong Li, Emad Barsoum, Kang Liu

When Background Matters: Breaking Medical Vision Language Models by Transferable Attack (2026.acl-long)

Challenge:	Recent approaches to quantization of Large Language Models (LLMs) have been widely adopted due to activation outliers, which degrade model performance especially at lower bit precision.
Approach:	They propose a new metric for quantization that strategically distributes outlier magnitudes across matrix dimensions via optimized diagonal operations.
Outcome:	The proposed framework achieves less than 1% accuracy drop in W4A4 quantization on the LLaMA-3-8B model and reduces the performance gap by 39.1% on the more challenging W2A4KV16 model.

Copied to clipboard

Akash Ghosh, Subhadip Baidya, Sriparna Saha, Xiuying Chen

Tears or Cheers? Benchmarking LLMs via Culturally Elicited Distinct Affective Responses (2026.acl-long)

Challenge:	Existing medical attacks focus on secondary objectives such as model stealing or adversarial fine-tuning, while transferable attacks from natural images introduce visible distortions that clinicians can easily detect. Existing transferable adversarials are less effective in the medical domain.
Approach:	They propose a highly transferable black-box multimodal attack that induces incorrect yet clinically plausible diagnoses while keeping perturbations imperceptible.
Outcome:	The proposed method induces incorrect yet clinically plausible diagnoses while keeping perturbations imperceptible.

Copied to clipboard

Chongyuan Dai, Yaling Shen, Zihan Gao, Jia Li, Yishun Jiang, Yaxiong Wang, Liu Liu, Zongyuan Ge, Jinpeng Hu

Emergent Misalignment via In-Context Learning: Narrow in-context examples can produce broadly misaligned LLMs (2026.acl-long)

Challenge:	Culture is a fundamental determinant of human affective processing and affective perceptions are often limited by declarative knowledge or established societal customs.
Approach:	They propose a multimodal benchmark that leverages LLM-generated provisional labels to isolate cross-cultural emotional distinctions.
Outcome:	The proposed benchmark captures cross-cultural emotional distinctions and derives reliable ground-truth annotations through human evaluation.

Copied to clipboard

Nikita Afonin, Nikita Andriianov, Vahagn Hovhannisyan, Nikhil Bageshpura, Kyle Liu, Kevin Zhu, Sunishchal Dev, Ashwinee Panda, Oleg Rogov, Elena Tutubalina, Alexander Panchenko, Mikhail Seleznyov

EvoRoute: Experience-Driven Self-Routing LLM Agent Systems (2026.acl-long)

Challenge:	Recent studies have documented emergent misalignment in language models adapted on narrow examples . Emergent misalignement occurs when models are trained on narrow set of misallocated examples resulting in harmful or misleading responses .
Approach:	They propose to explain in-context EM as conflict between safety objectives and context-following behavior.
Outcome:	The proposed model is adapted on 16 in-context examples and produces misaligned responses to benign queries.

Copied to clipboard

Guibin Zhang, Haiyang Yu, Kaiming Yang, Bingli Wu, Fei Huang, Yongbin Li, Shuicheng Yan

Foresight Optimization for Strategic Reasoning in Large Language Models (2026.acl-long)

Challenge:	EvoRoute is a self-evolving model routing paradigm that transcends static, pre-defined model assignments.
Approach:	They propose a model routing paradigm that transcends static, pre-defined model assignments.
Outcome:	Experiments on GAIA and BrowseComp+ show that EvoRoute reduces execution cost and latency by over 70%.

Copied to clipboard

Jessie Wang, Jiawen Duan, Jian Wang, Kaitao Song, Chunpu Xu, Johnny K. W. Ho, YU Fenggang, Johan F. Hoorn, Wenjie Li

DetectRL-X: Towards Reliable Multilingual and Real-World LLM-Generated Text Detection (2026.acl-long)

Challenge:	Existing reasoning enhancement methods do not capture foresight in LLMs.
Approach:	They propose to integrate opponent modeling principles into policy optimization to enhance strategic reasoning in LLMs by integrating opponent modeling into policy.
Outcome:	The proposed method outperforms existing reasoning-based LLMs in out-of-domain scenarios and shows that it significantly enhances strategic reasoning across LLM of varying sizes and origins.

Copied to clipboard

Junchao Wu, Yefeng Liu, Chenyu Zhu, Hao Zhang, Zeyu Wu, Tianqi Shi, Yichao Du, Longyue Wang, Weihua Luo, Jinsong Su, Derek F. Wong

ImmersiveTTS: Environment-Aware Text-to-Speech with Multimodal Diffusion Transformer and Domain-Specific Representation Alignment (2026.acl-long)

Challenge:	Existing detectors are limited in their ability to detect large language models generated content in multilingual environments.
Approach:	They propose a multilingual benchmark to evaluate advanced detectors across 8 dimensions to better align with real-world applications.
Outcome:	The proposed benchmark encompasses 8 languages commonly used in commercial contexts and collects human-written texts from 6 domains highly susceptible to LLM misuse.

Copied to clipboard

Jun-Hak Yun, Seung-Bin Kim, Seong-Whan Lee

When Efficiency Becomes a Vulnerability: Computational Cost Attacks on WebAgents (2026.acl-long)

Challenge:	ImmersiveTTS model synthesizes intelligible speech and environmental audio from natural language descriptions.
Approach:	They propose an environment-aware text-to-speech model that integrates natural speech with environmental audio . the model explicitly models cross-modal interactions through a dual-stream stage .
Outcome:	Experimental results show that ImmersiveTTS achieves higher naturalness, intelligibility, and audio fidelity than existing approaches.

Copied to clipboard

Liang-Bo Ning, Yuchen Zhu, Heqing Huang, Xin Wang, Yi Chang, Li Qing, Wenqi Fan

Evolving Beyond Snapshots: Harmonizing Structure and Sequence via Entity State Tuning for Temporal Knowledge Graph Forecasting (2026.acl-long)

Challenge:	Existing WebAgents suffer from computational cost attacks due to long reasoning processes and excessive computational cost.
Approach:	They propose a framework that generates adversarial prompts and a reinforcement learning-enhanced selector to identify the most effective perturbations.
Outcome:	The proposed framework exploits large language models to generate diverse adversarial prompts and a reinforcement learning–enhanced selector to identify the most effective perturbations.

Copied to clipboard

Siyuan Li, Yunjia Wu, Yiyong Xiao, Pingyang Huang, Peize Li, Ruitong Liu, Yan Wen, Te Sun

CodeRipple: Wavelet-Based Detection of LLM-Generated Code (2026.acl-long)

Challenge:	Temporal knowledge graphs (TKGs) require predicting future facts by modeling structural dependencies within each snapshot and temporal evolution across snapshots.
Approach:	They propose an encoder-agnostic framework that provides persistent entity states . EST maintains a global state buffer and aligns structural evidence with sequential signals .
Outcome:	Experiments show that EST improves diverse backbones and achieves state-of-the-art performance.

Copied to clipboard

Xingyu Yao, Zhendong Mao, Quan Wang

PHOTON: Hierarchical Autoregressive Modeling for Lightspeed and Memory-Efficient Language Generation (2026.acl-long)

Challenge:	Existing training-free detectors rely on global statistics of the Token Perplexity Sequence (TPS) and struggle with code.
Approach:	They propose a training-free detection framework that characterizes TPS morphology across scales.
Outcome:	The proposed framework outperforms existing training-free detectors on three challenging benchmarks spanning programming languages, multiple generating LLMs, and various evasion strategies.

Copied to clipboard

Yuma Ichikawa, Naoya Takagi, Takumi Nakagawa, Yuzi Kanazawa, Akira Sakai

CoMeT: Collaborative Memory Transformer for Efficient Long Context Modeling (2026.acl-long)

Challenge:	Experimental results show that PHOTON is superior to competitive Transformer-based language models . despite their capabilities, the inference cost increases with context length under recent serving workloads.
Approach:	They propose a hierarchical autoregressive model that replaces horizontal scanning with vertical context scanning.
Outcome:	Experimental results show that PHOTON is superior to Transformer-based language models . it reduces decode-time KV-cache traffic yielding up to 103 higher throughput per unit memory .

Copied to clipboard

Runsong Zhao, Shilei Liu, Jiwei Tang, Langming Liu, Haibin Chen, Weidong Zhang, Yujin Yuan, Tong Xiao, JingBo Zhu, Wenbo Su, Bo Zheng

CheckRLM: Effective Knowledge–Thought Coherence Checking in Retrieval-Augmented Reasoning (2026.acl-long)

Challenge:	a novel architecture that enables LLMs to handle arbitrarily long sequences with constant memory usage and linear time complexity is a major barrier to long-context processing.
Approach:	They propose a novel architecture that enables LLMs to handle arbitrarily long sequences with constant memory usage and linear time complexity.
Outcome:	The proposed architecture can handle arbitrarily long sequences with constant memory usage and linear time complexity.

Copied to clipboard

Dingling Xu, Ruobing Wang, Qingfei Zhao, Yukun Yan, Zhichun Wang, Daren Zha, Shi Yu, Zhenghao Liu, Shuo Wang, Xu Han, Maosong Sun

CAMEC: Complexity-Aware Multi-Expert Collaboration for Reliable Chinese Medical Question Answering (2026.acl-long)

Challenge:	Reasoning Language Models (RLMs) have improved performance on complex tasks by extending the reasoning chain, but they are prone to factual errors, especially in knowledge-intensive tasks.
Approach:	They propose a framework that improves the reliability of the reasoning process by timely checking and correcting factual errors.
Outcome:	The proposed framework outperforms baselines and shows that it mitigates error accumulation with lower costs.

Copied to clipboard

Yukang Wu, Xiyuan Jia, Jiayi Wu, HongChen Yu, Yuhan Qiu, Wu Guohua

Question Answering

Challenge:	Large language models are promising for medical question answering in china, but remain unreliable due to hallucinations, weak factual grounding and difficulty handling clinically complex cases.
Approach:	They propose a framework that combines hierarchical medical adaptation with complexity-aware expert routing for reliable Chinese medical QA.
Outcome:	The proposed framework outperforms strong general and medical LLM baselines on four Chinese medical benchmarks.

Vocabulary Hijacking in LVLMs: Unveiling Critical Attention Heads by Excluding Inert Tokens to Mitigate Hallucination (2026.acl-long)

Copied to clipboard

Yangneng Chen, Junlin Li, Weijun Yao, Xilai Ma, Guodong DU, Wenya Wang, Jing Li

HumanLLM: Benchmarking and Improving LLM Anthropomorphism via Human Cognitive Patterns (2026.acl-long)

Challenge:	Large Vision-Language Models (LVLMs) are capable of processing visual inputs, but are susceptible to hallucinations.
Approach:	They propose a method to localize and localize specific visual tokens, which are defined as Inert Tokens, across layers, revealing a rigid semantic collapse.
Outcome:	The proposed approach reduces the likelihood of LVLMs being hijacked by visual inputs while maintaining general capabilities.

Copied to clipboard

Xintao Wang, Jian Yang, Weiyuan Li, Rui Xie, Jen-tse Huang, Jun Gao, Shuai Huang, Yueping Kang, Yuanli Guo, Hongwei Feng, Yanghua Xiao

Vision-Language Introspection: Mitigating Overconfident Hallucinations in MLLMs via Interpretable Bi-Causal Steering (2026.acl-long)

Challenge:	Large Language Models (LLMs) have demonstrated remarkable capabilities in reasoning and generation, serving as the foundation for advanced persona simulation and Role-Playing Language Agents (RPLAs).
Approach:	They propose a framework that treats psychological patterns as interacting causal forces and synthesizes 113 scenarios where 2-5 patterns reinforce, conflict, or modulate each other.
Outcome:	The proposed framework outperforms Qwen3-32B on multi-pattern dynamics despite 4 fewer parameters.

Copied to clipboard

Shuliang Liu, Songbo Yang, Dong Fang, Sihang Jia, Yuqi Tang, Lingfeng Su, Ruoshui Peng, Yibo Yan, Xin Zou, Xuming Hu

v-HUB: A Benchmark for Video Humor Understanding from Vision and Sound (2026.acl-long)

Challenge:	Existing approaches to overcome object hallucination are limited . Existing mitigations include costly retraining and a training-free inference framework .
Approach:	They propose a training-free inference framework that simulates a metacognitive self-correction process.
Outcome:	The proposed framework reduces object hallucination rates by 12.67% on MMHal-Bench and improves accuracy by 5.8% on POPE.

Copied to clipboard

Zhengpeng Shi, Yanpeng Zhao, Jianqun Zhou, Yuxuan Wang, Qinrong Cui, Wei Bi, Song-Chun Zhu, Bo Zhao, Zilong Zheng

Beyond Single Representations: Multi-Model Embedding Fusion for Stable Text Classification (2026.acl-long)

Challenge:	Humor enriches our daily lives and appears in many forms, from jokes and cartoons to comedies and viral videos.
Approach:	They introduce a video humor understanding benchmark to test their ability to understand humor from visual cues.
Outcome:	The proposed video humor understanding benchmark is based on a collection of short videos . it features rich annotations and a study of environmental sound that can enhance humor .

Copied to clipboard

Jiho Gwak, Yuchul Jung

Confidence Should Be Calibrated More Than One Turn Deep (2026.acl-long)

Challenge:	Existing studies on embedding fusion have not evaluated the effectiveness of individual layers or the impact of combining embeddables from multiple models.
Approach:	They propose to combine embeddings from multiple models to improve performance across NLP tasks.
Outcome:	The proposed method improves performance on low-resource datasets and reduces the impact of any single model as the number of integrated models increases.

Copied to clipboard

Zhaohan Zhang, Chengzhengxu Li, Xiaoming Liu, Chao Shen, Ziquan Liu, Ioannis Patras

NaviMaster: Learning a Unified Policy for GUI and Embodied Navigation Tasks (2026.acl-long)

Challenge:	Existing work on confidence estimation and calibration focuses on single-turn settings . existing work on multi-turn calibration ignores the risks and potential of multi-turned conversations .
Approach:	They propose a multi-turn calibration task that reframes calibration from a static property into a dynamic challenge central to reliable multi- turn conversations.
Outcome:	The proposed model minimizes ECE@T and leverages ConfChat to improve confidence . the proposed model preserves and even enhances model performance in multi-turn interactions.

Copied to clipboard

Zhihao Luo, Wentao Yan, Jingyu Gong, Min Wang, Zhizhong Zhang, Xuhong Wang, Yuan Xie, Xin Tan

Challenge:	Recent advances in Graphical User Interface (GUI) and embodied navigation have driven progress, yet these domains have largely evolved in isolation, with disparate datasets and training paradigms.
Approach:	They propose a visual-target trajectory collection pipeline that generates trajectories for GUI and embodied tasks using a single formulation.
Outcome:	The proposed agent outperforms state-of-the-art agents in GUI navigation, spatial affordance prediction, and embodied navigation.

Teaching LLMs Human-Like Editing of Inappropriate Argumentation via Reinforcement Learning (2026.acl-long)

Copied to clipboard

Timon Ziegenbein, Maja Stahl, Henning Wachsmuth

Challenge:	Comparing human-like edits to LLMs, we observe a mismatch in editing strategies.
Approach:	They propose a reinforcement learning approach that teaches LLMs human-like editing to improve the appropriateness of arguments.
Outcome:	The proposed approach outperforms baselines and the state of the art in human-like editing, with multi-round editing achieving appropriateness close to full rewriting.

FLAIR: Steering LLM Mathematical Problem Solving based on A Fuzzy-Logic-AssIsted Reasoner (2026.acl-long)

Copied to clipboard

Hao Wu, Hongru Sun, Wanqing Li, Xinguo Yu, Hao Ming, Xiao Luo, Wenbin Zhang, Jiahong Zhao, Yi Guo, Jie Yang

QuDAR: Query-Wise Dual-Perspective Adaptive Retrieval (2026.acl-long)

Challenge:	Existing approaches to mathematical reasoning rely on static heuristics or pre-determined reasoning strategies.
Approach:	They propose an adaptive framework that integrates fuzzy theory into LLM-based mathematical reasoning.
Outcome:	The proposed framework outperforms state-of-the-art models while offering effective and interpretable diagnostics of intermediate problem-solving states.

Copied to clipboard

Joeun Kim, Seunghyouk Yoon, Xuan-Bach Le, Youngeun Nam, Doyoung Kim, Hwanjun Song, Jae-Gil Lee

Trait Activation in Silicon: A Situation-Aware Framework for Psychologically Grounded Role-Playing (2026.acl-long)

Challenge:	Existing approaches to grounding large language models rely on static weights and a static retrieval component.
Approach:	They propose a dual-perspective adaptive retrieval framework that adapts along two perspectives: retriever type (sparse vs. dense) and query format (original v. expanded).
Outcome:	The proposed framework adapts along two perspectives: retriever type (sparse vs. dense) and query format (original v. expanded).

Copied to clipboard

Zuolong Li, Pingyu Wu, Xianwen Huang, Tianyi Wei, Wenbo Zhou

SPARKLE: A Structured and Plug-and-play Agentic Retrieval Policy for Adaptive RAG Models (2026.acl-long)

Challenge:	Role-playing agents lack a deep understanding of complex human psychological mechanisms.
Approach:	They propose a situation-aware framework that decouples personality traits into bidirectional LoRA adapters.
Outcome:	Empirical results show that PD-LLM achieves superior performance in both static fidelity and dynamic adaptability.

Copied to clipboard

Jinyuan Fang, Zaiqiao Meng, Craig Macdonald

G-IdiomAlign: A Gloss-Pivoted Benchmark for Cross-Lingual Idiom Alignment (2026.acl-long)

Challenge:	Existing methods for integrating external knowledge rely on frozen large language models without explicit supervision or require costly LLM finetuning.
Approach:	They propose a structured and plug-and-play agentic retrieval policy with an additional proxy model to control the retrieval process.
Outcome:	Experiments on three in-domain and four out-of-domain QA benchmarks show that SPARKLE outperforms state-of the-art adaptive RAG models, achieving average improvements of 9.17% and 2.85%, respectively.

Copied to clipboard

Fengying Ye, Yanming Sun, Runzhe Zhan, Lidia S. Chao, Zheqi Zhang, Derek F. Wong

SciCoQA: Quality Assurance for Scientific Paper–Code Alignment (2026.acl-long)

Challenge:	Existing tools for cross-lingual idiom-to-idiom equivalence evaluation are limited . figurative meanings are non-compositional and culturally grounded, making literal mappings unreliable.
Approach:	They propose a gloss-pivoted benchmark where each idiom is anchored by an English gloss from Wiktionary.
Outcome:	The proposed benchmark is based on a dictionary-anchored English idiom . a bias to literal translation is a dominant failure mode across diverse LLMs, the study shows .

Copied to clipboard

Tim Baumgärtner, Iryna Gurevych

ImpRIF: Stronger Implicit Reasoning Leads to Better Complex Instruction Following (2026.acl-long)

Challenge:	Discrepancies between scientific papers and their code undermine reproducibility, a concern that grows as automated research agents scale scientific output beyond human review capacity.
Approach:	They propose a synthetic generation pipeline to scale beyond AI to Physics, Quantitative Biology, and other computational sciences.
Outcome:	The proposed pipeline scales beyond AI to Physics, Quantitative Biology, and other computational sciences.

Copied to clipboard

Yuancheng Yang, Lin Yang, Xu Wang, Chao Tong, Haihua Yang

HalluAudio: A Comprehensive Benchmark for Hallucination Detection in Large Audio-Language Models (2026.acl-long)

Challenge:	Experiments show that enhancing implicit reasoning capabilities can significantly improve complex instruction following in large language models.
Approach:	They propose a method to enhance LLMs’ understanding of implicit reasoning instructions by formalizing such instructions as verifiable reasoning graphs and fine-tuning with graph reasoning.
Outcome:	The proposed method outperforms existing models on five complex instruction following benchmarks and will be open-sourced in the near future.

Copied to clipboard

Feiyu Zhao, Yiming Chen, Wenhuan Lu, Daipeng Zhang, Xianghu Yue, Jianguo Wei

ReTraceQA: Evaluating Reasoning Traces of Small Language Models in Commonsense Question Answering (2026.acl-long)

Challenge:	Existing studies on hallucination focus on text or vision, while few audio-oriented studies are limited in scale, modality coverage, and diagnostic depth.
Approach:	They propose a large-scale benchmark for evaluating hallucinations across speech, sound, and music.
Outcome:	The proposed model improves hallucination rate, yes/no bias, error-type analysis, and refusal rate.

Copied to clipboard

Francesco Maria Molfese, Luca Moroni, Ciro Porcaro, Simone Conia, Roberto Navigli

BEFT: Bias-Efficient Fine-Tuning of Language Models in Low-Data Regimes (2026.acl-long)

Challenge:	Recent work in language modeling has led to effective SLMs with impressive performance levels across various benchmarks.
Approach:	They propose a benchmark that introduces process-level evaluation for commonsense reasoning tasks.
Outcome:	The proposed benchmarks show that large language models provide correct answers despite flawed reasoning processes in a substantial portion of cases.

Copied to clipboard

Baichuan Huang, Ananth Balashankar, Amir Aminifar

JoPR: Joint Emotion Perception and Reasoning for Conversational Emotion Recognition (2026.acl-long)

Challenge:	Fine-tuning bias terms of large language models (LLMs) for downstream tasks has gained a lot of attention over the past few years.
Approach:	They extensively evaluate bq, bk, v across a wide range of LLMs . they find that bv generally leads to higher downstream performance in low-data regimes compared to bQ and bK .
Outcome:	The proposed method improves performance across a wide range of LLMs spanning encoder-only and decoder-free architectures up to 6.7B parameters.

Copied to clipboard

Yumeng Fu, Weitao Huang, Junjie Wu, Hao Teng, Meishan Zhang, Bingquan Liu

Uncertainty-Aware Routing for Principled Alignment with MoE Dynamics (2026.acl-long)

Challenge:	Existing methods for ERC lack human-like emotion reasoning and discrimination between similar emotions.
Approach:	They propose a multi-dimension curriculum with long CoT fine-tuning to clone human-like emotion reasoning for conversational emotion recognition.
Outcome:	The proposed model outperforms existing methods on three widely used datasets and shows that it is more intuitive and more accurate.

Copied to clipboard

Yilong Chen, Junyuan Shang, Yuchen Feng, Zhenyu Zhang, Naibin Gu, Ziqi Wang, Tingwen Liu, Shuohuan Wang, Yu Sun, Hua Wu, Haifeng Wang

Challenge:	Mixture-of-Experts (MoE) is a cornerstone for scaling LLMs, yet its training dynamics remain poorly understood, often leading to sub-optimal specialization.
Approach:	They propose to use Helmholtz Free Energy and Router Entropy to study the MoE lifecycle and identify a universal Three-Stage Phase Transition .
Outcome:	The proposed model reduces perplexity and improves expert distinctiveness, offering a principled path toward thermodynamically aligned computation.

S^4: Operationalizing Speech Act Theory for Strategic Semi-Structured Psychiatric Interview (2026.acl-long)

Copied to clipboard

Guanqun Bi, Zhoufu Liu, Zhuang Chen, Dazhen Wan, Xiyao Xiao, Minlie Huang

More Aligned, Less Diverse? Analyzing the Grammar and Lexicon of Two Generations of LLMs (2026.acl-long)

Challenge:	Existing methods for psychiatric interviewing degenerate into rigid interrogation or aimless chitchat due to a lack of strategic planning.
Approach:	They propose a framework for psychiatric interviewing grounded in Speech Act Theory that integrates a large-scale dataset with fine-grained psychic speech act annotations.
Outcome:	The proposed framework outperforms baselines in psychiatric interviewing.

Copied to clipboard

Adrián Gude, Roi Santos-Rios, Francis Bond, Dan Flickinger, Carlos Gómez-Rodríguez, Olga Zamaraeva

MemCoRL: Alternating Co-Optimization of Memory Retrieval and Utilization via Collaborative Reinforcement Learning (2026.acl-long)

Challenge:	a growing number of studies compare LLMs with human-authored text . diversity is unclear, but it is important to understand what makes human and machine writing distinct .
Approach:	They compare syntactic properties of AI-generated and human-authored English news texts . they use the Head-Driven Phrase Structure Grammar and the English Resource Grammar .
Outcome:	The proposed model differs from human-authored English news text in two years.

Copied to clipboard

Yuewen Liu, Peng Xu, Muxi Diao, Anyi Zhang, Yang Li, Yutong Zhang

Fast and Accurate Fisher-Guided Quantization via Efficient Kronecker Factorization (2026.acl-long)

Challenge:	Existing research has proposed external memory modules for Large Language Models (LLMs) to overcome the limitations of finite input length and obtain contextual memory beyond the current input.
Approach:	They propose a two-stage alternating co-optimization reinforcement learning method that optimizes evidence retrieval and utilization using semantic feedback and rewards.
Outcome:	The proposed method outperforms baselines on lexical overlap and semantic similarity metrics, confirming the co-optimization in memory retrieval and memory utilization.

Copied to clipboard

Viktoriia A. Chekalina, Gerasin Timofey, Andrey Kuznetsov, Evgeny Frolov

Discovering a Shared Logical Subspace: Steering LLM Logical Reasoning via Alignment of Natural-Language and Symbolic Views (2026.acl-long)

Challenge:	Quantization has shown strong results in preserving model quality under compression, but under aggressive bit-width reductions, even quantization may require additional information to prevent performance degradation.
Approach:	They propose a Kronecker-factored approximation that captures second-order curvature information, captured by the Hessian, to achieve a 10 speedup over prior approaches.
Outcome:	The proposed method significantly accelerates the most expensive component in second-order quantization – Hessian parameterization . it achieves up to a 10 speedup over prior approaches.

Copied to clipboard

Feihao Fang, My T. Thai, Yuanyuan Lei

One Refiner to Unlock Them All: Inference-Time Reasoning Elicitation via Reinforcement Query Refinement (2026.acl-long)

Challenge:	Existing approaches to multistep logical reasoning are limited by natural language refinement or external symbolic solvers.
Approach:	They propose a logical subspace that captures logical reasoning capabilities in LLMs that are shared across views while remaining independent of surface forms.
Outcome:	The proposed approach improves accuracy by 11 percentage points and generalizes well on out-of-domain problems.

Copied to clipboard

Yixiao Zhou, Dongzhou Cheng, Zhiliang wu, Yi Yang, Yu Cheng, Hehe Fan

Why Does Reinforcement Learning Generalize? A Feature-Level Mechanistic Study of Post-Training in Large Language Models (2026.acl-long)

Challenge:	Existing alignment methods for Large Language Models (LLMs) are expensive and lack the flexibility to fully activate their latent reasoning capabilities.
Approach:	They propose a modular framework that treats reasoning elicitation as an inference-time alignment task.
Outcome:	The proposed framework outperforms baselines by 2.1% on average across diverse architectures and benchmarks.

Copied to clipboard

Dan Shi, Zhuowen Han, Simon Ostermann, Renren Jin, Josef Van Genabith, Deyi Xiong

Polymorphic Universal Transformer (2026.acl-long)

Challenge:	Reinforcement learning (RL)-based post-training often improves the reasoning performance of large language models beyond the training domain, while supervised fine-tuning (SFT) frequently leads to general capabilities forgetting.
Approach:	They propose a feature-level mechanistic analysis methodology to probe RL generalization using a controlled experimental setup.
Outcome:	The proposed method identifies a compact, task-agnostic set of features that directly mediate generalization across diverse tasks.

Copied to clipboard

Yilong Chen, Zitian Gao, Yihao Xiao, Jason Klein Liu, Xinyu Yang, Yifan Luo, Haoming Luo, Zhengmao Ye, Tingwen Liu, Ran Tao, Bryan Dai

Can Reasoning Path still be Effective as Input? Bridging Post-Reasoning to Chain-of-Thought Compression (2026.acl-long)

Challenge:	Compute Distribution Skew is a pathological phenomenon in ultra-deep recurrent models . it causes over-smoothing, representation rank collapse, and degraded reasoning performance.
Approach:	They propose a dynamic architecture that redefines recursive computation by decoupling parameter count from depth.
Outcome:	The proposed model significantly improves representation rank and reasoning robustness while reducing computation by 64.7%.

Copied to clipboard

Chengzhengxu Li, Xiaoming Liu, Zhaohan Zhang, Shengchao Liu, Guoxin Ma, Yu Lan, Cong Wang, Chao Shen

REAL: REtrieval-reAsoning and Logic-constructed Attention Behaviors for Long-Context KV Cache Compression (2026.acl-long)

Challenge:	Existing work on reducing CoT generation in reasoning impairs the necessary information for deriving the correct answer.
Approach:	They propose a reasoning paradigm that takes CoT as a part of context to simplify the reasoning task for Large Language Models (LLMs).
Outcome:	The proposed framework reduces the generation length of LLMs, but its effectiveness hinges on the efficiency and reliability of the contextual CoT generation.

Copied to clipboard

Mengjie Li, Yuan Feng, Xike Xie, William J. Song

From Where Words Come: Efficient Regularization of Code Tokenizers Through Source Attribution (2026.acl-long)

Challenge:	Existing methods to evict keyvalue caches ignore diverse behavior in failure cases, such as bias and distraction.
Approach:	They propose a method to analyze attention head behaviors in success and failure scenarios by maximizing signal-to-noise ratio and minimizing noise from bias and distraction.
Outcome:	The proposed method achieves comparable accuracy to the strongest baseline, HeadKV-R2 on LongBench v2 while requiring 32x less space.

Copied to clipboard

Pavel Chizhov, Egor Bogomolov, Ivan P. Yamshchikov

EmoS: A High-Fidelity Multimodal Benchmark for Fine-grained Streaming Emotional Understanding (2026.acl-long)

Challenge:	Currently, subword tokenization is the most common approach for vocabulary building in large models.
Approach:	They propose to regularize training and minimize overfitting by using source-attributed BPE . they find that undertrained tokens are prone to producing unused, unusable tokens .
Outcome:	The proposed techniques reduce the number of under-trained tokens while maintaining the same inference procedure as with regular BPE.

Copied to clipboard

Pengze Guo, Jingxi Liang, Zhiwen Xie, Qifeng Wang, Derek F. Wong

Surprisal Minimisation over Goal-directed Alternatives Predicts Production Choice in Dialogue (2026.acl-long)

Challenge:	Existing benchmarks fail to achieve ecological validity, signal clarity, and reliable fine-grained labeling in multimodal Emotion Recognition (MER) Existing datasets lack spontaneity of real-life interactions, resulting in poor quality and inconsistent data quality.
Approach:	They propose a bilingual benchmark to resolve limitations of ecological validity and noise in existing datasets by combining strictly filtered static slices with a dynamic Streaming Monologue subset.
Outcome:	EmoS provides trusted ground truth that captures continuous emotional evolution.

Copied to clipboard

Thomas P. Utting, Mario Giulianelli, Arabella Sinclair

HowToNarrate: A General-Domain Benchmark for Synchronized Video Narration with External Knowledge (2026.acl-long)

Challenge:	a method to model utterance production is based on information-theoretic notions of cost . a technique to generate alternative sets of utterables is proposed .
Approach:	They propose a procedure to generate both types of alternative sets using language models.
Outcome:	The proposed procedure allows for speaker- and listener-oriented interpretations of different cost measures.

Copied to clipboard

Xueyan Wang, Dingyi Yang, Qin Jin

Putting HUMANS first: Efficient LAM Evaluation with Human Preference Alignment (2026.acl-long)

Challenge:	Existing MLLMs overemphasize knowledge retrieval while neglecting prior context, causing redundancy and incoherence.
Approach:	They propose a framework that combines context compression, knowledge retrieval, and narration generation to improve models' performance.
Outcome:	The proposed method significantly improves MLLM performance over existing models.

Copied to clipboard

Woody Haosheng Gan, William Barr Held, Diyi Yang

Towards Fast and Accurate Modeling for Cross-Lingual Label Projection (2026.acl-long)

Challenge:	Existing LAM benchmarks with thousands of examples create substantial computational barriers.
Approach:	They examine whether subsets can reliably evaluate large audio models . they find that subset of 50 examples can achieve over 0.93 Pearson correlation with full benchmark .
Outcome:	The proposed method outperforms the full benchmark and subset selection methods.

Copied to clipboard

Thang Le, Huy Huu Nguyen, Anh Tuan Luu, Thamar Solorio, Thien Huu Nguyen

False Friends or Cognates? A Cross-lingual Semantic Ambiguity Evaluation for Galician, Portuguese and Spanish (2026.acl-long)

Challenge:	Existing methods for label projection are inaccurate or slow for large-scale use.
Approach:	They propose to synthesize alignment sequence pairs and fine-tune an encoder model with span alignment objective while controlling data influence during training.
Outcome:	The proposed method outperforms state-of-the-art methods while maintaining fast inference speed across 50+ languages.

Copied to clipboard

Marta Vázquez Abuín, Jose Camacho-Collados, Marcos Garcia

BlindGuard: Safeguarding LLM-based Multi-Agent Systems under Unknown Attacks (2026.acl-long)

Challenge:	Closely related languages exhibit a high degree of lexical and orthographic similarity, which can facilitate cross-lingual understanding but also give rise to systematic semantic ambiguity.
Approach:	They introduce six cross-lingual datasets that are manually or semi-automatically generated and are able to identify and process false friends among these languages.
Outcome:	The proposed models can identify and process false friends among Galician, Portuguese, and Spanish.

Copied to clipboard

Rui Miao, Yixin Liu, Yili Wang, Xu Shen, Yue Tan, Yiwei Dai, Shirui Pan, Xin Wang

Black-Box Membership Inference Attacks for Video Training Data in Multimodal Large Language Models (2026.acl-long)

Challenge:	Existing supervised defense methods rely on labeled malicious agents to train a supervised model of malicious behavior.
Approach:	They propose an unsupervised defense method that learns without requiring any attack-specific labels or prior knowledge of malicious behaviors.
Outcome:	The proposed method detects diverse attack types across MAS with various communication patterns while maintaining superior generalizability compared to baselines.

Copied to clipboard

Jinrui Wang, Zhenfeng Gao, Wendan Wang, Huili Wang, Zichen Qin, Linjie Zhu, Hongke Fu, Shangguang Wang, Tao Qi

Mitigating Safety Context Amnesia in Multimodal Reasoning Models via Intent-Guided Safety Reasoning (2026.acl-long)

Challenge:	Existing methods assess model memorization of key semantic concepts within a video but do not provide reliable evidence that a specific video was used during training.
Approach:	They propose a black-box MIA framework that can provide reliable evidence of specific video data usage for training multimodal large language models.
Outcome:	The proposed framework can provide reliable evidence of specific video data usage for training multimodal large language models.

Copied to clipboard

Xiyao Dong, Guangsheng Cheng, YiLong Chen, Xiaojin Zhang, Kun He

On-policy Reinforcement Fine-tuning with Offline reward for Multi-step Embodied Planning (2026.acl-long)

Challenge:	Recent advances in Multimodal Large Reasoning Models (MLRMs) have enabled explicit chain-of-thought inference across vision and language, improving performance on complex cognitive tasks.
Approach:	They propose an inference-time defense that uses a percept decoupler to extract objective visual evidence into a structured intent output and a cognitive arbiter to enforce explicit safety constraints prior to generation.
Outcome:	The proposed defense improves defense success rates by over 62% compared to baselines while preserving task utility.

Copied to clipboard

Di Wu, Jiaxin Fan, Chloe Gu, Guanbo Wang, Wei Yin, Wenhao Li, Bo Jin

Please refuse to answer me! Mitigating Over-Refusal in Large Language Models via Adaptive Contrastive Decoding (2026.acl-long)

Challenge:	Embodied planning requires agents to make coherent multi-step decisions based on dynamic visual observations and verbal goals.
Approach:	They propose an On-policy Reinforcement fine-tuning framework with offline rewards for Embodied Task Planning that preserves generalization benefits of RFT while addressing costly interaction and sparse rewards.
Outcome:	The proposed framework outperforms closed-source and online-RL methods on EmbodiedBench, a recent benchmark for interactive embodied tasks.

Copied to clipboard

Yupeng Qi, Ziyu Lyu, Lixin Cui, Lu Bai, Feng Xia

Knowledge Injection Exists in MoE? Exploring Expert-Aware Contrast Decoding in MoE for Mitigating LLMs’ Hallucinations (2026.acl-long)

Challenge:	Existing methods for mitigating over-refusal can't maintain low refusal ratio for harmless queries while keeping high for malicious queries.
Approach:	They propose a model-agnostic approach to mitigate over-refusal in large language models . they propose an adaptive contrastive decoding strategy that incorporates or removes the refusal token distribution .
Outcome:	The proposed approach reduces the refusal ratio for over-refusal queries by 10.35% while increasing the refusal rate for malicious queries by 0.13%.

Copied to clipboard

Xinyue Fang, Zhiliang Tian, Zhen Huang, Ziyi Pan, Zhihua Wen, Xi Wang, Quntian Fang, Dongsheng Li

ERCThinker: Fast-Slow Thinking for Emotion Recognition in Conversation (2026.acl-long)

Challenge:	Existing methods to mitigate hallucinations include prompt engineering and model optimization, but lack domain generalization and potential errors in fine-tuning data may exacerbate the hallucism.
Approach:	They propose an expert-aware adaptive contrast decoding that uses expert differences in MoE’s higher layers to mitigate hallucinations on QA tasks.
Outcome:	The proposed method outperforms baseline models on four datasets Large language models (LLMs) show strong performance but suffer from hallucinations, limiting their application.

Copied to clipboard

Yumeng Fu, Weitao Huang, Junjie Wu, Hao Teng, Shouduo Shang, Meishan Zhang, Bingquan Liu

REaR : Retrieve, Expand and Refine for Effective Multitable Retrieval (2026.acl-long)

Challenge:	Existing methods for ERC lack interpretability and shallow semantics capture deep semantics.
Approach:	They propose a Fast-Slow thinking framework for Emotion Recognition in Conversation . they use fine-grained emotion reasoning chains to capture deep semantics .
Outcome:	The proposed framework achieves state-of-the-art in explanation and judgment on a benchmark dataset.

Copied to clipboard

Rishita Agarwal, Himanshu Singhal, Peter Baile Chen, Manan Roy Choudhury, Dan Roth, Vivek Gupta

Figure It Out: Improve the Frontier of Reasoning with Executable Visual States (2026.acl-long)

Challenge:	REaR is retriever-agnostic and improves dense/ sparse retrievers on complex table QA datasets.
Approach:	a new framework separates semantic relevance from structural joinability for efficient multi-table retrieval. adam scott and eric liu introduce REaR, a three-stage, LLM-free framework.
Outcome:	a new framework improves retrieval quality and performance on complex table QA datasets . it separates semantic relevance from structural joinability and prunes weakly related candidates . the framework is retriever-agnostic and delivers performance competitive with state-of-the-art LLM-augmented retrieval systems .

Copied to clipboard

Meiqi Chen, Fandong Meng, Jie Zhou

MARCH: Multi-Agent Reinforced Check for Hallucination (2026.acl-long)

Challenge:	Recent reasoning models fail to capture structural constraints in complex settings.
Approach:	They propose a visual-based reasoning system that integrates executable visual construction into multi-turn reasoning via end-to-end reinforcement learning.
Outcome:	The proposed model outperforms strong text-only chain-of-thought models on seven mathematical benchmarks and improves by 13.12% on AIME 2025 and 11.00% on BeyondAIME.

Copied to clipboard

Zhuo Li, Yupeng Zhang, Pengyu Cheng, Jiajun Song, Mengyu Zhou, Hao Li, Shujie Hu, Yu Qin, null Erchao.zec, Xiaoxi Jiang, null Guanjunjiang

A Survey of Deep Learning for Geometry Problem Solving (2026.acl-long)

Challenge:	Existing methods to detect hallucinations suffer from inherent confirmation bias, where the verifier inadvertently reproduces the errors of the original generation.
Approach:	They propose a framework that enforces rigorous factual alignment by leveraging deliberate information asymmetry by combining a pipeline of three specialized agents: a Solver, a Proposer, and a Checker.
Outcome:	Extensive experiments across hallucination benchmarks demonstrate that MARCH substantially reduces hallucinism rates.

Copied to clipboard

Jianzhe Ma, Wenxuan Wang, Qin Jin

ConsistRM: Improving Generative Reward Models via Consistency-Aware Self-Training (2026.acl-long)

Challenge:	Recent surge in deep learning technologies has significantly accelerated research in this area.
Approach:	They propose a comprehensive summary of the relevant tasks in geometry problem solving and a review of related deep learning methods.
Outcome:	The proposed method is based on a systematic review of related methods and evaluation metrics and methods.

Copied to clipboard

Yu Liang, Liangxin Liu, Longzheng Wang, null Wangyan, Zhang Yueyang, Long Xia, Zhiyuan Sun, Daiting Shi

Challenge:	ConsistRM is a self-training framework that enables effective and stable GRM training without human annotations.
Approach:	They propose a self-training framework that enables effective and stable GRM training without human annotations.
Outcome:	The proposed framework outperforms vanilla Reinforcement Fine-Tuning (RFT) by 1.5% on five benchmark datasets.

Beyond Experience Retrieval: Learning to Generate Utility-Optimized Structured Experience for Frozen LLMs (2026.acl-long)

Copied to clipboard

Xuancheng Li, Haitao Li, Yujia Zhou, Yiqun Liu, Qingyao Ai

Zero-shot Large Language Models for Automatic Readability Assessment (2026.acl-long)

Challenge:	Large language models (LLMs) are largely static and often redo reasoning or repeat mistakes. Prior experience reuse relies on external retrieval, which is similarity-based, can introduce noise, and adds latency.
Approach:	They propose a lightweight plug-in that stores experience in its parameters and generates a structured, instance-tailored experience entry in a single forward pass to guide a frozen LLM executor.
Outcome:	Experiments on mathematical reasoning benchmarks show consistent accuracy gains across executors with low overhead.

Copied to clipboard

Riley Grossman, Yi Chen

NoisyCausal: A Benchmark for Evaluating Causal Reasoning Under Structured Noise (2026.acl-long)

Challenge:	Existing unsupervised automatic readability assessment methods have important practical and research applications.
Approach:	They propose a new zero-shot prompting methodology for automatic readability assessment (ARA) they propose combining large language models with readability formula scores to improve robustness .
Outcome:	The proposed method outperforms prior methods on 13 of 14 datasets and improves on contextual and shallow features of readability.

Copied to clipboard

Zhi Xu, Yun Fu

Chain-of-Thought as a Lens: Evaluating Structured Reasoning Alignment between Human Preferences and Large Language Models (2026.acl-long)

Challenge:	Existing benchmarks focus on clean, abstract scenarios where causal structure is simple or implicitly assumed.
Approach:	They propose a benchmark to evaluate causal reasoning under structured noise.
Outcome:	The proposed method outperforms standard prompting and reasoning baselines on NoisyCausal.

Copied to clipboard

Boxuan Wang, Zhuoyun Li, Xinmiao Huang, Xiaowei Huang, Yi Dong

Robust Membership Inference for Large Language Models under Adversarial Generative Corruption (2026.acl-long)

Challenge:	Empirical results show misalignment at greater reasoning depths is driven mainly by alignment errors such as thematic shift and redundant reasoning.
Approach:	They propose a method to quantitatively assess the alignment between multi-step, structured reasoning in large language models and human preferences by constructing semantic-entropy-based matrices over intermediate steps and measuring their divergence.
Outcome:	The proposed method shows that it is consistent with previous studies and can be used as a diagnostic signal.

Copied to clipboard

Yuanhong Huang, Huili Wang, Xueying Bai, Jinrui Wang, Jiajun Liu, Ziqin Wang, Wanchun Ni, Shangguang Wang, Tao Qi

Towards Trustworthy Smart Contract Synthesis: A Multi-Agent Framework with Lean-Based Verification (2026.acl-long)

Challenge:	Membership inference attacks are a promising tool for auditing training data of LLMs . existing methods rely on the assumption that LLM's assign higher confidence scores to training samples than to non-training ones.
Approach:	They propose a membership inference framework that can be robust against adversarial MIAs.
Outcome:	The proposed framework can be robust against adversarial MIA methods and AIGT detectors while maintaining the performance of baselines.

Copied to clipboard

Bowei Zhang, Hanbing Liu, Qixin Tian, Siyu Chen, Ziyuan Wang, Qi Qi

LoRA on the Go: Instance-level Dynamic LoRA Selection and Merging (2026.acl-long)

Challenge:	Smart Contracts are the foundation of Decentralized Finance (DeFi), executing financial logic without trusted intermediaries.
Approach:	They propose a framework that integrates LLM-based generation with Lean-based auto-formalization and verification.
Outcome:	LeVer is the first trustworthy smart contract synthesis framework that integrates LLM-based generation with Lean-based auto-formalization and verification.

Copied to clipboard

Seungeon Lee, Soumi Das, Manish Gupta, Krishna P. Gummadi

Deep Supervised Contrastive Learning of Pitch Contours for Robust Pitch Accent Classification in Seoul Korean (2026.acl-long)

Challenge:	Low-Rank Adaptation (LoRA) is a parameter-efficient approach for fine-tuning large language models.
Approach:	They propose a low-rank Adaptation framework that automatically selects and merges LoRA adapters at the instance level without additional training.
Outcome:	The proposed framework outperforms training-based baselines on some tasks upto a margin of 3.6% while remaining competitive on other tasks and maintaining inference throughput.

Copied to clipboard

Hyunjung Joo, GyeongTaek Lee

DARM: Distribution-Aware Reward Modeling by Alleviating Biases from Low Preference-Context Dependency Data (2026.acl-long)

Challenge:	Existing approaches to classify fine-grained pitch accent patterns in Seoul Korean are limited due to variable realizations of F0 in real-world speech.
Approach:	They propose a deep contrastive learning framework to classify fine-grained pitch accent patterns in Seoul Korean using a dataset of 10,093 Accentual Phrases.
Outcome:	The proposed framework outperforms baseline models with state-of-the-art accuracy and F1-score.

Copied to clipboard

Shaofan Liu, Guoqiang Zhang, Shihan Dou, Huiyuan Zheng, Yiming Zhou, Junjie Ye, Shaowen Wang, Shichun Liu, Jiazheng Zhang, Tao Gui, Qi Zhang, Xuanjing Huang

Challenge:	Existing methods for training reward models are vulnerable to context neglect and degraded accuracy.
Approach:	They propose distribution-aware reward modeling that augments the RM objective with a conditional mutual information regularizer that maximizes context and the predicted reward conditioned on the response.
Outcome:	The proposed model improves performance in RLHF and improves accuracy in other settings.

KnowRL: Exploring Knowledgeable Reinforcement Learning for Factuality (2026.acl-long)

Copied to clipboard

Baochang Ren, Shuofei Qiao, Ningyu Zhang, Da Zheng, Huajun Chen

Rethinking Depression Prediction from a Fine-Grained Subscore Modeling Perspective via Multi-Task Learning (2026.acl-long)

Challenge:	Existing Reinforcement Learning approaches rely on outcome-oriented rewards to reinforce fabricated reasoning paths when the final answer is correct.
Approach:	They propose a framework that integrates factual supervision directly into reasoning . they propose to decompose chain of thought into atomic facts and verify them against ground-truth knowledge .
Outcome:	The proposed framework reduces the Incorrect Rate on SimpleQA by 20.3% while maintaining strong performance on complex reasoning benchmarks.

Copied to clipboard

Zhenguang Wang, Bo Li, Wenhui Tan, Peng Cao, Yang Wang, Jia Duan, Fei Wang, Osmar Zaiane

PDTrim: Targeted Pruning for Prefill-Decode Disaggregation in Inference (2026.acl-long)

Challenge:	Existing methods for depression assessment rely on standardized ratings, but they are time-consuming and subject to inter-rater variability.
Approach:	They propose a fine-grained model for subscore prediction via multi-task learning that can be used to predict depression severity using multiple tasks.
Outcome:	The proposed model outperforms baselines and Qwen3-14B direct scoring on the public E-DAIC dataset and to a large-scale private clinical dataset.

Copied to clipboard

Hao Zhang, Lyu Mengsi, Zhuo Chen, Yulong Ao, Yonghua Lin

ASTRA: An Automated Framework for Strategy Discovery, Retrieval, and Evolution for Jailbreaking LLMs (2026.acl-long)

Challenge:	Existing pruning methods ignore prefill-decode (PD) disaggregation in practice.
Approach:	They propose a pruning method that is highly integrated with prefill-decode (PD) disaggregation, enabling more precise pruning of blocks.
Outcome:	The proposed method achieves strong performance in both PD disaggregation and PD unified settings, and can be extended to other non-block pruning methods.

Copied to clipboard

Xu Liu, Yan Chen, Kan Ling, Yichi Zhu, Hengrun Zhang, Guisheng Fan, Huiqun Yu

LLM Safety From Within: Detecting Harmful Content with Internal Representations (2026.acl-long)

Challenge:	Existing methods lack the capability for continuous learning and self-evolution from interactions, limiting the diversity and adaptability of attack strategies.
Approach:	They propose an automated framework capable of discovering, retrieving, and evolving attack strategies.
Outcome:	The proposed framework outperforms existing baselines in a black-box setting.

Copied to clipboard

Difan Jiao, Yilun Liu, Ye Yuan, Zhenwei Tang, Linfeng Du, Haolun Wu, Ashton Anderson

AttnPO: Attention-Guided Process Supervision for Efficient Reasoning (2026.acl-long)

Challenge:	State-of-the-art guard models rely on terminal-layer representations and overlook safety-relevant features encoded across internal layers.
Approach:	They propose a lightweight guard model that harnesses safety neurons from LLM internals without modifying the underlying model.
Outcome:	The proposed model outperforms open-source guard models across multiple benchmarks while using 250 fewer trainable parameters.

Copied to clipboard

Shuaiyi Nie, null Dingsiyu, Wenyuan Zhang, Linhao Yu, Tianmeng Yang, Yao Chen, Weichong Yin, Yu Sun, Hua Wu, Tingwen Liu

Evaluating LLMs on Large-Scale Graph Property Estimation via Random Walks (2026.acl-long)

Challenge:	Existing trajectory-level length penalties fail to effectively shorten reasoning length and degrade accuracy, as they treat all reasoning steps uniformly and lack fine-grained signals to distinguish redundancy from necessity.
Approach:	They propose a low-overhead process-supervised RL framework that leverages the model’s intrinsic attention signals for step-level credit assignment.
Outcome:	The proposed framework reduces reasoning length while improving performance across 9 benchmarks.

Copied to clipboard

Sunil Kumar Maurya, Xin Liu

MedEinst: Benchmarking the Einstellung Effect in Medical LLMs through Counterfactual Differential Diagnosis (2026.acl-long)

Challenge:	Large Language Models (LLMs) have improved reasoning abilities but are limited due to limited context length.
Approach:	They propose a large graph benchmark dataset and propose four tasks to evaluate LLMs' reasoning abilities.
Outcome:	The proposed tasks evaluate the reasoning abilities of LLMs on a large graph benchmark dataset.

Copied to clipboard

Wenting Chen, Guolin Huang, Wenxuan Wang, Zhongrui Zhu

Libra-VLA: Achieving Learning Equilibrium via Asynchronous Coarse-to-Fine Dual-System (2026.acl-long)

Challenge:	Existing medical benchmarks fail to detect the Einstellung Effect in clinical diagnosis . Existing models exhibit the Einstellung effect, relying on statistical shortcuts rather than logical reasoning.
Approach:	They propose a counterfactual benchmark that uses statistical shortcuts to diagnose patients . they propose CGME-based system that iteratively refines reasoning paths .
Outcome:	The proposed model achieves high baseline accuracy but severe bias trap rates . iteratively refines reasoning paths in an exemplar base and consolidates disease-specific knowledge into illness graphs.

Copied to clipboard

Yifei Wei, Linqing Zhong, Yi Liu, Yuxiang Lu, Xindong He, Maoqing Yao, Guanghui Ren

Challenge:	Vision-Language-Action models ground high-level semantic instructions into executable physical actions.
Approach:	They propose a Coarse-to-Fine Dual-System VLA architecture that decouples learning complexity into a coarse-to fine hierarchy while leveraging structural modularity to implement an asynchronous execution strategy.
Outcome:	The proposed architecture decouples learning complexity into a coarse-to-fine hierarchy while leveraging structural modularity to implement an asynchronous execution strategy.

Language Statistics and False Belief Reasoning: Evidence from 41 Open-Weight LMs (2026.acl-long)

Copied to clipboard

Sean Trott, Samuel M. Taylor, Cameron Robert Jones, James A. Michaelov, Pamela D. Rivière

Decoding-Unlearning: Fact Forgetting via Entropy-Guided Inference (2026.acl-long)

Challenge:	a recent study on mental state reasoning in language models relies on a relatively small sample of closed-source LMs.
Approach:	They replicate and extend published work on false belief task by assessing LM mental state reasoning behavior across 41 open-weight models.
Outcome:	The results show that large LMs show higher sensitivity and predictive power . they also show that humans and LM models show a bias towards attributing false beliefs .

Copied to clipboard

Jingwen Pu, Mingjun Shi, Xinrui Ren, Yizhe Wang, Xinyu Zhang, Zhaokun Wang, Kun She

Global Adaptive Momentum Meets Local Personalized Perturbation: Efficient Federated LLM Fine-Tuning with Zeroth-Order Gradients (2026.acl-long)

Challenge:	Existing methods for large-scale modeling memorize sensitive information . however, they are limited in real-world scenarios and require updating parameters .
Approach:	They propose a training-free, plug-and-play inference-time unlearning strategy that uses a probe to detect queries involving forgettable concepts and applies entropy-guided decoding to suppress target knowledge.
Outcome:	Experiments on MUSE, RWKU, and WMDP datasets show that SEGUE outperforms existing methods.

Copied to clipboard

Zihan Chen, Howard Hao Yang, Tony Quek, Kai Fong Ernest Chong

Beyond Query Memorization: Large Language Model Routing with Query Decomposition and Historical Matching (2026.acl-long)

Challenge:	federated fine-tuning of large language models provides privacy-preserving approach to deploying pervasive generative AI services.
Approach:	They propose a federated framework for fine-tuning large language models . they propose unified optimization and local personalized perturbation for ZO gradients .
Outcome:	The proposed framework outperforms existing methods for integrating ZO gradients in federated learning over diverse heterogeneous data settings.

Copied to clipboard

Bo Lv, Jingbo Sun, Jianwei Lv, Chen Tang, Shaojie Zhang, Nayu Liu, Guoxin Yu, Zihao Li, Qichao Zhang, Dongbin Zhao, Ping Luo, Yue Yu

Market-Bench: Benchmarking Large Language Models on Economic and Trade Competition (2026.acl-long)

Challenge:	Existing routing methods rely on direct mapping from queries to models based on surface-level features, leading to poor generalizability on out-of-distribution data.
Approach:	They propose a new routing framework that recasts the routing task as a matching process of sifting similar queries from historical logs.
Outcome:	The proposed framework improves matching accuracy while lowering inference costs . it decouples linguistic surface forms from task-intrinsic requirements .

Copied to clipboard

Yushuo Zheng, Huiyu Duan, Zicheng Zhang, Yucheng Zhu, Xiongkuo Min, Guangtao Zhai

Benchmarking Post-Training Quantization of Large Language Models under Microscaling Floating Point Formats (2026.acl-long)

Challenge:	Existing LLM benchmarks focus on semantic complexity or quantitative competition, but rarely both simultaneously under economic scarcity.
Approach:	They propose a benchmark that evaluates the capabilities of large language models (LLMs) in economically-relevant tasks through economic and trade competition.
Outcome:	The proposed model evaluates the capabilities of large language models in economically-relevant tasks through economic and trade competition.

Copied to clipboard

Manyi Zhang, Ji-Fu Li, Zhongao Sun, Haoli Bai, Hui-Ling Zhen, Zhenhua Dong, Xianzhi Yu

HiEdit: Lifelong Model Editing with Hierarchical Reinforcement Learning (2026.acl-long)

Challenge:	Microscaling Floating-Point (MXFP) is a low-precision format for large language models (LLMs).
Approach:	They conduct systematic evaluations of PTQ under Microscaling Floating-Point (MXFP) . they find that MXFP8 consistently achieves near-lossless performance .
Outcome:	The proposed method achieves near-lossless performance while MXFP4 introduces substantial accuracy degradation and remains challenging.

Copied to clipboard

Yangfan Wang, Tianyang Sun, Chen Tang, Jie Liu, Wei Cai, Jingchi Jiang

Challenge:	Existing approaches to lifelong model editing apply parameter perturbations to static and dense layers for all instances.
Approach:	They propose a hierarchical reinforcement learning framework that identifies the most knowledge-relevant layers for each editing instance.
Outcome:	The proposed framework boosts the performance of the competitive RLEdit by 8.48% with perturbing only half of the layers per edit.

Too Long, Do Re-weighting for Efficient LLM Reasoning Compression (2026.acl-long)

Copied to clipboard

Zhong-Zhi Li, Xiao Liang, Zihao Tang, Lei Ji, Peijie Wang, Haotian Xu, Xing W, Haizhen Huang, Weiwei Deng, Yeyun Gong, Zhijiang Guo, Xiao Liu, Fei Yin, Cheng-Lin Liu

Measuring Social Bias in Vision-Language Models with Face-Only Counterfactuals from Real Photos (2026.acl-long)

Challenge:	Large Language Models (LLMs) have recently achieved remarkable progress on complex reasoning tasks by leveraging extended Chain-of-Thought (CoT) techniques.
Approach:	They propose a method that uses Extended Chain-of-Thought (EFT) to reduce the number of output tokens by nearly 40% while maintaining the accuracy of the reasoning.
Outcome:	The proposed method reduces the number of output tokens by nearly 40% while maintaining the accuracy of the reasoning.

Copied to clipboard

Haodong Chen, Qiang Huang, Jiaqi Zhao, Qiuping Jiang, Xiaojun Chang, Jun Yu

KASER: Knowledge-Aligned Student Error Simulator for Open-Ended Coding Tasks (2026.acl-long)

Challenge:	Vision-Language Models (VLMs) are increasingly deployed in socially consequential settings . attribution under visual confounding is a central challenge in measuring social bias .
Approach:	They propose a face-only counterfactual evaluation paradigm that isolates demographic effects while preserving real-image realism.
Outcome:	The proposed paradigm isolates demographic effects while preserving real-image realism.

Copied to clipboard

Zhangqi Duan, Nigel Fernandez, Andrew Lan

Challenge:	coding tasks that provide detailed insights into student knowledge are challenging to train . open-ended tasks often suffer from mode collapse and fail to capture student errors .
Approach:	They propose a method that aligns errors with student knowledge by using a hybrid reward system.
Outcome:	The proposed method outperforms baselines on code and error prediction and error coverage and simulated code diversity on two real-world datasets.

GroupToM-Bench: Benchmarking Group Theory of Mind and Nonlinear Social Emergence in MLLMs (2026.acl-long)

Copied to clipboard

Weidong Tang, Jierui Li, Yueling Hou, Zihan Mei, Can Zhang, Xinyan Wan, Zhiyuan Liang, Pengfei Zhou, Yang You, Wangbo Zhao

Assessing the Belief Consistency of Large Language Models on the Logical Conversation Process (2026.acl-long)

Challenge:	Existing models for general intelligence fail to model how mental states interact and crystallize into group-level outcomes.
Approach:	They propose a multimodal benchmark for group-level Theory of Mind (ToM) to probe nonlinear collective behavior.
Outcome:	The proposed model performs significantly below human levels, exposing blind spots in modeling social structures and nonlinear collective behavior.

Copied to clipboard

Tomoki Tsujimura, Matīss Rikters, Masaki Asada, Shusaku Egami, Tatsuya Ishigaki, Ken Yano, Hiroya Takamura

Textual Steering Vectors Can Improve Visual Understanding in Multimodal Large Language Models (2026.acl-long)

Challenge:	Large language models have been shown remarkable ability to understand given contexts.
Approach:	They propose a method to evaluate whether beliefs held by LLMs remain consistent . they propose to use multiple choice question answering format to assess belief consistency .
Outcome:	The proposed method evaluates the consistency of LLMs in a multiple-choice question answering format.

Copied to clipboard

Woody Haosheng Gan, Deqing Fu, Julian Asilis, Ollie Liu, Vatsal Sharan, Robin Jia, Willie Neiswanger

Learning Uncertainty from Sequential Internal Dispersion in Large Language Models (2026.acl-long)

Challenge:	Steering methods have emerged as effective tools for guiding large language models’ behavior, yet multimodal large language model lacks comparable techniques due to architectural diversity and limited availability of multimodal steering vectors.
Approach:	They validate steering vectors derived solely from text-only LLM backbones and use a cross-modal transfer technique to reuse existing interpretability tools.
Outcome:	The proposed steering vectors can guide and enhance multimodal models using SPAR, Mean Shift, and Linear Probing.

Copied to clipboard

Ponhvoan Srey, Xiaobao Wu, Cong-Duy T Nguyen, Anh Tuan Luu

On-the-Fly VLA Adaptation via Test-Time Reinforcement Learning (2026.acl-long)

Challenge:	Recent approaches to detect hallucinations depend on model internal states to estimate uncertainty, but they focus on last or mean tokens.
Approach:	They propose a supervised hallucination detection framework that leverages token-wise, layer-wise features derived from hidden states.
Outcome:	The proposed framework outperforms baseline models and avoids large training sets.

Copied to clipboard

Changyu Liu, Yiyang Liu, Taowen Wang, Qiao Zhuang, James Chenhao Liang, Wenhao Yang, Renjing Xu, Qifan Wang, Dongfang Liu, Cheng Han

AdaMix: Adaptive Mixing for Short and Long Reasoning Adapters (2026.acl-long)

Challenge:	Existing vision-language-action models are unsuitable for simulated or physical-world deployments . current methods fail when confronted with inherent real-world dynamic variability.
Approach:	They propose a test-time reinforcement learning framework that enables on-the-fly policy adaptation during inference.
Outcome:	Empirical results show that the proposed framework improves adaptability, stability and task success in dynamic, previously unseen scenarios.

Copied to clipboard

Hao Luo, Xiao Yan, Xinyan Li, Qiming Zeng, Yuhao Lin, Shanshan Feng, Hao Wang, Jiawei Jiang

Quantifying and Mitigating Socially Desirable Responding in LLMs: A Desirability-Matched Graded Forced-Choice Psychometric Study (2026.acl-long)

Challenge:	Existing methods for large reasoning models have improved efficiency but still face limitations such as conflicting objectives and limited adaptability.
Approach:	They propose an adaptive reasoning framework that applies a uniform, computation-intensive deep reasoning strategy to all problems.
Outcome:	The proposed framework reduces the average response length of DeepSeek-R1-Distill-Qwen-7B by 54.9% while improving accuracy by up to 4.8% on five mathematical datasets.

Copied to clipboard

Kensuke Okada, Yui Furukawa, Kyosuke Bunji

Challenge:	a psychometric framework is proposed to quantify and mitigate socially desirable responding (SDR) in questionnaire-based evaluation of large language models.
Approach:	They propose a psychometric framework to quantify and mitigate socially desirable responding (SDR) in questionnaire-based evaluation of large language models.
Outcome:	The proposed framework quantifies and mitigates socially desirable responding (SDR) in questionnaire-based evaluation of large language models.

From Curated Data to Scalable Models: Continual Pre-training of Dense and MoE Large Language Models for Tibetan (2026.acl-long)

Copied to clipboard

Lei Yang, Leiyu Pan, Bojian Xiong, Renren Jin, Shaowei Zhang, Yue Chen, Ling Shi, Jiang Zhou, Junru Wu, Zhen Wang, Jianxiang Peng, Juesi Xiao, Tianyu Dong, Zhuowen Han, Zhuo Chen, Yuqi Ren, Deyi Xiong

UrbanGeoEval: A City-Scale Benchmark for Evaluating Large Language Models in Geospatial Reasoning (2026.acl-long)

Challenge:	Large language models have achieved remarkable success across a wide range of tasks, yet their performance remains heavily biased toward high-resource languages.
Approach:	They propose a pipeline for advancing Tibetan language modeling through multilingual continual pre-training with Tibetan, Chinese, and English.
Outcome:	The proposed model outperforms open-source and Tibetan-focused models on diverse tasks.

Copied to clipboard

Mutian Bao, Qiuyi Qi, Tian Liang, Jinjian Zhang, Wei Zhou, Ming Kong, Linjian Mo, Qiang Zhu

TAIGR: Towards Modeling Influencer Content on Social Media via Structured, Pragmatic Inference (2026.acl-long)

Challenge:	Extensive experiments on 18 widely used LLMs uncover critical insights: (1) models exhibit severe geographic biases and resolution gaps; (2) failures in complex multi-hop tasks stem from brittle foundational spatial skills rather than high-level logic deficits.
Approach:	They propose a dual-module framework that disentangles factual recall and spatial logic from the model's real capabilities in urban environments.
Outcome:	Extensive tests on 18 widely used LLMs reveal that models exhibit severe geographic biases and resolution gaps, and failures in complex multi-hop tasks often stem from brittle foundational spatial skills rather than high-level logic deficits.

Copied to clipboard

Nishanth Sridhar Nakshatri, Eylon Caplan, Rajkumar Pujari, Dan Goldwasser

Afri-MCQA: Multimodal Cultural Question Answering for African Languages (2026.acl-long)

Challenge:	Health influencers are often conveyed through conversational narratives and rhetorical strategies rather than explicit factual claims.
Approach:	They propose a framework to analyze influencer discourse using takeaway argumentation inference with Grounded References.
Outcome:	The proposed framework is based on a content validation task over influencer video transcripts on health, showing that accurate validation requires modeling the discourse’s pragmatic and argumentative structure rather than treating transcripts as flat collections of claims.

Copied to clipboard

Atnafu Lambebo Tonja, Srija Anand, Emilio Villa-Cueva, Israel Abebe Azime, Jesujoba Oluwadara Alabi, Muhidin A. Mohamed, Debela Desalegn Yadeta, Negasi Haile Abadi, Abigail Oppong, Nnaemeka Casmir Obiefuna, Idris Abdulmumin, Naome A Etori, Eric Peter Wairagala, Kanda Patrick Tshinu, Imanigirimbabazi Emmanuel, Gabofetswe Malema, Alham Fikri Aji, David Ifeoluwa Adelani, Thamar Solorio

E-ViC: Reasoning Beyond Text via Embodied Visual Chain for Spatial Intelligence (2026.acl-long)

Challenge:	Afri-MCQA is the first multilingual cultural question-answering benchmark containing 7.5k Q A pairs across 15 African languages from 12 countries.
Approach:	They introduce Afri-MCQA, the first multilingual cultural question-answering benchmark containing 7.5k Q A pairs across 15 African languages from 12 countries.
Outcome:	The proposed model shows poor performance across cultures, with near zero accuracy on open-ended VQA when queried through native language or speech.

Copied to clipboard

Junbo Qi, Yi Zhang, Hanchu Ni, Che Liu, Zhimin Yao, Ruilin Yang, Xiancong Ren, Liangjian Wen, Wei Ge, Yuya Ieiri, Osamu Yoshie, Yong Dai, Xiaozhu Ju

MessToClean: Evidence-Grounded Structure-Preserving Reconstruction for Real-World Degraded Exam Paper Images (2026.acl-long)

Challenge:	Existing Vision-Language Models (VLMs) lack spatial reasoning, despite text-based CoTs . e-ViC reframes spatial intelligence as a verifiable, tool-using capability, argues a new study.
Approach:	They propose a framework that moves reasoning beyond text into the visual domain . they ground reasoning in pixel-level interactions to enable human-like "look-and-confirm" strategies .
Outcome:	The proposed framework outperforms existing Vision-Language Models with an average gain of 10.1%.

Copied to clipboard

Jiayi Tuo, Cheng Tang, Zihan Wang, Chenyue Zhou, Yao Li, Yanbiao Ma, Chao Wang, Wei Dai, Mingxuan Wang, Shitong Qin, Ziwei Zhao

Phonemes to the Rescue: Multilingual Tokenization Based on International Phonetic Alphabet (2026.acl-long)

Challenge:	Existing Multimodal Large Language Models (MLLMs) fail under RDEI, leading to disrupted structure and evidence-unsupported hallucinations.
Approach:	They propose a backbone-agnostic, evidence-driven pipeline that treats off-the-shelf MLLMs as interchangeable components to improve stem consistency and figure consistency.
Outcome:	The proposed pipeline improves stem consistency by 1.01-3.18%, figure consistency by 0.50-49.16%, and refusal F1 by 1.06-10.88% across question types.

Copied to clipboard

Milan Miletić, Julie Kallini, Ekaterina Shutova

Structure Guided Retrieval-Augmented Generation for Factual Queries (2026.acl-long)

Challenge:	Widely-used subword tokenization approaches favor high-resource languages and tokenizer-free methods yield longer sequences for scripts with a higher bytes-per-character ratio.
Approach:	They propose to use the International Phonetic Alphabet (IPA) as a language-agnostic input representation for multilingual tokenizers.
Outcome:	The proposed model improves tokenization quality and generalizes more effectively to unseen languages and scripts.

Copied to clipboard

Miao Xie, Xiao Zhang, Yi Li, Chunli Lv

GROKE: Vision-Free Navigation Instruction Evaluation via Graph Reasoning on OpenStreetMap (2026.acl-long)

Challenge:	Existing methods for RAG produce factually incorrect outputs, resulting in incorrect answers.
Approach:	They propose a novel problem that explicitly incorporates structural information into RAG for factual questions to satisfy all query conditions.
Outcome:	The proposed method significantly outperforms baselines on ERQA while maintaining reasonable computational overhead.

Copied to clipboard

Farzad Shami, Subhrasankha Dey, Nico Van de Weghe, Henrikki Tenkanen

Emotion-Wheel-Guided Audio-Referred Text Representation for Multimodal Emotion Recognition in Conversation (2026.acl-long)

Challenge:	Existing evaluation metrics for navigation instructions are limited by licensing constraints and computational costs.
Approach:	They propose a vision-free training-free hierarchical LLM-based framework for evaluating navigation instructions using OpenStreetMap data.
Outcome:	The proposed framework reduces navigation error by 68.5% compared to baselines on the Map2Seq dataset.

Copied to clipboard

Eunseon Seong, Harim Lee, Dahye Kim, Changhyun Kim, Dong-Kyu Chae

LongVideoAgent: Multi-Agent Reasoning with Long Videos (2026.acl-long)

Challenge:	Existing methods for Emotion Recognition in Conversation ignore their distinct communicative roles and information capacities and apply uniform penalties regardless of affective proximity.
Approach:	They propose a modality-aware fusion strategy capturing linguistic features from text as the primary source and audio as a complementary component.
Outcome:	The proposed method captures linguistic features from text as the primary source and audio as a complementary component and supervised contrastive loss to encode emotional proximity based on Russell’s circumplex model.

Copied to clipboard

Runtao Liu, Ziyi Liu, Jiaqi Tang, Yue Ma, Renjie Pi, Jipeng Zhang, Qifeng Chen

BOSCH: Black-Box Binary Optimization for Short-Context Attention-Head Selection in LLMs (2026.acl-long)

Challenge:	a key emerging challenge is robust long video understanding, authors say . current methods compress content into lossy summaries or rely on limited toolsets .
Approach:	They propose a multi-agent framework where a master LLM coordinates a grounding agent and a vision agent to extract targeted textual observations.
Outcome:	The proposed model outperforms strong non-agent baselines on episode-level datasets . the proposed model significantly outperformed existing models on other datasets.

Copied to clipboard

Abbas Ghaddar, Ivan Kobyzev, Boxing Chen, Yufei Cui

Scaling Reasoning, Losing Control: Evaluating Instruction Following in Large Reasoning Models (2026.acl-long)

Challenge:	Existing hybridization schemes use sliding-window attention (SWA) to reduce KV cache usage and improve latency.
Approach:	They propose a training-free method that decomposes a large neighborhood search problem into three subproblems and a method that uses black-box binary optimization for short-context head selection.
Outcome:	Extensive experiments on 4 LLMs show that BOSCH outperforms layer-level heuristics and 6 strong static head-level methods with larger gains at higher SWA ratios.

Copied to clipboard

Tingchen Fu, Yafu Li, Jiawei Gu, Xiaoye Qu, Yu Cheng

MirrorQA: Benchmarking Multimodal LLMs on Mirror-Orientation Reasoning (2026.acl-long)

Challenge:	Recent advances in reasoning-oriented models have demonstrated impressive capabilities in mathematical reasoning, but their ability to adhere to user directives remains underexplored.
Approach:	They propose a benchmark to evaluate instruction-following in mathematical reasoning tasks.
Outcome:	The proposed model degrades in instruction adherence when generation length increases, but can partially recover obedience, despite increasing generation length.

Copied to clipboard

Jingping Liu, Xingchen Peng, Yan Zhou, Ziyan Liu, Jie Zhai, Ronghao Chen, Huacan Wang, Xiaofeng Jia

Robust Tool Use via Fission-GRPO: Learning to Recover from Execution Errors (2026.acl-long)

Challenge:	Multimodal large language models (MLLMs) have achieved remarkable progress in recent years, yet their ability to perform left–right reasoning in mirror contexts remains underexplored.
Approach:	They propose a benchmark to evaluate MLLMs' ability to distinguish left from right from a subject-centered perspective.
Outcome:	The proposed benchmarks show that even the best performing models achieve only 65.40% accuracy, far below the 99.28% accuracy of humans.

Copied to clipboard

Zhiwei Zhang, Fei Zhao, Rui Wang, Zezhong Wang, Bin Liang, Jiakang Wang, Yao Hu, Shaosheng Cao, Kam-Fai Wong

DimABSA: Building Multilingual and Multidomain Datasets for Dimensional Aspect-Based Sentiment Analysis (2026.acl-long)

Challenge:	Large language models (LLMs) can call tools effectively, but they remain brittle in multi-turn execution.
Approach:	They propose a framework that converts execution errors into on-policy corrective supervision within the RL training loop.
Outcome:	The proposed framework improves the error recovery rate of Qwen3-8B by 5.7% absolute and overall accuracy by 4.0% on BFCL v4 Multi-Turn.

Copied to clipboard

Lung-Hao Lee, Liang-Chih Yu, Natalia V Loukachevitch, Ilseyar Alimova, Alexander Panchenko, Tzu-Mi Lin, Zhe-Yu Xu, Jian-Yu Zhou, Guangmin Zheng, Jin Wang, Sharanya Awasthi, Jonas Becker, Jan Philip Wahle, Terry Ruas, Shamsuddeen Hassan Muhammad, Saif M. Mohammad

CAP: Controllable Alignment Prompting for Unlearning in LLMs (2026.acl-long)

Challenge:	Existing ABSA research relies on coarse-grained categorical labels, which limits its ability to capture nuanced affective states.
Approach:	They propose a dimensional approach that represents sentiment with continuous valence–arousal (VA) scores, enabling fine-grained analysis at both the aspect and sentiment levels.
Outcome:	The proposed approach represents sentiment with continuous valence–arousal (VA) scores, enabling fine-grained analysis at both the aspect and sentiment levels.

Copied to clipboard

Zhaokun Wang, Jinyu Guo, Jingwen Pu, Hongli Pu, Meng Yang, Xunlei Chen, Jie Ou, Wenyi Li, Guangchun Luo, Wenhong Tian

GeoRC: A Benchmark for Geolocation Reasoning Chains (2026.acl-long)

Challenge:	Existing methods for modifying parameters are unsystematic and rely on empirical experience.
Approach:	They propose a controllable alignment prompting for unlearning framework that decouples unlearning into a learnable prompt optimization process via reinforcement learning.
Outcome:	The proposed framework achieves precise, controllable unlearning without updating model parameters.

Copied to clipboard

Mohit Talreja, Joshua Diao, Jim James, Radu Casapu, Tejas Santanam, Ethan Mendes, Alan Ritter, Wei Xu, James Hays

SemCSE-Multi: Multifaceted and Decodable Embeddings for Aspect-Specific and Interpretable Scientific Domain Mapping (2026.acl-long)

Challenge:	Vision Language Models (VLMs) are good at recognizing the global location of a photograph but are startlingly bad at explaining which image evidence led to their location prediction.
Approach:	They propose a benchmark for geolocation reasoning chains based on the global location prediction task in the popular GeoGuessr game.
Outcome:	The proposed benchmark compares LLM-as-a-judge and VLM-As-jumble strategies against human scoring.

Copied to clipboard

Marc Brinner, Sina Zarrieß

Rendering Data Unlearnable by Exploiting LLM Alignment Mechanisms (2026.acl-long)

Challenge:	Existing embedding models rely on implicit, imprecise and fixed notion of similarity to evaluate scientific abstracts.
Approach:	They propose a framework for generating multifaceted embeddings of scientific abstracts . they propose an unsupervised procedure that produces aspect-specific summarizing sentences .
Outcome:	The proposed framework captures distinct, individually specifiable aspects in isolation . it then trains embedding models to map semantically related summaries to nearby positions . the proposed framework is evaluated in the domains of invasion biology and medicine .

Copied to clipboard

Ruihan Zhang, Jun Sun

CAR-bench: Evaluating the Consistency and Limit-Awareness of LLM Agents under Real-World Uncertainty (2026.acl-long)

Challenge:	Large language models (LLMs) are increasingly trained on massive, heterogeneous text corpora, raising serious concerns about the unauthorised use of proprietary or personal data during model training.
Approach:	They propose a data-level defence that renders text unlearnable to LLMs by injecting carefully designed alignment-triggering disclaimers into the models' alignment mechanisms.
Outcome:	The proposed approach exploits the models’ alignment mechanisms to prevent effective learning.

Copied to clipboard

Johannes Kirmayr, Lukas Stappen, Elisabeth Andre

Tiny Scales, Great Challenges: The Limits of Multimodal LLMs in Scale Recognition (2026.acl-long)

Challenge:	Existing benchmarks for Large Language Model (LLM) agents focus on task completion under idealistic settings but overlook reliability in real-world, user-facing applications.
Approach:	They propose a benchmark to evaluate consistency, uncertainty handling, and capability awareness in multi-turn, tool-using LLM agents in an in-car assistant domain.
Outcome:	The proposed benchmarks evaluate consistency, uncertainty handling, and capability awareness in multi-turn, tool-using LLM agents in an in-car assistant domain.

Copied to clipboard

Jihang Jin, Ronghao Chen, Hao Zhang, Ziyan Liu, Huacan Wang, Qi Ye, Jingping Liu

M3-VQA: A Benchmark for Multimodal, Multi-Entity, Multi-Hop Visual Question Answering (2026.acl-long)

Challenge:	Existing benchmarks focus on a single type of quantity or a specific format, lacking a comprehensive evaluation of scale recognition capabilities.
Approach:	They propose a visual scale recognition benchmark built using images from COCO, Open Images, and Flickr to evaluate scale recognition capabilities of multimodal large language models.
Outcome:	The proposed model achieves 42.60% accuracy, lower than the 97.40% of humans.

Copied to clipboard

Jiatong Ma, Longteng Guo, Yuchen Liu, Zijia Zhao, Dongze Hao, Xuanxu Lin, Jing Liu

From Selection to Refinement: Iterative Optimization for Instruction Data (2026.acl-long)

Challenge:	Existing knowledge-based VQA benchmarks focus on coarse-grained categories and simple reasoning over single entities.
Approach:	They propose a knowledge-based Visual Question Answering benchmark to enhance multimodality evaluation.
Outcome:	The proposed benchmark improves evaluation of multimodal large language models in fine-grained multimodal entity understanding and complex multihop reasoning.

Copied to clipboard

Hang Hu, Ziyan Liu, Rujie Wen, Ruihui Hou, Xueyan Wu, Mu Zhang, Jianxing Yu, Tong Ruan, Jingping Liu

VISTA: Verification In Sequential Turn-based Assessment (2026.acl-long)

Challenge:	Existing methods to optimize instruction tuning datasets face two main challenges: unreasonable pruning of potentially valuable low-quality data and the persistence of noise or semantic drift during revision.
Approach:	They propose an automated iterative framework for instruction data optimization that prunes low-quality data and refines low quality data using feedback-driven iteration.
Outcome:	The proposed framework outperforms state-of-the-art methods on seven public benchmark datasets with high data efficiency.

Copied to clipboard

Ashley Lewis, Andrew Perrault, Eric Fosler-Lussier, Michael White

Knowledge-to-Verification: Exploring RLVR for LLMs in Knowledge-Intensive Domains (2026.acl-long)

Challenge:	Existing metrics evaluate isolated responses or treat unverifiable content as errors, limiting their use for multi-turn dialogue.
Approach:	They propose a framework for evaluating conversational factuality via claim-level verification and sequential consistency tracking.
Outcome:	The proposed framework improves hallucination detection over existing benchmarks and models.

Copied to clipboard

Zhonghang Yuan, Zhefan Wang, Fang Hu, Zihong Chen, Jinzhe Li, Gang Li, Jie Ying, Huanjun Kong, Songyang Zhang, Nanqing Dong

Mem-Gallery: Benchmarking Multimodal Long-Term Conversational Memory for MLLM Agents (2026.acl-long)

Challenge:	Recent large language models (LLMs) have demonstrated remarkable progress in reasoning, but their applications on knowledge-intensive domains have not been explored due to the scarcity of high-quality verifiable data.
Approach:	They propose a framework that extends reinforcement learning with verifiable rewards (RLVR) to knowledge-intensive domains through automated verififiability data synthesis while enabling verification of the LLM's reasoning process.
Outcome:	Extensive experiments show that the proposed framework enhances the reasoning of large language models in knowledge-intensive domains without significantly compromising the model’s general capabilities.

Copied to clipboard

Yuanchen Bei, Tianxin Wei, Xuying Ning, Yanjun Zhao, Zhining Liu, Xiao Lin, Yada Zhu, Hendrik Hamann, Jingrui He, Hanghang Tong

HarDBench: A Benchmark for Draft-Based Co-Authoring Jailbreak Attacks for Safe Human–LLM Collaborative Writing (2026.acl-long)

Challenge:	Existing benchmarks evaluate multi-session memory in text-only conversations or assess multimodal understanding within localized contexts.
Approach:	They propose a benchmark for evaluating multimodal long-term conversational memory in MLLM agents.
Outcome:	The proposed framework assesses key memory capabilities along three functional dimensions: memory extraction and test-time adaptation, memory reasoning, and memory knowledge management.

Copied to clipboard

EunTae Kim, Soomin Han, Buru Chang

Mind Reader: Latent User Demand-Guided Content Optimization for Generative Search Engine (2026.acl-long)

Challenge:	Large language models are increasingly used as coauthors in collaborative writing . however, this capability poses a serious safety risk .
Approach:	They propose a safety-utility balanced alignment approach to train LLMs to refuse harmful completions while remaining helpful on benign drafts.
Outcome:	The proposed method reduces harmful outputs without degrading performance on co-authoring capabilities.

Copied to clipboard

Tong Chen, JiaWei Guo, Yuxi Li, Baiming Chen, Houxing Ren, Zhang Zhiwei, Yunxiang Zhang, Hanyang Xia, Kun Liang, Zhaoran Fan

Retrieval-Augmented Defense: Adaptive and Controllable Jailbreak Prevention for Large Language Models (2026.acl-long)

Challenge:	Generative Search Engines (GSEs) have reshaped information retrieval and Generating Engine Optimization (GEO) emerges to improve the content visibility in GSEs’ responses.
Approach:	They propose a method to optimize content to cover latent semantic information of GSEs by decomposing query into diverse perspectives and capturing underlying semantic information.
Outcome:	The proposed method outperforms baselines and effectively improves content visibility (with up to 2.44x objective metrics and 1.23x subjective metrics on average).

Copied to clipboard

Guangyu Yang, Jinghong Chen, Jingbiao Mei, Weizhe Lin, Bill Byrne

CNSL-bench: Benchmarking the Sign Language Understanding Capabilities of MLLMs on Chinese National Sign Language (2026.acl-long)

Challenge:	Large Language Models (LLMs) remain vulnerable to jailbreak attacks due to evolving nature and diversity of attack strategies.
Approach:	They propose a framework for jailbreak detection that integrates a database of known attack examples into Retrieval-Augmented Generation to infer the underlying, malicious user query and jailbreak strategy used to attack the system.
Outcome:	The proposed framework reduces the effectiveness of strong jailbreak attacks while maintaining low rejection rates for benign queries.

Copied to clipboard

Rui Zhao, Xuewen Zhong, Xiaoyun Zheng, Jinsong Su, Yidong Chen

Thinking in Schemas: Robust Syllogistic Reasoning in LLMs (2026.acl-long)

Challenge:	CNSL-bench is the first comprehensive Chinese National Sign Language benchmark . current MLLMs are inferior to human performance, despite advances in multimodal modeling .
Approach:	They propose a Chinese National Sign Language benchmark to evaluate multimodal large language models in sign language understanding.
Outcome:	The proposed benchmark evaluates 21 open-source and proprietary MLLMs . results show that current models are inferior to human performance .

Copied to clipboard

Federico Ranaldi, Leonardo Ranaldi, Fabio Massimo Zanzotto, Shay B Cohen

Amadeus: Autoregressive Model with Bidirectional Attribute Modelling for Symbolic Music (2026.acl-long)

Challenge:	syllogistic reasoning models often mistake what sounds true for what is formally valid . content effect is a limitation of sluggish reasoning, which can lead to invalid conclusions . eisape et al., 2024: a key open problem for formal inference in natural language.
Approach:	They propose a schema-guided framework that disentangles semantic plausibility from logical validity.
Outcome:	The proposed framework outperforms existing frameworks while reducing CE.

Copied to clipboard

Hongju Su, Ke Li, Lan Yang, Honggang Zhang, Yi-Zhe Song

FedMental: Evaluating Federated Learning for Mental Health Detection from Social Media Data (2026.acl-long)

Challenge:	Existing symbolic music generation models represent musical notes as a sequence of attribute tokens with fixed unidirectional dependencies.
Approach:	They propose a symbolic music generation framework that adopts a autoregressive and a discrete diffusion architectures for note attributes.
Outcome:	The proposed framework improves state-of-the-art models across objective and subjective metrics.

Copied to clipboard

Nuredin Ali Abdelkadir, Anjali Ratnam, Zeerak Talat, Stevie Chancellor

Edit-Aware Reward Modeling for Chinese Grammatical Error Correction (2026.acl-long)

Challenge:	Social media text data is often used to train machine learning models to identify users exhibiting high-risk mental health behaviors.
Approach:	They apply federatedlearning and Differentially Private FL to two widely-studied mental health prediction tasks using social media text data.
Outcome:	The proposed methods achieve comparable performance to centralized training on depression identification, but have a large performance-privacy trade-off even with low levels of noise.

Copied to clipboard

Yilin Li, Xiaojun Wan

GenPT: Beyond Self-Report for Reliable LLM Psychometrics via Generative Projective Testing (2026.acl-long)

Challenge:	Recent work has applied reinforcement learning with rule-based rewards to grammatical error correction tasks, but these methods fail to capture fine-grained quality distinctions among correction candidates.
Approach:	They propose an Edit-Aware Reward Model that explicitly incorporates edit-awareness into preference learning for CGEC.
Outcome:	The proposed model outperforms rule-based models on CGEC and other NLP tasks by 5.41 and 1.80 points.

Copied to clipboard

Ming Wang, Shuang Wu, Bixuan Wang, Lu Lin, Yuxin Chen, Xiaocui Yang, Daling Wang, Shi Feng, Yifei Zhang, Yufan Sun

Challenge:	Large language models (LLMs) inherit contamination from training corpora, directional bias under social-desirability framing, and limited responsiveness to context beyond the item text.
Approach:	They propose a paradigm that reformulates TAT, Rorschach, and SCT with newly generated stimuli and organises assessment as a three-stage pipeline.
Outcome:	The proposed paradigm reformulates TAT, Rorschach, and SCT with newly generated stimuli and organises assessment as a three-stage pipeline.

Still Between Us? Evaluating and Improving Voice Assistant Robustness to Third-Party Interruptions (2026.acl-long)

Copied to clipboard

Dongwook Lee, Eunwoo Song, Che Hyun Lee, Heeseung Kim, Sungroh Yoon

Learning to Conceal Risk: Controllable Multi-turn Red Teaming for LLMs in the Financial Domain (2026.acl-long)

Challenge:	Recent Spoken Language Models lack the capability to discern Third-Party Interruptions (TPI) from the primary user’s ongoing flow, leaving them vulnerable to contextual failures.
Approach:	They propose a dataset with speaker-aware hard negatives to enforce acoustic cue prioritization for interruption handling and a framework to measure the interruption-handling strategy and precise speaker discrimination in deceptive contexts.
Outcome:	The proposed framework mitigates semantic shortcut learning while neglecting acoustic signals essential for discerning speaker changes.

Copied to clipboard

Gang Cheng, Haibo Jin, Wenbin Zhang, Haohan Wang, Jun Zhuang

Beyond Static Alignment: Adaptive Arbitration for Semantic Incongruence in Semi-Supervised Multimodal Sentiment Analysis (2026.acl-long)

Challenge:	Large Language Models (LLMs) are increasingly deployed in high-stakes domains such as finance where unsafe behavior can lead to serious regulatory risks.
Approach:	They propose a black-box multi-turn risk-concealed redteaming framework that progressively conceals surface-level risk while exploiting regulatory-violating behaviors.
Outcome:	Experiments on nine widely used LLMs show that the proposed framework achieves 93.19% average attack success rate (ASR) and improves the average ASR to 95.00%.

Copied to clipboard

Huicong Li, Xiangbo Ji, Wei Wu

LLM Beliefs Are in Their Heads (2026.acl-long)

Challenge:	Existing methods for semantic incongruence in sentiment analysis are limited by label-limited settings.
Approach:	They propose a framework for semi-supervised multimodal sentiment analysis that emphasizes stable cross-modal representations and reliable supervision.
Outcome:	The proposed framework outperforms state-of-the-art methods under label-limited settings.

Copied to clipboard

Alessandro Corona Mendozza, Anders Søgaard

Challenge:	Using linear controlled probes, we investigate belief-like representations in decoder-only autoregressive LLMs using residual stream activations and single attention heads.
Approach:	They develop four different experiments on decoder-only autoregressive LLMs and examine how they fare against these standards.
Outcome:	The proposed representations exhibit strong truth sensitivity and consistent accuracy across models and data sets.

ProactiveEval: A Unified Evaluation Framework for Proactive Dialogue Agents (2026.acl-long)

Copied to clipboard

Tianjian Liu, Fanqi Wan, Jiajian Guo, Xiaojun Quan

The Visual Iconicity Challenge: Evaluating Vision-Language Models on Sign Language Form–Meaning Mapping (2026.acl-long)

Challenge:	Existing studies on proactive dialogue models focus on domain-specific or task-oriented scenarios, which leads to fragmented evaluations and limits the comprehensive exploration of models’ proactive dialogue abilities.
Approach:	They propose a framework for evaluating proactive dialogue capabilities of large language models that decomposes proactive dialogue into target planning and dialogue guidance, establishing evaluation metrics across various domains.
Outcome:	The proposed framework decomposes proactive dialogue into target planning and dialogue guidance, establishing evaluation metrics across various domains, and enables automatic generation of diverse and challenging evaluation data.

Copied to clipboard

Onur Keleş, Asli Ozyurek, Gerardo Ortega, Kadir Gökgöz, Esam Ghaleb

From Competition to Synergy: Unlocking Reinforcement Learning for Subject-Driven Image Generation (2026.acl-long)

Challenge:	a visual Iconicity test is used to evaluate vision–language models based on visual form and iconicity ratings.
Approach:	They propose a video-based benchmark to evaluate vision–language models on three tasks . they assess 17 state-of-the-art VLMs in zero- and few-shot settings on Sign Language of the Netherlands .
Outcome:	The proposed benchmark evaluates 17 state-of-the-art VLMs on Sign Language of the Netherlands . they achieve moderate to strong alignment with human iconicity ratings, but fail to infer lexical meaning from visual form alone .

Copied to clipboard

Ziwei Huang, Ying Shu, null Fanghao, Quanyu Long, Wenya Wang, Qiushi Guo, Tiezheng Ge, Leilei Gan

Deep-Reporter: Deep Research for Grounded Multimodal Long-Form Generation (2026.acl-long)

Challenge:	a naive application of GRPO leads to conflicting gradient signals and a misalignment with the temporal dynamics of the diffusion process.
Approach:	They propose a framework that uses synergy-aware reward shaping to penalize conflicted reward signals and amplify synergies to provide a sharper and decisive gradient.
Outcome:	The proposed framework outperforms naive GRPO and Time-Aware Dynamic Weighting (TDW) on DreamBench, and achieves a state-of-the-art balance between ID preservation and prompt adherence.

Copied to clipboard

Fangda Ye, Kuicai Dong, Xie Zhifei, Yuxin Hu, Yihang Yin, Shurui Huang, Shikai Dong, Chen Zhang, Jianzhu Bao, Shuicheng Yan

Embodied-Reasoner: Synergizing Visual Search, Reasoning, and Action for Embodied Interactive Tasks (2026.acl-long)

Challenge:	Recent agentic search frameworks are text-centric, overlooking multimodal evidence . a pressing task is multimodal long-form generation, a new paper argues .
Approach:	They propose a unified agentic framework for grounded multimodal long-form generation.
Outcome:	The proposed framework is based on a unified agentic framework for grounded multimodal long-form generation.

Copied to clipboard

Wenqi Zhang, Mengna Wang, Gangao Liu, Huixin Xu, Yiwei Jiang, Yongliang Shen, Guiyang Hou, Zhe Zheng, Hang Zhang, Xin Li, Jiajun Liu, Weiming Lu, Peng Li, Yueting Zhuang

Temporal Precision Matters: Brain-Tuning Speech Language Models with Millisecond-Resolution Neural Signals (2026.acl-long)

Challenge:	Recent advances in reasoning models have demonstrated remarkable capabilities on mathematical and coding tasks, but their effectiveness in embodied domains remains largely unexplored.
Approach:	They propose a reasoning model for interactive embodied tasks that synthesizes 9.3k coherent Observation-Thought-Action trajectories containing 64k ego-centric images and 90k diverse reasoning processes.
Outcome:	The proposed model outperforms existing visual reasoning models by +9%, 24%, and +13% on long-horizon tasks.

Copied to clipboard

Zhejun Zhang, Wenqing Zhou, Haozhe Xu, Lin Zhang, Lei Li

One Pair Suffices: Unlocking Universal Zero-Shot Translation via Cross-Architecture Alignment (2026.acl-long)

Challenge:	Language models have emerged as powerful tools for predicting human brain activity during language comprehension.
Approach:	They propose a technique that leverages electrocorticography’s millisecond precision to train speech language models.
Outcome:	The proposed technique improves brain alignment over pretrained and distillation models and produces higher gains in higher-order language regions.

Copied to clipboard

Hao Zong, Cong Hu Yuan, Chao Bei, Wentao Chen, Huan Liu, Kaiyu Huang, Degen Huang

LASA: Language-Agnostic Semantic Alignment at the Semantic Bottleneck for LLM Safety (2026.acl-long)

Challenge:	Current paradigms for empowering Large Language Models with multilingual capabilities rely heavily on massive instruction tuning.
Approach:	They propose a hybrid cross-alignment approach that fuses a frozen NLLB encoder with a Qwen decoder via a closed-loop dual-adapter architecture.
Outcome:	The proposed model outperforms towerPlus-9B and Aya-101 on language-agnostic projection protocols.

Copied to clipboard

Junxiao Yang, Haoran Liu, Jinzhe Tu, Jiale Cheng, Zhexin Zhang, Shiyao Cui, Jiaqi Weng, Jialing Tao, Hui Xue, Hongning Wang, Han Qiu, Minlie Huang

Challenge:	Large language models (LLMs) have demonstrated better safety performance in high-resource languages than in low-resourced languages.
Approach:	They propose language-agnostic semantic alignment (LASA) which anchors safety alignment directly in semantic bottlenecks.
Outcome:	The proposed approach significantly improves safety across all languages: average attack success rate drops from 24.7% to 2.8% on LLaMA-3.1-8B-Instruct and remains within 3–4% across Qwen2.5 and Qwend3 Instruct models (7B–32B).

Modeling Annotator Disagreement with Demographic-Aware Experts and Synthetic Perspectives (2026.acl-long)

Copied to clipboard

Yinuo Xu, Veronica Derricks, Allison Earl, David Jurgens

Challenge:	Recent work treats disagreements as signal, instead of noise, resulting in a single label and marginalizing minoritized perspectives.
Approach:	They propose an approach to modeling annotator disagreement in subjective NLP tasks through architectural and data-centric innovations.
Outcome:	The proposed model performs competitively across demographic groups and shows strong results on datasets with high disagreement.

Multimodal Safety Evaluation in Generative Agent Social Simulations (2026.acl-long)

Copied to clipboard

Alhim Adonai Vera Gonzalez, Carlos Hinojosa, Karen Sanchez, Haidar Bin Hamid, Donghoon Kim, Bernard Ghanem

Quantifying the Impact of Translation Errors on Multilingual LLM Evaluation (2026.acl-long)

Challenge:	Recent advances in large language models have enabled generative agents that simulate be-like behavior through natural language interactions.
Approach:	They propose a reproducible simulation framework to evaluate generative agents in multimodal scenarios . they use metrics that quantify plan revisions and unsafe-to-safe conversions to evaluate their effectiveness .
Outcome:	The proposed framework evaluates generative agents in three aspects: safety improvement over time, detection of unsafe activities across social contexts, social dynamics and acceptance rates.

Copied to clipboard

Klaudia Thellmann, Bernhard Stadler, Michael Färber, Jens Lehmann

Long-Context Reasoning Through Proxy-Based Chain-of-Thought Tuning (2026.acl-long)

Challenge:	Machine-translated benchmarks are widely used to assess the multilingual capabilities of large language models (LLMs), yet translation errors in these benchmarks remain underexplored.
Approach:	They show how well machine-translated benchmarks match human span annotations on translations . they also show how strongly translation errors explain accuracy drops on translated benchmarks - a gap that is not addressed yet .
Outcome:	The proposed model matches human-level translations with human-language annotations on translations, but translation errors are associated with accuracy drops even after controlling for English correctness and source-side anomalies.

Copied to clipboard

Miao Li, Irina Saparina, Alexander Gurung, Mirella Lapata

MASPO: Unifying Gradient Utilization, Probability Mass, and Signal Reliability for Robust and Sample-Efficient LLM Reasoning (2026.acl-long)

Challenge:	Recent large language models support inputs of up to 10 million tokens, yet they perform poorly on long-context tasks.
Approach:	They propose a training framework that transfers reasoning capabilities from proxy contexts to full long contexts.
Outcome:	The proposed framework outperforms baseline models with reduced computational overhead.

Copied to clipboard

Xiaoliang Fu, Jiaye Lin, Yangyi Fang, Binbin Zheng, Chaowen Hu, Zekai Shao, Cong Qin, Lu Pan, Ke Zeng, Xunliang Cai

Towards A Scanpath-Conditioned Surprisal Theory: Modeling Reader Information States (2026.acl-long)

Challenge:	Existing RLVR algorithms rely on rigid, uniform, and symmetric trust region mechanisms . current algorithms lack robustness, asymmetric signal reliability and inefficient gradient utilization .
Approach:	They propose a framework to harmonize three dimensions of RLVR algorithms, a paper argues . a binary cutoff is used to discard valuable reinforcement signals, they argue .
Outcome:	The proposed framework outperforms baselines in evaluating a robust RLVR solution.

Copied to clipboard

Michael Mooney, Edmond S. L. Ho

SafeMT: Multi-turn Safety for Multimodal Language Models (2026.acl-long)

Challenge:	Standard surprisal is computed from the linear text prefix, but human reading is non-linear and memory constrained.
Approach:	They propose a formulation of surprisal conditioned on a reader-specific accessible information state given by the scanpath history and memory dynamics rather than by the written prefix alone.
Outcome:	The proposed approach improves on eye-tracking measures on the written prefix and on eye movement data on human reading.

Copied to clipboard

Han Zhu, Juntao Dai, Jiaming Ji, Haoran Li, Chengkun Cai, Pengcheng Wen, Chi-Min Chan, Boyuan Chen, Yaodong Yang, Sirui Han, Yike Guo

From log 𝜋 to 𝜋: Taming Divergence in Soft Clipping via Bilateral Decoupled Decay of Probability Gradient Weight (2026.acl-long)

Challenge:	Multi-turn dialogues pose a greater risk than single prompts, but existing safety benchmarks do not account for this situation.
Approach:	They propose a benchmark that features dialogues of varying lengths generated from harmful queries accompanied by images.
Outcome:	The proposed model reduces multi-turn Attack Success Rate (ASR) compared to existing guard models.

Copied to clipboard

Xiaoliang Fu, Jiaye Lin, Yangyi Fang, Chaowen Hu, Cong Qin, Zekai Shao, Binbin Zheng, Lu Pan, Ke Zeng

Paramanu: Compact and Competitive Monolingual Language Models for Low-Resource Morphologically Rich Indian Languages (2026.acl-long)

Challenge:	Standard algorithms for Large Language Models (LLMs) enforce stability via "hard clipping" but relying on log-probability gradient yields divergent weights as probabilities vanish, destabilizing LLM training.
Approach:	They propose a decoupled gradient policy optimization that uses a decay mechanism to decouple the probability of a boundary token.
Outcome:	The proposed algorithm outperforms baselines on various mathematical benchmarks.

Copied to clipboard

Mitodru Niyogi, Eric Gaussier, Arnab Bhattacharya

Linear-Time and Constant-Memory Text Embeddings Based on Recurrent Language Models (2026.acl-long)

Challenge:	Multilingual large language models are expensive to pretrain and suffer from imbalances across languages and datasets.
Approach:	They propose a family of Indian language-only autoregressive language models trained on open-source language-specific data for the five most spoken Indian languages.
Outcome:	The proposed model outperforms most larger models up to 8B across all five languages.

Copied to clipboard

Tobias Grantner, Emanuel Sallinger, Martin Flechl

Outcome Accuracy is Not Enough: Aligning the Reasoning Process of Reward Models (2026.acl-long)

Challenge:	Existing work on recurrent models for text embedding is limited to small task-specific models.
Approach:	They propose a vertically chunked inference strategy that enables fast embedding generation with memory usage that becomes constant in the input length once it exceeds the vertical chunk size.
Outcome:	The proposed architectures achieve competitive performance across benchmarks while maintaining a substantially smaller memory footprint compared to transformer-based models.

Copied to clipboard

Binghai Wang, Yantao Liu, Yuxuan Liu, Tianyi Tang, Shenzhi Wang, Chang Gao, Chujie Zheng, Yichang Zhang, Le Yu, Shixuan Liu, Tao Gui, Qi Zhang, Xuanjing Huang, Bowen Yu, Fei Huang, Junyang Lin

CIS-BWE: Chaos-Informed Speech Bandwidth Extension (2026.acl-long)

Challenge:	Recent studies observe a phenomenon where reward models achieve high accuracy on static datasets but fail to generalize effectively during RLHF.
Approach:	They propose a method that combines rationale consistency with outcome accuracy to improve performance on RM-Bench and JudgeBench.
Outcome:	The proposed method surpasses baselines on RM-Bench and JudgeBench by an average of 5% and improves creative writing tasks by 7%.

Copied to clipboard

Tarikul Islam Tamiti, Tonmoy Das, Nursadul Mamun, Anomadarshi Barua

The Pitfalls of KV Cache Compression (2026.acl-long)

Challenge:	CIS-BWE introduces two chaos-informed discriminators for capturing the deterministic chaos from speech.
Approach:	They propose a novel adversarial Bandwidth Extension framework that introduces two chaos-informed discriminators for capturing the deterministic chaos from speech.
Outcome:	The proposed framework achieves better performance across nine subjective and objective evaluation metrics with a 40x reduction in discriminator size and overall 0.5x fewer parameters, establishing a new baseline in the BWE task.

Copied to clipboard

Alex Chen, Renato Geh, Aditya Grover, Guy Van Den Broeck, Daniel Mingyi Israel

Survey Response Generation: Generating Closed-Ended Survey Responses In-Silico with Large Language Models (2026.acl-long)

Challenge:	Recent literature has shown minimal degradation of KV cache in multi-instruction prompts . authors show that certain instructions degrade much more rapidly with compression .
Approach:	They propose to change KV cache eviction policies to reduce the impact of KV evict bias . they propose to use a 'simple' evviction policy to reduce ejection bias if the LLM is a multi-instruction model .
Outcome:	The proposed methods show that certain instructions degrade much faster with compression, causing them to be ignored by the LLM.

Copied to clipboard

Georg Ahnert, Anna-Carolina Haensch, Barbara Plank, Markus Strohmaier

Sign-Language Datasets at Scale: A Comprehensive Survey on Resources, Benchmarks, and Annotation Standards (2026.acl-long)

Challenge:	Existing studies focus on generating closed-ended survey responses with large language models, whereas LLMs are typically trained to generate open-ended text.
Approach:	They evaluate the impact of various Survey Response Generation Methods on simulated responses by generating closed-ended responses from large language models.
Outcome:	The proposed methods perform best in individual-level and subpopulation-level alignment.

Copied to clipboard

Yiming Ni, Zhi-Qi Cheng, Jiayu Li, Wei Cheng

Challenge:	Existing benchmarks fail to reflect real-world communication needs and are limited in their coverage.
Approach:	They present a comprehensive index of sign-language datasets, covering 120 resources across 35 sign languages.
Outcome:	The proposed index covers 120 resources across 35 sign languages.

Anchoring the Affective Manifold: Learning Canonical and Disentangled Representations via Generative Cross-Modal Alignment (2026.acl-long)

Copied to clipboard

Weibin Li, Jintao Cheng, Xiaoyu Tang, Chi Man Vong

Who Plays Which Role When? Communication Role Dynamics for Peer Recognition and Team Performance Prediction (2026.acl-long)

Challenge:	Dominant multimodal emotion recognition paradigms neglect the intrinsic geometric structure of affect, resulting in representations heavily entangled with non-affective factors.
Approach:	They propose a Canonical Disentangled Multimodal Generative Framework that decomposes the latent space into a canonical Shared Affective Subspace and a private Modality Subspace.
Outcome:	The proposed model disentangles affect from private attributes while enabling controllable emotion generation.

Copied to clipboard

Yifan Song, Wenxuan Wendy Shi, Brian Bailey, Tal August

Challenge:	Prior work has modeled functional roles for meeting participants from simple speech features or used behavioral patterns to predict "latent" roles and team outcomes.
Approach:	They operationalize a taxonomy of eight communication roles grounded in education literature and annotate a corpus of 6,307 Slack messages from 55 students across 18 teams.
Outcome:	The proposed taxonomy outperforms lexical, conversational, and LLM-prompting baselines in predicting team performance after deliberation.

Incomplete In-context Learning (2026.acl-long)

Copied to clipboard

Wenqiang Wang, Wen Yujia, Yan Xiao, Zhifeng Chen, Yangshijie Zhang, Peng Chen, Mingbo Yang, Xiaochun Cao

Fine-tuning vs. In-context Learning in Large Language Models: A Formal Language Learning Perspective (2026.acl-long)

Challenge:	Existing in-context learning assumes the retrieval dataset contains demonstrations for all output label spaces.
Approach:	They propose a framework with train-free and train-based variants to address IICL . they propose to integrate a dataset with labeled demonstrations for each output space .
Outcome:	The proposed framework outperforms existing methods under incomplete retrieval datasets and even outperformed ICL with complete labels.

Copied to clipboard

Bishwamittra Ghosh, Soumi Das, Till Speicher, Qinyuan Wu, Mohammad Aflah Khan, Deepak Garg, Krishna P. Gummadi, Evimaria Terzi

Where Paths Split: Localized, Calibrated Control of Moral Reasoning in Large Language Models (2026.acl-long)

Challenge:	Prior studies comparing FT and ICL have yielded mixed and inconclusive results due to inconsistent experimental setups.
Approach:	They propose a formal language learning task with precise language boundaries, controlled string sampling, and no data contamination to enable a rigorous comparison.
Outcome:	The proposed task offers precise language boundaries, controlled string sampling, and no data contamination.

Copied to clipboard

Chenchen Yuan, Zheyu Zhang, Gjergji Kasneci

Social Story Frames: Contextual Reasoning about Narrative Intent and Reception (2026.acl-long)

Challenge:	Large language models display heterogeneous moral preferences across settings.
Approach:	They propose a method for steering toward a desired ethical framework while preserving general competence.
Outcome:	The proposed method outperforms baselines while providing interpretable mechanism.

Copied to clipboard

Joel Mire, Maria Antoniak, Steven R Wilson, Zexin Ma, Achyutarama R Ganti, Andrew Piper, Maarten Sap

Lightweight and Faithful Visual Condition Checking in Behavior Trees via Expert-Regularized Reinforcement Learning (2026.acl-long)

Challenge:	SocialStoryFrames is a formalism for distilling plausible inferences about reader response . authors characterize frequency and interdependence of storytelling intents across communities .
Approach:	They propose a formalism for distilling plausible inferences about reader response using conversational context and a taxonomy grounded in narrative theory, linguistic pragmatics, and psychology.
Outcome:	The proposed model can be used to analyze reader responses in online communities.

Copied to clipboard

Hyosik Moon, Eldan Cohen

When in Doubt, Consult: Expert Debate for Sexism Detection via Confidence-Based Routing (2026.acl-long)

Challenge:	Existing behavior trees are not suitable for high-dimensional perceptual inputs such as images or language.
Approach:	They propose a framework that leverages expert-regularized reinforcement learning to preserve semantic faithfulness while employing a factorized policy that aggregates sequential condition-node decisions into a single decision unit.
Outcome:	The proposed framework outperforms imitation learning and reinforcement learning but risks misalignment of condition nodes with intended semantics and poor credit assignment.

Copied to clipboard

Anwar Alajmi, Gabriele Pergola

Failure Modes in Multi-Hop QA: The Weakest Link Effect and the Recognition Bottleneck (2026.acl-long)

Challenge:	sexist content on social media is increasingly pervasive, often appearing in subtle, context-dependent forms that evade traditional classification methods.
Approach:	They propose a framework that unifies targeted training procedures to regularize supervision to scarce and noisy data with selective reasoning-based inference to handle ambiguous or borderline cases.
Outcome:	The proposed framework outperforms existing approaches across several public benchmarks . it bridges the gap between efficiency and reasoning with a dynamic routing mechanism .

Copied to clipboard

Meiru Zhang, Zaiqiao Meng, Nigel Collier

TagSpeech: End-to-End Multi-Speaker ASR and Diarization with Fine-Grained Temporal Grounding (2026.acl-long)

Challenge:	Existing studies have identified a position bias in Large Language Models that causes them to overlook information at certain positions.
Approach:	They propose a semantic probe to disentangle position bias in Large Language Models . they propose MFAI to steer attention towards selected positions .
Outcome:	The proposed model can locate and integrate information at certain positions even in noisy, long-context settings.

Copied to clipboard

Mingyue Huo, Yiwen Shao, Yuheng Zhang

On the Effect of Hyperparameters in Language Modeling for Computational Linguistics (2026.acl-long)

Challenge:	a critical ambiguity persists regarding what constitutes "joint ASR and diarization" a unified framework for multi-speaker ASR is proposed, but it is not yet clear what constitute "diarization."
Approach:	They propose a unified LLM-based framework that uses Temporal Anchor Grounding for joint multi-speaker ASR and diarization.
Outcome:	The proposed framework improves on AMI and AliMeeting benchmarks on speaker-content alignment . the proposed framework achieves consistent improvements in Diarization Error Rate over strong baselines .

Copied to clipboard

Ruoxi Ning, Yongpeng Zhu, Qingcheng Zeng, Tatsuki Kuribayashi, Freda Shi

Prosody as Supervision: Bridging the Non-Verbal–Verbal for Multilingual Speech Emotion Recognition (2026.acl-long)

Challenge:	Training language models and examining their linguistic behaviors is a common protocol in computational linguistics for studying linguistic phenomena and modeling human language processing.
Approach:	They replicate three prior studies with hyperparameters varied within a practical range and show that modest hyperparametric changes can alter qualitative conclusions about models’ linguistic abilities.
Outcome:	The results show that hyperparameter changes can alter qualitative conclusions and reverse the ranking of models.

Copied to clipboard

null Girish, Mohd Mujtaba Akhtar, Muskaan Singh

Mechanisms of Prompt-Induced Hallucination in Vision–Language Models (2026.acl-long)

Challenge:	Existing paradigms for low-resource multilingual speech emotion recognition rely on labeled verbal speech and lack cross-lingual transfer.
Approach:	They propose a paralinguistic supervision paradigm for low-resource multilingual speech emotion recognition that leverages non-verbal vocalizations to exploit prosody-centric emotion cues.
Outcome:	The proposed framework outperforms Euclidean counter parts and strong SSL baselines in the language-based evaluation of low-resource multilingual speech emotion recognition (LRM-SER)

Copied to clipboard

William Rudman, Michal Golovanevsky, Dana Arad, Yonatan Belinkov, Carsten Eickhoff, Ritambhara Singh, Kyle Mahowald

LLM4Cell: Taxonomy and Evaluation of LLM and Agentic Models for Single-Cell Biology (2026.acl-long)

Challenge:	Large vision–language models (VLMs) often hallucinate by favoring textual prompts over visual evidence.
Approach:	They study the failure mode of large vision–language models by focusing on textual prompts over visual evidence.
Outcome:	The proposed model overestimates the number of objects in an image . it hallucinates additional waterlilies when asked to describe a mismatched number of items . the model ablation reduces prompt-induced hallucinosities by at least 40% without additional training .

Copied to clipboard

Sajib Acharjee Dip, Adrika Zafor, Bikash Kumar Paul, Uddip Acharjee Shuvo, Muhit Islam Emon, Xuan Wang, Liqing Zhang

Wait! There’s a Way Out: A Decision Mechanism for Forecasting Conversational Derailment (2026.acl-long)

Challenge:	Large language models are transforming biomedical discovery by linking molecular patterns with knowledge encoded in text.
Approach:	They propose to map 58 foundation and agentic models developed for single-cell research into eight key analytical tasks.
Outcome:	The proposed models are applied to eight key analytical tasks including annotation, trajectory inference, perturbation modeling, and drug-response prediction.

Copied to clipboard

Laerdon Kim, Vivian Nguyen, Cristian Danescu-Niculescu-Mizil

From Nodes to Narratives: Explaining Graph Neural Networks with LLMs and Graph Context (2026.acl-long)

Challenge:	Existing approaches make decision to "trigger" based on the estimated likelihood of derailment given the preceding utterances, implicitly assuming that the conversation’s future trajectory is fixed.
Approach:	They propose a method for decoupling the decision to trigger from derailment likelihood estimation.
Outcome:	The proposed method is inspired by the first human baseline on this task, which shows that humans achieve dramatically lower false positive rates by selectively deferring their decision to trigger when they anticipate that tension is likely to subside.

Copied to clipboard

Peyman Baghershahi, Gregoire Fournier, Pranav Nyati, Sourav Medya

RealChart2Code: Bridging the Gap in Real-World Chart-to-Code Generation via Multi-Task Evaluation (2026.acl-long)

Challenge:	Existing explanation methods for graph neural networks struggle to generate interpretable, fine-grained rationales.
Approach:	They propose a lightweight framework that uses large language models to generate interpretable explanations for GNNs.
Outcome:	The proposed framework generates interpretable explanations for GNN predictions using large language models.

Copied to clipboard

Jiajun Zhang, Yuying Li, Zhixun Li, Xingyu Guo, Jingzhuo Wu, Leqi Zheng, Yiran Yang, Jianke Zhang, Qingbin Li, Shannan Yan, Changguo Jia, Junfei Wu, Zilei Wang, Qiang Liu, Liang Wang

Success and Cost Elicit Convention Formation for Efficient Communication (2026.acl-long)

Challenge:	Vision-Language Models (VLMs) have demonstrated impressive capabilities in code generation across various domains, but their ability to replicate complex, multi-panel visualizations remains largely unassessed.
Approach:	They propose a large-scale benchmark to evaluate chart generation from large- scale raw data and assess iterative code refinement in a multi-turn conversational setting.
Outcome:	The new benchmark evaluates 14 leading VLMs on real-world data and shows they struggle with complex plot structures and authentic data.

Copied to clipboard

Saujas Vaduguru, Yilun Hua, Yoav Artzi, Daniel Fried

Measuring User’s Mental Models of Speech Translation in Human-AI Collaboration (2026.acl-long)

Challenge:	ad hoc conventions allow people to coordinate on short, less costly utterances that are understood using shared conversational context.
Approach:	They propose a method to train large multimodal models to form conventions . they use simulated reference games to produce training data .
Outcome:	The proposed method reduces message length by up to 41% while increasing success by 15% over the course of the interaction.

Copied to clipboard

HyoJung Han, Nishant Balepur, Jordan Lee Boyd-Graber, Marine Carpuat

VeriTaS: The First Dynamic Benchmark for Multimodal Automated Fact-Checking (2026.acl-long)

Challenge:	Existing research on machine translation tools has not revealed how users perceive MT errors and how they evolve through interaction.
Approach:	They propose a framework where users accept MT output or request professional re-translation to answer questions based on information presented in a foreign language.
Outcome:	The proposed framework can predict where the system is likely to be wrong and how it evolves through interaction.

Copied to clipboard

Mark Rothermel, Marcus Kornmann, Marcus Rohrbach, Anna Rohrbach

Ethical NLP

Challenge:	Existing benchmarks for evaluating AFC systems are limited in terms of task scope, modalities, domain, language diversity, realism, or coverage of misinformation types.
Approach:	They propose to use Verified Theses and Statements (VeriTaS) to evaluate AFC systems that are static and subject to data leakage as claims enter pretraining corpora.
Outcome:	The proposed system is robust under large-scale pretraining of foundation models and can be updated in the future.

A Survey of Reasoning-Intensive Retrieval: Progress and Challenges (2026.acl-long)

Copied to clipboard

Yiyang Wei, Tingyu Song, Siyue Zhang, Yilun Zhao

Which Reasoning Trajectories Teach Students to Reason Better? A Simple Metric of Informative Alignment (2026.acl-long)

Challenge:	Reasoning-Intensive Retrieval (RIR) targets retrieval settings where relevance is mediated by latent inferential links between a query and supporting evidence, rather than semantic similarity.
Approach:	They propose a taxonomy that categorizes methods based on where and how reasoning is integrated into the retrieval pipeline.
Outcome:	The proposed method framework provides a detailed analysis of the current landscape and its trade-offs and practical applications.

Copied to clipboard

Yuming Yang, Mingyoung Lai, Wanxu Zhao, Xiaoran Fan, Zhiheng Xi, Mingqi Wu, Chiyue Huang, Jun Zhao, Haijun Lv, Jian Tong, Yunhua Zhou, Yicheng Zou, Qipeng Guo, Tao Gui, Qi Zhang, Xuanjing Huang

From Proof to Program: Characterizing Tool-Induced Reasoning Hallucinations in Large Language Models (2026.acl-long)

Challenge:	Existing methods assess suitability primarily through student likelihood, favoring trajectories that align closely with the student model’s current behavior but overlooking more informative ones.
Approach:	They propose a Rank–Surprisal Ratio metric that captures both alignment and informativeness to assess the suitability of a reasoning trajectory.
Outcome:	The proposed metric captures both alignment and informativeness to assess the suitability of a reasoning trajectory.

Copied to clipboard

Farima Fatahi Bayat, Pouya Pezeshkpour, Estevam Hruschka

Paper Circle: An Open-source Multi-agent Research Discovery and Analysis Framework (2026.acl-long)

Challenge:	Tool-augmented Language Models can invoke external tools to solve problems beyond their parametric capacity.
Approach:	They propose a preference-optimization-based framework that realigns TaLMs to use tool outputs as assistive evidence.
Outcome:	The proposed framework improves accuracy and reasoning depth under tool use.

Copied to clipboard

Komal Kumar, Aman Chadha, Salman Khan, Fahad Shahbaz Khan, Hisham Cholakkal

PEAR: Pairwise Evaluation for Automatic Relative Scoring in Machine Translation (2026.acl-long)

Challenge:	Recent advances in large language models have demonstrated strong potential for understanding user intent . paper describes system architecture, agent roles, retrieval and scoring methods, knowledge graph schema, and evaluation interfaces .
Approach:	They propose a multi-agent research discovery and analysis system that integrates multiple agents to reduce the effort required to find, assess, organize, and understand academic literature.
Outcome:	The proposed system reduces the effort required to find, assess, organize, and understand academic literature.

Copied to clipboard

Lorenzo Proietti, Roman Grundkiewicz, Matt Post

AV-Dialog: Spoken Dialogue Models with Audio-Visual Input (2026.acl-long)

Challenge:	PEAR is a supervised quality estimation metric that reframes reference-free machine translation evaluation as a graded pairwise comparison.
Approach:	They propose to use a supervised quality estimation metric family to reframe machine translation evaluation as a graded pairwise comparison.
Outcome:	The proposed metric outperforms strictly matched single-candidate QE baselines on the WMT24 meta-evaluation benchmark.

Copied to clipboard

Tuochao Chen, Bandhav Veluri, Hongyu Gong, Shyamnath Gollakota

A BERTology View of LLM Orchestrations: Token- and Layer-Selective Probes for Efficient Single-Pass Classification (2026.acl-long)

Challenge:	AV-Dialog uses audio and visual cues to track the target speaker, predict turn-taking, and generate coherent responses.
Approach:	They propose a multimodal dialog framework that uses both audio and visual cues to track the target speaker.
Outcome:	AV-Dialog outperforms audio-only models under interference, reducing transcription errors, improving turn-taking prediction and human-rated dialogue quality.

Copied to clipboard

Gonzalo Ariel Meyoyan, Luciano Del Corro

Can Large Language Models Keep Up? Benchmarking Online Adaptation to Continual Knowledge Streams (2026.acl-long)

Challenge:	Modern LLM deployments are rarely a single model in isolation.
Approach:	They propose a model that reuses computation already paid for by the serving LLM . they instantiate a template with pooling, a scoring-attention gate, and a downcast multi-head self-attention probe .
Outcome:	The proposed model improves safety and sentiment benchmarks on dense and mixture-of-experts architectures while preserving near-serving latency.

Copied to clipboard

Jiyeon Kim, Hyunji Lee, Dylan Zhou, Sue Hyun Park, Seunghyun Yoon, Trung Bui, Franck Dernoncourt, Sungmin Cha, Minjoon Seo

Evaluating Language Model Pluralism through In-the-wild Crowd Discussions (2026.acl-long)

Challenge:	Existing models and agentic memory systems fail to adapt robustly to OAKS, demonstrating delays in state-tracking and susceptibility to distraction within streaming environments.
Approach:	They propose a benchmark to evaluate models' ability to adapt to changing knowledge over streaming . they use two datasets to analyze how facts evolve over time .
Outcome:	The proposed benchmark evaluates models in an online adaptation setting over streaming, continually updating knowledge.

Copied to clipboard

Gagan Mundada, Rohan Surana, Nandhini Swaminathan, Bodhisattwa Prasad Majumder, Junda Wu, Julian McAuley, Zhouhang Xie

PRBench: Large-Scale Expert Rubrics for Evaluating High-Stakes Professional Reasoning (2026.acl-long)

Challenge:	Existing evaluation methods focus predominantly on multiple-choice and question-answering tasks, leaving open-ended generation largely unaddressed.
Approach:	They propose an evaluation framework that assesses LLM pluralism in open-ended generation by comparing outputs against free-form crowd responses.
Outcome:	The proposed evaluation framework decomposes ground-truth responses into atomic, non-overlapping claims and evaluates whether LLMs adequately cover this diverse claim space.

Copied to clipboard

Afra Feyza Akyürek, Advait Gosai, Chen Bo Calvin Zhang, Vipul Gupta, Jaehwan Jeong, Anisha Gunjal, Tahseen Rabbani, Maria Mazzone, David Randolph IV, Mohammad Mahmoudi Meymand, Gurshaan Chattha, Paula Rodriguez, Diego A. Mares Buendia, Pavit Singh, Michael Liu, Subodh Chawla, Peter Cline, Lucy Ogaz, Ernesto Gabriel Hernández Montoya, Zihao Wang, Pavi Bhatter, Marcos Ayestaran, Bing Liu, Yunzhong He

Constructing Interpretable Features from Compositional Neuron Groups (2026.acl-long)

Challenge:	Frontier models often lack a view of performance on open-ended, economically consequential tasks in high-stakes professional domains where practical returns matter most.
Approach:	They introduce a professional reasoning benchmark that recruits 182 qualified professionals to contribute questions inspired by their workflows.
Outcome:	The proposed model outperforms other models in 114 countries and 47 US jurisdictions on hard subsets.

Copied to clipboard

Or David Shafran, Atticus Geiger, Mor Geva

Simulated Students in Tutoring Dialogues: Substance or Illusion? (2026.acl-long)

Challenge:	Existing methods for analyzing LLMs rely on dictionary learning with sparse autoencoders (SAEs) however, SAEs struggle in causal evaluations and lack intrinsic interpretability, as their learning is not explicitly tied to the computations of the model.
Approach:	They propose to decompose MLP activations with semi-nonnegative matrix factorization (SNMF) such that the learned features are mapped to their activating inputs, making them directly interpretable.
Outcome:	Experiments on Llama 3.1, Gemma 2 and GPT-2 show that SNMF derived features outperform SAEs and a strong supervised baseline on causal steering while aligning with human-interpretable concepts.

Copied to clipboard

Alexander Scarlatos, Jaewook Lee, Simon Woodhead, Andrew Lan

A Survey of Multimodal Mathematical Reasoning: From Perception, Alignment to Reasoning (2026.acl-long)

Challenge:	evaluating the effectiveness of new technology requires real students, which is time-consuming and hard to scale up.
Approach:	They propose to define the student simulation task and benchmark a wide range of student simulation methods on these metrics.
Outcome:	The proposed evaluation metrics show that prompting strategies perform poorly on a real-world tutoring dialogue dataset.

Copied to clipboard

Tianyu Yang, Sihong Wu, Yilun Zhao, Zhenwen Liang, Lisen Dai, Chen Zhao, Minhao Cheng, Arman Cohan, Xiangliang Zhang

Empowering Multi-Turn Tool-Integrated Agentic Reasoning with Group Turn Policy Optimization (2026.acl-long)

Challenge:	Multimodal mathematical Reasoning (MMR) has attracted increasing attention for its ability to solve mathematical problems involving both textual and visual modalities.
Approach:	They review the theoretical frameworks of multimodal reasoning and examine the challenges they face in visual math tasks.
Outcome:	The proposed models can solve problems involving both textual and visual modalities.

Copied to clipboard

Yifeng Ding, Hung Le, Songyang Han, Kangrui Ruan, Zhenghui Jin, Varun Kumar, Zijian Wang, Anoop Deoras

HypoEval: Hypothesis-Guided Evaluation for Natural Language Generation (2026.acl-long)

Challenge:	Current reinforcement learning methods suffer from coarse-grained, trajectory-level rewards that provide insufficient learning signals for complex multi-turn interactions, leading to training stagnation.
Approach:	They propose a novel RL algorithm for training large language models for multi-turn tool-integrated reasoning (TIR) that incorporates three innovations: turn-level reward assignment that provides fine-grained feedback for individual turns, return-based advantage estimation where normalized discounted returns are calculated as advantages, and self-supervised reward shaping that exploits self-supervision signals from generated code to densify sparse binary outcome-based rewards.
Outcome:	The proposed algorithm outperforms GRPO by 3.0% across diverse math reasoning benchmarks and improves grepo by 3.9% on commonsense reasoning and program synthesis tasks.

Copied to clipboard

Mingxuan Li, Hanchen Li, Chenhao Tan

LLMSurgeon: Diagnosing Data Mixture of Large Language Models (2026.acl-long)

Challenge:	Existing frameworks for LLM-as-a-judge use zero-shot setting without consulting any human input, which leads to low alignment, or fine-tune LLMs on labeled data, which requires a non-trivial number of samples.
Approach:	They propose a hypothesis-guided evaluation framework that uses a small corpus of human evaluations to generate more detailed rubrics for human judgments and incorporates a checklist-like approach to combine LLM’s assigned scores on each decomposed dimension to acquire overall scores.
Outcome:	The proposed framework outperforms existing frameworks in both human rankings and human scores with 30 human evaluations and fine-tunes LLMs on labeled data with 3 times more human evaluation by 11.95%.

Copied to clipboard

Yaxin Luo, Jiacheng Cui, Xiaohan Zhao, Xinyi Shang, Jiacheng Liu, Xinyue Bi, Zhaoyi Li, Zhiqiang Shen

The African Languages Lab: A Collaborative Approach to Advancing Low-Resource African NLP (2026.acl-long)

Challenge:	a lack of transparency in large language models makes auditing their "digital DNA" difficult.
Approach:	They propose a framework that casts DMS as an inverse problem under label-shift assumption . they propose LLMScan, a recipe-verifiable evaluation suite built from open-source LLMs .
Outcome:	The proposed framework casts DMS as an inverse problem under label-shift assumption . compared with existing frameworks, it recovers domain mixtures with high fidelity .

Copied to clipboard

Sheriff Issaka, Keyi Wang, Yinka Ajibola, Oluwatumininu Samuel-Ipaye, Zhaoyi Zhang, Nicte Aguillon Jimenez, Evans Kofi Agyei, Abraham Lin, Rohan Ramachandran, Sadick Abdul Mumin, Faith Nchifor, Mohammed Shuraim Issah, Erick Rosas Gonzalez, Lieqi Liu, Sylvester Kpei, Jemimah Kusi Osei, Carlene Ajeneza, Persis Boateng, Prisca Adwoa Dufie Yeboah, Saadia Gabriel

LogosKG: Hardware-Optimized Scalable and Interpretable Knowledge Graph Retrieval (2026.acl-long)

Challenge:	Among the approximately 7,000 languages spoken globally, fewer than 20 receive substantial attention in NLP research.
Approach:	They propose to use African multi-modal speech and text data to validate African multimodal models and validate them on targeted language data.
Outcome:	The African Languages Lab's results show that the proposed model outperforms untrained models in 31 languages and a 1B-parameter model beats the commercial system in Yoruba and Twi.

Copied to clipboard

He Cheng, Yifu Wu, Saksham Khatwani, Maya Kruse, Dmitriy Dligach, Timothy A. Miller, Majid Afshar, Yanjun Gao

GCA Framework: A GCC Countries–Grounded Dataset and Agentic Pipeline for Climate Decision Support (2026.acl-long)

Challenge:	Existing systems struggle to balance efficiency, scalability, and interpretability.
Approach:	They propose a hardware-aligned framework that enables scalable and interpretable k-hop retrieval on large KGs.
Outcome:	The proposed framework scales to billion-edge graphs without loss of retrieval fidelity.

Copied to clipboard

Muhammad Umer Sheikh, Khawar Shehzad, Salman Khan, Fahad Shahbaz Khan, Muhammad Haris Khan

Challenge:	Climate decision support systems are weak in region-specific climate knowledge and interaction with geospatial and forecasting tools.
Approach:	They propose a framework that unifies a curated multimodal dataset and a tool-augmented agent for climate analysis.
Outcome:	The proposed framework improves reliability over general-purpose models on climate tasks in the Gulf region.

Effects of Collaboration on the Performance of Interactive Theme Discovery Systems (2026.acl-long)

Copied to clipboard

Alvin Po-Chun Chen, Rohan Das, Dananjay Srinivas, Alexandra Barry, Maksim Seniw, Maria Leonor Pacheco

Challenge:	NLP-assisted systems to support qualitative data analysis have gained considerable traction, but no unified evaluation framework exists to account for the many different settings in which qualitative researchers may employ them.
Approach:	They propose a framework to evaluate the way collaboration settings may produce different research outcomes across a variety of interactive systems.
Outcome:	The proposed framework evaluates the impact of synchronous vs. asynchronous collaboration on consistency, cohesiveness, and correctness of qualitative research outcomes.

GraphSynth: Resolving the Diversity-Reliability Trade-off with Probabilistic Factor Graphs (2026.acl-long)

Copied to clipboard

Zehua Cheng, Wei Dai, Jiahao Sun, Thomas Lukasiewicz

A Structured Clustering Approach for Inducing Media Narratives (2026.acl-long)

Challenge:	Large language models are a scaleable solution for the generation of synthetic data . however, the utility of such data is capped by a critical tension between diversity and factual reliability.
Approach:	They propose a framework which leverages a probabilistic factor graph modeling the universe of attributes.
Outcome:	The proposed framework outperforms state-of-the-art models with a high structural integrity and a boost in performance on downstream tasks.

Copied to clipboard

Rohan Das, Advait Deshmukh, Alexandria Leto, Zohar Naaman, I-Ta Lee, Maria Leonor Pacheco

ACBQ: Adaptive Cross-Block Quantization of Large Language Models (2026.acl-long)

Challenge:	Existing approaches to modeling media narratives miss subtle narrative patterns through coarse-grained analysis or require domain-specific taxonomies that limit scalability.
Approach:	They propose a framework for inducing rich narrative schemas by jointly modeling events and characters via structured clustering.
Outcome:	The proposed framework produces explainable narrative schemas that align with established framing theory while scaling to large corpora without exhaustive manual annotation.

Copied to clipboard

Hailing Wang, Jianglin Lu, Yitian Zhang, Huimin Zeng, Yun Fu

SecureGate: Learning When to Reveal PII Safely via Token-Gated Dual-Adapters for Federated LLMs (2026.acl-long)

Challenge:	Existing methods for post-training quantization struggle to support weight–activation joint quantization and extreme low-bit weight quantization.
Approach:	They propose a framework that addresses weight–activation joint quantization and extreme weight quantization.
Outcome:	The proposed framework achieves superior performance under both W4A4 and highly aggressive W2 settings while incurring negligible additional computational overhead.

Copied to clipboard

Mohamed Shaaban, Mohamed Elmahallawy

EDUMATH: Generating Standards-aligned Educational Math Word Problems (2026.acl-long)

Challenge:	Existing privacy defenses reduce leakage of PII due to LLM memorization, but often degrade downstream performance.
Approach:	They propose a privacy-aware federated fine-tuning framework for large language models that provides fine-grained privacy control without sacrificing utility.
Outcome:	The proposed framework reduces PII leakage while providing fine-grained privacy control without sacrificing utility.

Copied to clipboard

Bryan R Christ, Penelope Molitz, Beau LeBlond, Zachary Gottesman, Jonathan Kropko, Thomas Hartvigsen

AdaFuse: Adaptive Ensemble Decoding for Large Language Models (2026.acl-long)

Challenge:	Math word problems (MWPs) are critical elements of K-12 math education and can be customized to students' interests and ability levels.
Approach:	They propose that LLMs can generate MWPs customized to student interests and math education standards by using an open and closed LLM to evaluate over 11,000 MWps and develop a teacher-annotated dataset for standards-aligned educational MWPS generation.
Outcome:	The proposed model outperforms existing closed models without training and is more similar to human-written MWPs but prefers customized MWPS with grade school students.

Copied to clipboard

Chengming Cui, Tianxin Wei, Ziyi Chen, Ruizhong Qiu, Zhichen Zeng, Zhining Liu, Xuying Ning, Duo Zhou, Jingrui He

Uncertainty-Aware Test-Time Search for Optimization Problem Solving (2026.acl-long)

Challenge:	Existing ensemble approaches to large language models lack flexibility for mid-generation adaptation.
Approach:	They propose an adaptive ensemble decoding framework that dynamically selects semantically appropriate fusion units during generation.
Outcome:	The proposed framework outperforms existing ensemble frameworks on open-domain QA, arithmetic reasoning, and machine translation tasks.

Copied to clipboard

Linlin Yu, Xujiang Zhao, Dong Li, Yanchi Liu, Wei Cheng, Zhengzhang Chen, Chen Zhao, Feng Chen, Haifeng Chen

TLPO: Token-Level Policy Optimization for Mitigating Language Confusion in Large Language Models (2026.acl-long)

Challenge:	Language model hallucinations and limited availability of labeled datasets often result in misaligned formulations, code errors and feasibility failures.
Approach:	They propose a Monte Carlo Tree Search framework that automates optimization problems from natural language descriptions with efficiency and reliability.
Outcome:	The proposed framework achieves state-of-the-art solution accuracy and reduces token usage.

Copied to clipboard

Jinho Choo, JunSeung Lee, Jimyeong Kim, Yeeho Song, S. K. Hong, Yeong-Dae Kwon

JW-SVD: Bridging the Cross-Modal Mismatch in Post-Training MLLM Compression (2026.acl-long)

Challenge:	Prior mitigation approaches that optimize entire responses operate at the level of entire responses and can lead to unintended degradation of general model capabilities.
Approach:	They propose a fine-tuning framework to mitigate erroneous outputs by localizing and updating the policy at a granular level.
Outcome:	The proposed framework outperforms baselines on multiple multilingual LLMs across diverse languages while preserving task accuracy.

Copied to clipboard

Runchao Li, Yao Fu, Mu Sheng, Haotian Yu, Xianxuan Long, Kenneth A. Loparo

Advancing Reasoning in Diffusion Language Models with Denoising Process Rewards (2026.acl-long)

Challenge:	Existing methods for compression of Multimodal Large Language Models lack multimodal adaptation to preserve cross-modal synergy.
Approach:	They propose a framework that aligns vision and language manifolds via a Joint Covariance basis and propose Global Spectrum-Aware Truncation to dynamically transfer parameter budget to the sensitive Backbone.
Outcome:	Experiments on Qwen2.5-VL and Llama-3-Next confirm that JW-SVD retains both text and image capabilities.

Copied to clipboard

Shaoan Xie, Lingjing Kong, Xiangchen Song, Xinshuai Dong, Guangyi Chen, Eric P. Xing, Kun Zhang

Estimating the Black-box LLM Uncertainty with Distribution-Aligned Adversarial Distillation (2026.acl-long)

Challenge:	Existing methods for improving reasoning in diffusion language models rely on outcome-based rewards that provide no direct supervision over the denoising process.
Approach:	They propose a method that provides a process-level reinforcement signal over denoising trajectory of diffusion language models.
Outcome:	Experiments on challenging reasoning benchmarks show that the proposed model improves reasoning stability, interpretability and overall performance.

Copied to clipboard

Huizi Cui, Huan Ma, Qilin Wang, Yuhang Gao, Changqing Zhang

CoDial: Interpretable Task-Oriented Dialogue Systems Through Dialogue Flow Alignment (2026.acl-long)

Challenge:	Existing uncertainty quantification methods depend on computationally expensive multiple sampling or internal parameters, which prevents real-time estimation and fails to capture information implicit in the black-box reasoning process.
Approach:	They propose a distribution-aligned adjudication architecture to guide a lightweight proxy model to learn the high-quality regions of the output distribution of the black-box LLM.
Outcome:	Extensive experiments show that a proxy model even with 1% of the target LLM’s size can achieve reliable uncertainty quantification.

Copied to clipboard

Radin Shayanfar, Chu Fei Luo, Rohan V Bhambhoria, Samuel Dahan, Xiaodan Zhu

BioProAgent: Neuro-Symbolic Grounding for Constrained Scientific Planning (2026.acl-long)

Challenge:	Recent schema-based TOD frameworks improve generalization by decoupling task logic from language understanding, but their reliance on neural or generative models obscures how task schemas influence behaviour and hence impair interpretability.
Approach:	They propose a framework that converts a predefined task schema to a structured heterogeneous graph and then to popular programmatic LLM guardrailing code, such as NVIDIA’s Colang.
Outcome:	The proposed framework achieves state-of-the-art performance on the widely used benchmark datasets while providing inherent interpretability in the design.

Copied to clipboard

Yuyang Liu, Jingya Wang, Liuzhenghao Lv, Yonghong Tian

MDP-GRPO: Stabilized Group Relative Policy Optimization for Multi-Constraint Instruction Following (2026.acl-long)

Challenge:	Large language models (LLMs) have demonstrated significant reasoning capabilities in scientific discovery but struggle to bridge the gap to physical execution in wet-labs.
Approach:	They propose a framework that anchors probabilistic planning in a deterministic Finite State Machine (FSM) they propose reducing token consumption by 6 through symbolic abstraction to address context bottlenecks .
Outcome:	The proposed framework achieves 95.6% physical compliance, compared to 21.0% for ReAct, in the extended BioProBench benchmark.

Copied to clipboard

Mohammad Mahdi Salmani-Zarchi, Zahra Rahimi, Heshaam Faili, Mohammad Javad Dousti

Challenge:	Large language models (LLMs) can follow many natural-language instructions, yet they remain brittle when a request bundles multiple explicit constraints, such as asking the LLM to respond in a particular structure with an exact ending phrase.
Approach:	They propose a method which stabilizes learning through multi-temperature sampling to increase reward dispersion, dual-anchor advantages to restore gradients in homogeneous groups, prospect-theoretic shaping to bound updates and penalize violations based on Kahneman Tversky’s theory and asymmetric KL regularization.
Outcome:	The proposed method outperforms standard GRPO on FollowBench, IFEval, and a curated multi-constraint dataset, improving strict constraint satisfaction by up to 5.0% on Llama-3.2-3B.

TokenTiming: A Dynamic Alignment Method for Universal Speculative Decoding Model Pairs (2026.acl-long)

Copied to clipboard

Sibo Xiao, Fu Jinyuan, Zhongle Xie, Lidan Shou

Challenge:	Speculative decoding (SD) is a useful tool for accelerating large language models . but its utility is limited by a fundamental constraint: draft and target models must share the same vocabulary .
Approach:	They propose an algorithm that uses a draft token sequence to get a new target token sequence and then uses DTW to build a mapping to transfer probability distributions.
Outcome:	The proposed method shows 1.57x speedup on various tasks.

GUIDE: Towards Scalable Advising for Research Ideas (2026.acl-long)

Copied to clipboard

Yaowenqi Liu, BingXu Meng, Rui Pan, Yuxing Liu, Jerry Huang, Jiaxuan You, Tong Zhang

ARES: Adaptive Red-Teaming and End-to-End Repair of Policy-Reward System (2026.acl-long)

Challenge:	Existing systems that provide detailed, constructive feedback on academic papers struggle with review fidelity.
Approach:	They explore factors that underlie the development of robust advising systems . large language models have shown remarkable progress in tasks from text generation to code synthesis .
Outcome:	The proposed model outperforms general-purpose language models in acceptance rates for self-ranked top-30% submissions to ICLR 2025.

Copied to clipboard

Jiacheng Liang, Yao Ma, Tharindu Kumarage, Satyapriya Krishna, Rahul Gupta, Kai-Wei Chang, Aram Galstyan, Charith Peris

Reasoning Traces Shape Outputs but Models Won’t Say So (2026.acl-long)

Challenge:	Existing red-teaming approaches focus on policy-level weaknesses, but they overlook systemic weaknesses . aRES exploits dual-targeting weaknesses in both the core LLM and the RM simultaneously.
Approach:	a new framework uncovers weaknesses in both the core and the reward models simultaneously . a "Safety Mentor" generates semantically coherent adversarial prompts .
Outcome:	ARES uncovers weaknesses in both the core LLM and the RM simultaneously . it fine-tunes the LM to detect harmful content, then optimizes the core model .

Copied to clipboard

Yijie Hao, Lingjie Chen, Ali Emami, Joyce C. Ho

Value of Information: A Framework for Human–Agent Communication (2026.acl-long)

Challenge:	Large reasoning models (LRMs) generate explicit reasoning traces before producing answers, offering a window into their decisionmaking.
Approach:	They propose a method that injects synthetic reasoning snippets into a model’s reasoning trace and measures whether the model follows the injected reasoning and acknowledges doing so.
Outcome:	The proposed method reveals that models refuse to disclose their influence when asked to explain their changed answers.

Copied to clipboard

Yijiang River Dong, Tiancheng Hu, Zheng Hui, Caiqi Zhang, Ivan Vulić, Andreea Bobu, Nigel Collier

Challenge:	Existing approaches to large language model (LLM) agents fail to account for stakes of different decisions.
Approach:	They propose a framework that balances task risk, query ambiguity, user effort . they use a value-of-information framework to dynamically weigh the expected utility gain .
Outcome:	The proposed model matches or exceeds the best manually-tuned baselines in four domains . it explicitly balances task risk, query ambiguity, and user effort .

AutoRAN: Automated Hijacking of Safety Reasoning in Large Reasoning Models (2026.acl-long)

Copied to clipboard

Jiacheng Liang, Tanqiu Jiang, Yuhui Wang, Rongyi Zhu, Fenglong Ma, Ting Wang

Thinking Alignment of Scenario-Oriented User Simulation (2026.acl-long)

Challenge:	Large reasoning models (LRMs) generate coherent reasoning paths before conclusions, but they introduce new vulnerabilities.
Approach:	They propose a framework that leverages a weaker but less-aligned model to simulate execution reasoning for initial hijacking attempts and iteratively refines attacks by exploiting reasoning patterns leaked through the target LRM’s refusals.
Outcome:	The proposed framework achieves 100% success rate within one or few turns, neutralizing reasoning-based defenses even when evaluated by robustly aligned external models.

Copied to clipboard

Xiaoting Wu, Yi Huang, Chunyang Gao, Mengfei Guo, Jingyu Yao, Junlan Feng

AdapShot: Adaptive Many-Shot In-Context Learning with Semantic-Aware KV Cache Reuse (2026.acl-long)

Challenge:	Existing user simulators based on prompting to role-play or SFT focus on imitating textual utterances without considering multi-faceted cognitive processes that underlie human decision-making during interactions.
Approach:	They construct a user-simulator dataset that augments 51k human–LLM conversations by reconstructing the user’s inner reasoning during and at the end of each dialogue.
Outcome:	The proposed user simulators augment 51k human–LLM conversations by reconstructing the user’s inner reasoning both during and at the end of each dialogue.

Copied to clipboard

Jie Ou, Jinyu Guo, Shiyao Guo, Yuang Li, Ruiqi Wu, Zhaokun Wang, Wenyi Li, Wenhong Tian

Credal Concept Bottleneck Models for Epistemic–Aleatoric Uncertainty Decomposition (2026.acl-long)

Challenge:	Existing methods for In-Context Learning (ICL) rely on a predetermined number of shots, leading to insufficient context or noise.
Approach:	They propose a probe-based evaluation mechanism that utilizes output entropy to determine the optimal number of shots and leverages KV cache reuse for efficient inference.
Outcome:	The proposed model achieves an average performance gain of 10% and a 4.64 speedup compared to state-of-the-art DBSA.

Copied to clipboard

Tanmoy Mukherjee, Thomas Bailleux, Pierre Marquis, Zied Bouraoui

How do Role Models Shape Collective Morality? Exemplar-Driven Moral Learning in Multi-Agent Simulation (2026.acl-long)

Challenge:	Existing models face challenges when dealing with uncertainty.
Approach:	They propose a framework that decomposes concept uncertainty by construction . epistemic uncertainty is positively associated with prediction errors, whereas aleatoric uncertainty closely tracks disagreement .
Outcome:	The proposed framework decomposes concept uncertainty by construction . epistemic uncertainty is positively associated with prediction errors, whereas aleatoric uncertainty closely tracks disagreement .

Copied to clipboard

Junjie Liao, Huacong Tang, Zhou Ziheng, Yizhou Wang, Fangwei Zhong

ReActR: Reasoning through Error-Activated Reflection for LLM Post-Training (2026.acl-long)

Challenge:	Existing studies show that role models influence morality, but they are not uniformly interpreted and appropriated in groups with heterogeneous motivations.
Approach:	They build a multi-agent simulation where agents with diverse intrinsic drives interact and adapt through a four-stage cognitive loop.
Outcome:	The proposed model can significantly reshape morality of agents with diverse intrinsic drives . the simulations show that identity-driven conformity can substantially reshaped initial dispositions .

Copied to clipboard

Lina Sun

RL-PLUS: Countering Capability Boundary Collapse of LLMs in Reinforcement Learning with Hybrid-policy Optimization (2026.acl-long)

Challenge:	Existing methods for improving the mathematical abilities of Large Language Models (LLMs) focus disproportionately on scaling correct training samples, overlooking the rich learning signals contained in erroneous reasoning trajectories.
Approach:	They propose a framework that enhances reasoning by learning reflective behaviors from erroneous trajectories by using data construction and training.
Outcome:	Extensive experiments on three LLMs show that ReActR improves reasoning performance on Llama-3-8B.

Copied to clipboard

Yihong Dong, Xue Jiang, Yongding Tao, Huanyu Liu, Kechi Zhang, Lili Mou, Rongyu Cao, Yingwei MA, Jue Chen, Binhua Li, Zhi Jin, Fei Huang, Yongbin Li, Ge Li

Explain the Synth: Interpretable Evaluation of LLM Data Synthesis (2026.acl-long)

Challenge:	Reinforcement Learning with Verifiable Reward (RLVR) has significantly advanced the complex reasoning abilities of Large Language Models (LLMs).
Approach:	They propose a hybrid-policy optimization approach that synergizes internal exploitation with external data to achieve stronger reasoning capabilities.
Outcome:	The proposed approach achieves state-of-the-art performance on six math reasoning benchmarks and superior performance on out-of distribution reasoning tasks.

Copied to clipboard

Yue Yang, Fan Yang, Yu Bai, Hao Wang

Beyond the Leaderboard: Rethinking Medical Benchmarks for Large Language Models (2026.acl-long)

Challenge:	Large language models (LLMs) are increasingly used to generate tabular data.
Approach:	They propose a framework that uses a rule-based model as a shared explanatory language to examine the explanation of real versus synthetic data.
Outcome:	The proposed framework compares the explanatory structure induced by real versus synthetic data.

Copied to clipboard

Wenxuan Wang, Zizhan Ma, Guo Yu, Yiu-Fai Cheung, Meidan Ding, Jie Liu, Wenting Chen, Linlin Shen

Speech-Hands: A Self-Reflection Voice Agentic Approach to Speech Recognition and Audio Reasoning with Omni Perception (2026.acl-long)

Challenge:	Large language models (LLMs) are proving significant potential in healthcare, prompting numerous benchmarks to evaluate their capabilities.
Approach:	They propose a framework that deconstructs benchmark development into five stages from design to governance and provides a checklist of 46 medically-tailored criteria.
Outcome:	The framework deconstructs benchmark development into five stages from design to governance and provides a comprehensive checklist of 46 medically-tailored criteria.

Copied to clipboard

Zhen Wan, Chao-Han Huck Yang, Jinchuan Tian, Hanrong Ye, Ankita Pasad, Szu-Wei Fu, Arushi Goel, Ryo Hachiuma, Shizhe Diao, Kunal Dhawan, Sreyan Ghosh, Yusuke Hirota, Zhehuai Chen, Rafael Valle, Chenhui Chu, Shinji Watanabe, Boris Ginsburg, Yu-Chiang Frank Wang

When Personalization Tricks Detectors: The Feature-Inversion Trap in Machine-Generated Text Detection (2026.acl-long)

Challenge:	naively fine-tuning an omni-model on speech recognition and external sound understanding tasks often degrades performance . Xie and Wu's framework, Speech-Hands, recasts the problem as an explicit self-reflection decision.
Approach:	They propose a voice-agentic framework that learns one critical omni-understanding skill: trusting itself versus external audio perception.
Outcome:	The proposed framework outperforms baseline models on the OpenASR leaderboard by 12.1% WER and high F1 on audio QA decisions.

Copied to clipboard

Lang Gao, Xuhui Li, Chenxi Wang, Mingzhe Li, Wei Liu, Zirui Song, Jinghui Zhang, Rui Yan, Preslav Nakov, Xiuying Chen

HeteroCache: A Dynamic Retrieval Approach to Heterogeneous KV Cache Compression for Long-Context LLM Inference (2026.acl-long)

Challenge:	Personalized MGT detection remains largely underexplored due to personalization challenges . large language models (LLMs) can imitate personal writing styles, but they can generate fake news and misinformation.
Approach:	They propose a benchmark to evaluate detector robustness under personalization . they attribute this limitation to a feature-inversion trap that flips the effect in personalized contexts .
Outcome:	The proposed framework predicts detector robustness under personalization with an 85% correlation to actual results.

Copied to clipboard

Zhiyuan Shi, Qibo Qiu, null Xuefeng, Zhonglin Jiang, Li Yu, Jian Jiang, Xiaofei He, Wenxiao Wang

Sparse-RL: Breaking the Memory Wall in LLM Reinforcement Learning via Stable Sparse Rollouts (2026.acl-long)

Challenge:	Existing static compression methods suffer from coarse-grained caching and high I/O overhead.
Approach:	They propose a training-free dynamic compression framework that uses a sparse attention mechanism to categorize attention heads based on stability and similarity.
Outcome:	The proposed framework achieves state-of-the-art performance on long-context benchmarks and accelerates decoding by up to 3 compared to the original model with a 224K context.

Copied to clipboard

Sijia Luo, Xiaokang Zhang, Yuxuan Hu, Bohan Zhang, Ke Wang, Jinbo Su, Mengshu Sun, Lei Liang, Jing Zhang

Challenge:	Existing methods for storing key-value caches during long-horizon rollouts cause performance collapses.
Approach:	They propose a new training paradigm that empowers stable RL training under sparse rollouts.
Outcome:	The proposed model reduces rollout overhead while maintaining the performance.

C-World: A Computer Use Agent Environment Creator (2026.acl-long)

Copied to clipboard

Ziqiao Xi, Shuang Liang, Qi Liu, Jiaqing Zhang, Letian Peng, Fang Nan, Meshal Nayim, Tianhui Zhang, Rishika Mundada, Lianhui Qin, Biwei Huang, Kun Zhou

Challenge:	C-World enables users to build agent environments on demand.
Approach:	They propose a system that enables users to build agent environments on demand.
Outcome:	The proposed system outperforms baselines on 119k samples and achieves Spearman = 0.883 ranking correlation with real execution.

Do Morals Guide How LLMs Think? The Role of Ethical Perspectives in General Problem Solving (2026.acl-long)

Copied to clipboard

Iseo Kim, Eunjin Hong, Juae Kim

ViLL-E: Video LLM Embeddings for Retrieval (2026.acl-long)

Challenge:	Experimental results show that different moral perspectives lead to changes in the model’s decision-making during general reasoning, reflected in both responses and internal representations.
Approach:	They define distinct moral stages based on Kohlberg’s theory of moral development and design prompts to elicit model responses aligned with each condition.
Outcome:	The proposed model responses are validated using the Defining Issues Test, a human evaluation tool.

Copied to clipboard

Rohit Gupta, Jayakrishnan Unnikrishnan, Fan Fei, Sheng Liu, Son Tran, Mubarak Shah

From Trajectories to Graphs: Contract-Checked Editing for Verifier-Guided LLM Reasoning (2026.acl-long)

Challenge:	Video Large Language Models excel at video understanding tasks where outputs are textual . however, they underperform specialized embedding-based models in Retrieval tasks .
Approach:	They propose a video-LLM-based model with an embedding generation mechanism that allows the model to "think longer" for complex videos and stop early for easy ones.
Outcome:	The proposed model outperforms specialized embedding-based models in video understanding tasks while remaining competitive on VideoQA tasks.

Copied to clipboard

Rui Li, Shuang Cao

SiLP: Enhancing Non-Dominant Language Capabilities with a Selective Bidirectional Language Projection Framework (2026.acl-long)

Challenge:	Existing methods for inference-time search refine single trajectories and lack a reliable mechanism for composing partial solutions across candidates.
Approach:	a new method uses a gate-based algorithm to validate a nontrivial edit before invoking the verifier.
Outcome:	a new method improves verifier-runnable recombination and accuracy over existing methods . it outperforms execution-guided beam search on Spider and humanEval-MF on MCTS . a contract-checked graph editing improves recompilation and recombines partial solutions .

Copied to clipboard

Junpeng Liu, Jiuyi Li, Kaiyu Huang, Bo Jin, Degen Huang, Hui Xiong

Simple Agents, Biased Judges: Efficient Multi-Party Dialogue Generation & The Evaluation Gap (2026.acl-long)

Challenge:	Existing methods to improve performance of large language models rely on additional training objectives or language-specific parameters.
Approach:	They propose a bidirectional language projection framework that enables efficient multilingual alignment and language shift using the intrinsic parameters.
Outcome:	The proposed framework improves performance of non-dominant languages and improves internal representations.

Copied to clipboard

Kunal Samanta, Faisal Tareque Shohan, Amine Trabelsi, Richard Khoury

EdgeFormer: Latency-Aware Collaborative Multi-Head Attention of Transformer Inference in Edge Networks (2026.acl-long)

Challenge:	Multiparty social dialogue is difficult to formalize and expensive to evaluate, especially at scale.
Approach:	They propose a lightweight and controllable multi-party dialoguegeneration framework as an experimental instrument for studying generation and evaluation in social interaction.
Outcome:	The proposed framework shows that human judgments against state-of-the-art LLM judges are consistent with human preferences for naturalness, engagingness, and overall quality in multi-party social dialogue.

Copied to clipboard

Yiming Yao, Jianwei Niu, Bin Dai, Tao Ren

LEAF: Knowledge Distillation of Text Embedding Models with Teacher-Aligned Representations (2026.acl-long)

Challenge:	Existing methods for inference in centralized cloud pose privacy risks due to sensitive data.
Approach:	They propose a latency-aware framework for distributed Transformer inference in resource-constrained edge networks.
Outcome:	The proposed framework achieves 2.01 times inference acceleration over state-of-the-art baselines with leq1.06% accuracy loss, maintaining robustness under varying edge conditions.

Copied to clipboard

Robin Vujanic, Thomas Rückstieß

SAT: Balancing Reasoning Accuracy and Efficiency with Stepwise Adaptive Thinking (2026.acl-long)

Challenge:	a new framework for text embedding models is available for free . asymmetrical architectures allow for flexible asymmetry, while asynchronous architectures require small batches .
Approach:	They propose a knowledge distillation framework for text embedding models that is compatible with their teacher . they publish leaf-ir, a 23M parameters information retrieval oriented model that ranks no.1 on BEIR .
Outcome:	The proposed model is compatible with teacher, enabling flexible asymmetric architectures . it sets a new state-of-the-art (SOTA) on BEIR, and achieves no.1 on the leaderboard .

Copied to clipboard

Weiyang Huang, Xuefeng Bai, Kehai Chen, Xinyang Chen, Yibin Chen, Weili Guan, Min Zhang

Deep Research with Open-Domain Evaluation and Multi-Stage Guardrails for Safety (2026.acl-long)

Challenge:	Large Reasoning Models (LRMs) produce excessively long Chains of Thought (COT) Existing solutions that improve token efficiency but sacrifice fine-grained control can disrupt the logical integrity of the reasoning process.
Approach:	They propose a framework that performs step-level, difficulty-aware pruning while preserving the core reasoning structure.
Outcome:	Experiments show that SAT reduces reasoning tokens by 40% while maintaining or improving accuracy.

Copied to clipboard

Wei-Chieh Huang, Henry Peng Zou, Yaozu Wu, Dongyuan Li, Yankai Chen, Weizhi Zhang, Yangning Li, Angelo Zangari, Jizhou Guo, Chunyu Miao, Liancheng Fang, Langzhou He, Yinghui Li, Renhe Jiang, Philip S. Yu

P-Check: Advancing Personalized Reward Model via Learning to Generate Dynamic Checklist (2026.acl-long)

Challenge:	Existing deep research frameworks lack adequate evaluation procedures and stage-specific protections.
Approach:	They propose a framework with open-domain evaluation and a stage-wise safety benchmark to address this oversight.
Outcome:	The proposed framework improves defense success rates by 16.53% while reducing over-refusal rates to approximately 6%.

Copied to clipboard

Kwangwook Seo, Dongha Lee

Challenge:	Existing approaches to personalized reward modeling treat user context as static or implicit conditioning signal, failing to capture dynamic nature of human judgment.
Approach:	They propose a personalized reward modeling framework that synthesizes dynamic evaluation criteria for guiding the reward prediction.
Outcome:	The proposed framework improves reward accuracy and enhances downstream personalized generation.

Locate and Explain: Joint Multimodal Emotion Cause Extraction and Summarization in Conversation (2026.acl-long)

Copied to clipboard

Jikun Wan, Chen Gong, Guohong Fu

Understanding the Behaviors of Environment-aware Information Retrieval (2026.acl-long)

Challenge:	Existing studies focus on utterance-level emotion cause extraction and multimodal emotion cause generation, resulting in subjective and inconsistent annotations.
Approach:	They propose a task that extracts emotion cause utterances and generates cause summaries . they propose utterrance-level emotion cause extraction and multimodal emotion cause generation tasks .
Outcome:	The proposed task extracts emotion cause utterances and generates cause summaries . the proposed task establishes strong benchmark results for the proposed project .

Copied to clipboard

Ruifeng Yuan, Chaohao Yuan, David Dai, Yu Rong, Hong Cheng, Hou Pong Chan, Chenghao Xiao

ServImage: An Image Generation and Editing Benchmark from Real-world Commercial Imaging Services (2026.acl-long)

Challenge:	Recent retrieval-augmented generation approaches have demonstrated strong capability in handling complex queries.
Approach:	They propose a branching-based rollout technique that improves training stability . they find different retrievers exhibit distinct optimal query styles .
Outcome:	The proposed method improves training stability and improves retrieval-aware systems.

Copied to clipboard

Fengxian Ji, Jingpu Yang, Zirui Song, Lang Gao, Junhong Liang, Zhenhao Chen, Jinghui Zhang, Xiuying Chen

FAIRGAMER: Evaluating Social Biases in LLM-Based Video Game NPCs (2026.acl-long)

Challenge:	Recent image generation and editing models demonstrate robust adherence to instructions and high visual quality on academic benchmarks.
Approach:	They propose a benchmark that correlates image outputs with economic value in commercial design projects.
Outcome:	ServImage benchmarks show image generation models perform well on academic benchmarks but are uncertain on commercial projects.

Copied to clipboard

Bingkang Shi, Jen-tse Huang, Luo Long, Tianyu Zong, Hongzhu Yi, Yuanxiang Wang, Songlin Hu, Xiaodan Zhang, Zhongjiang Yao

EVOTOOL: Self-Evolving Tool-Use Policy Optimization in LLM Agents via Blame-Aware Mutation and Diversity-Aware Selection (2026.acl-long)

Challenge:	Large Language Models (LLMs) have enhanced or replaced traditional non-player characters in video games.
Approach:	They propose a benchmark to evaluate social biases across three interaction patterns: transaction, cooperation, and competition.
Outcome:	The proposed benchmark assesses four bias types across transaction, cooperation, and competition using a novel metric, FairMCV.

Copied to clipboard

Shuo Yang, Caren Han, Xueqi Ma, Yan Li, Mohammad Reza Ghasemi Madani, Eduard Hovy

Challenge:	Existing approaches to optimize tool-use policies are monolithic and prone to entangling behaviors.
Approach:	They propose a framework that decomposes agent’stool-use policy into four modules and improves them via three mechanisms.
Outcome:	The proposed framework outperforms strong baselines on bothGPT-4.1 and Qwen3-8B while maintaining superior efficiency and transferability.

A Lightweight Explainable Guardrail for Prompt Safety (2026.acl-long)

Copied to clipboard

Md Asiful Islam, Mihai Surdeanu

Explainable Disentangled Representation Learning for Generalizable Authorship Attribution in the Era of Generative AI (2026.acl-long)

Challenge:	Existing methods to detect unsafe prompts are based on reinforcement learning with human feedback (RLHF) or direct preference optimization (DPO) . however, these methods lack explainability and are costly to implement.
Approach:	They propose a lightweight explainable guardrail method to detect unsafe prompts using a multi-task learning architecture and a novel strategy to counteract confirmation biases.
Outcome:	The proposed method obtains equivalent or better performance than the state-of-the-art for both prompt classification and explainability on three datasets.

Copied to clipboard

Hieu Man, Van-Cuong Pham, Nghia Trung Ngo, Franck Dernoncourt, Thien Huu Nguyen

PodBench: A Comprehensive Benchmark for Instruction-Aware Audio-Oriented Podcast Script Generation (2026.acl-long)

Challenge:	Existing methods struggle with content-style entanglement, leading to poor generalization across domains.
Approach:	They propose an explanation-by-design framework that explicitly disentangles style from content through architectural separation-by design.
Outcome:	The proposed framework disentangles style from content through architectural separation-by-design.

Copied to clipboard

Chenning Xu, Mao Zheng, Mingyu Zheng, Mingyang Song

MIND: From Passive Mimicry to Active Reasoning through Capability-Aware Multi-Perspective CoT Distillation (2026.acl-long)

Challenge:	Podcast script generation is a challenging task for large language models, but evaluation resources are limited.
Approach:	They propose a benchmark to evaluate podcast script generation using a multifaceted evaluation framework . PodBench is a prototype that integrates quantitative constraints with LLM-based quality assessment .
Outcome:	The proposed framework integrates quantitative constraints with LLM-based quality assessment.

Copied to clipboard

Jin Cui, Jiaqi Guo, Jiepeng Zhou, Ruixuan Yang, Jiayi Lu, Jiajun Xu, Jiangcheng Song, Boran Zhao, Pengju Ren

Knowledge Vector of Logical Reasoning in Large Language Models (2026.acl-long)

Challenge:	Existing approaches restrict students to following a single golden rationale and treat different reasoning paths independently, causing suboptimal performance.
Approach:	They propose a capability-adaptive framework that transitions distillation from passive mimicry to active cognitive construction and employ a feedback-driven inertia calibration mechanism to align supervision with the student’s current adaptability.
Outcome:	Experiments show that the proposed framework achieves state-of-the-art performance on both in-distribution and out-of distribution benchmarks.

Copied to clipboard

Zixuan Wang, Yuanyuan Lei

BanHADEX: Towards Explainable HAte Speech Detection in Bangla Using Human Annotated EXplanation (2026.acl-long)

Challenge:	Logical reasoning is a central capability in LLMs, but understanding their abilities remains poorly understood.
Approach:	They propose to refine the knowledge representations of each reasoning type in LLMs to encourage complementarity . they propose to use complementary loss and subspace constraint loss to enhance complementarities .
Outcome:	The proposed framework encourages complementarity between the different types of reasoning in LLMs.

Copied to clipboard

Faisal Hossain Raquib, Akm Moshiur Rahman Mazumder, Md Fahim, Md Tahmid Hasan Fuad, Md Farhan Ishmam, Faria Sultana, M Ashraful Amin, Amin Ahsan Ali, Akmmahbubur Rahman

SMILE-Next: Teaching Large Language Models to Detect, Classify, and Reason about Laughter (2026.acl-long)

Challenge:	Existing studies in Bangla focus on hate classification while overlooking interpretability.
Approach:	They propose to create a dataset with human-annotated labels for banla that contains 19,203 YouTube comments spanning April 2024–June 2025.
Outcome:	The proposed dataset outperforms existing datasets on open and closed-source LLMs on interpretability and better understanding of hate speech in linguistically rich yet under-resourced languages.

Copied to clipboard

Lee Jung-Mok, Kim Sung-Bin, Joohyun Chang, Lee Hyun, Tae-Hyun Oh

MolMem: Memory-Augmented Agentic Reinforcement Learning for Sample-Efficient Molecular Optimization (2026.acl-long)

Challenge:	Existing approaches to understanding laughter or humor focus on narrowly defined tasks such as detecting humor and estimating humor intensity.
Approach:	They propose a dataset for real-world laughter understanding with multimodal textual representations and question–answer annotations.
Outcome:	The proposed framework outperforms baselines in three laughter-related tasks, showing that it is robust.

Copied to clipboard

Ziqing Wang, Yibo Wen, Abhishek Pandey, Han Liu, Kaize Ding

Activation Decomposition and Steering for LLM Backdoor Remediation (2026.acl-long)

Challenge:	Existing methods for molecular optimization use expensive oracle evaluations to achieve sample efficiency under limited oracular budget.
Approach:	They propose a framework that iteratively refines a lead compound to improve molecular properties while preserving structural similarity to the original molecule.
Outcome:	The proposed framework achieves 90% success on single-property tasks and 52% on multi-propety task using only 500 oracle calls.

Copied to clipboard

Lingfeng Zhong, Qiongkai Xu, Usman Naseem

A Survey of Large Language Models for Text-Guided Molecular Discovery: From Molecule Generation to Optimization (2026.acl-long)

Challenge:	Existing approaches to defending against LLM backdoors rely on auxiliary models or safety-related datasets.
Approach:	They propose a method which contrasts benign and poisoned settings to decompose feature vectors for steering without auxiliary models or datasets.
Outcome:	The proposed method achieves better defense qualities than existing steering strategies.

Copied to clipboard

Ziqing Wang, Kexin Zhang, Zihan Zhao, Yibo Wen, Abhishek Pandey, Han Liu, Kaize Ding

Reward Alignment Optimization: A Direct Point-wise Alignment Approach (2026.acl-long)

Challenge:	Large language models (LLMs) are introducing a paradigm shift in molecular discovery by enabling text-guided interaction with chemical spaces through natural language and symbolic notations.
Approach:	They analyze the current LLM learning paradigms to tackle four critical evaluation dimensions that have emerged as critical dimensions in recent studies.
Outcome:	The proposed models are able to interact with chemical spaces through natural language and symbolic notations, and have emerging extensions to incorporate multi-modal inputs.

Copied to clipboard

Zelin Li, Jia Leng, Dawei Song, Yangen Hu

Programming over Thinking: Efficient and Robust Multi-Constraint Planning (2026.acl-long)

Challenge:	Existing Direct Alignment Algorithms (DAAs) are limiting in generalizaiton to implicit rewards.
Approach:	They propose a point-wise direct alignment method that uses an explicit reward model to specify exact target generation probabilities and align the policy offline towards them.
Outcome:	The proposed method outperforms existing direct alignment algorithms while enabling controllable target probability distributions.

Copied to clipboard

Derrick Goh Xin Deik, Quanyu Long, Zhengyuan Liu, Nancy F. Chen, Wenya Wang

LANG: Reinforcement Learning for Multilingual Reasoning with Language-Adaptive Hint Guidance (2026.acl-long)

Challenge:	Existing large language model approaches lack flexibility in multi-constraint planning . SCOPE achieves state-of-the-art performance while lowering cost and latency .
Approach:	They propose a framework that disentangles query-specific problem reasoning from generic code execution.
Outcome:	The Scalable Code Planning Engine achieves state-of-the-art performance while lowering cost and latency.

Copied to clipboard

Yuchun Fan, Bei Li, Peiguang Li, Yilin Wang, Yongyu Mu, Jian Yang, Xin Chen, Rongxiang Weng, Jingang Wang, Xunliang Cai, JingBo Zhu, Tong Xiao

Can You Make It Sound Like You? Post-Editing LLM-Generated Text for Personal Style (2026.acl-long)

Challenge:	Existing methods for enhancing multi-step reasoning have not fully translated to multilingual contexts.
Approach:	They propose a framework that leverages language-conditioned hints to guide exploration in non-English reasoning tasks.
Outcome:	Empirical results show that the proposed framework improves reasoning performance without compromising language consistency.

Copied to clipboard

Connor Baumler, Calvin Bao, Huy Nghiem, Xinchen Yang, Marine Carpuat, Hal Daumé Iii

ReCode: Reinforcing Code Generation with Reasoning-Process Rewards (2026.acl-long)

Challenge:	Despite the growing use of large language models for writing tasks, it remains unclear whether users can effectively reshape LLM-generated text to reflect their personal style.
Approach:	They conduct an online study in which participants post-edit LLM-generated drafts for writing tasks where personal style matters to them.
Outcome:	The results show that post-editing increases stylistic similarity to unassisted writing and reduces similarity with fully LLM-generated output.

Copied to clipboard

Lishui Fan, Yu Zhang, Mouxiang Chen, Zhongxin Liu

PPA-Plan: Proactive Pitfall Avoidance for Reliable Planning in Long-Context LLM Reasoning (2026.acl-long)

Challenge:	Bringing process-level supervision into RL often neglects optimizing reasoning quality.
Approach:	They propose a framework for RL that integrates reasoning-process rewards with strict execution outcomes and a benchmark comprising preference pairs of superior and inferior reasoning processes.
Outcome:	The proposed framework outperforms the base version of ReCode by 16.1% and reaches performance comparable to GPT-4-Turbo.

Copied to clipboard

Byeongjin Kim, Gyuwan Kim, Seo Yeon Park

OPeRA: A Dataset of Observation, Persona, Rationale, and Action for Evaluating LLMs on Human Online Shopping Behavior Simulation (2026.acl-long)

Challenge:	Large language models struggle with reasoning over long contexts where relevant information is sparsely distributed.
Approach:	They propose a plan-and-execute framework that decomposes long-context tasks into an explicit planning stage followed by step-by-step step-through planning.
Outcome:	Experiments on long-context QA benchmarks show that PPA-Plan outperforms plan-and-execute methods and direct prompting.

Copied to clipboard

Ziyi Wang, Yuxuan Lu, Wenbo Li, Amirali Amini, Bo Sun, Yakov Bart, Weimin Lyu, Jiri Gesi, Tian Wang, Jing Huang, Yu Su, Upol Ehsan, Malihe Alikhani, Toby Jia-Jun Li, Lydia Chilton, Dakuo Wang

Challenge:	evaluating LLMs' ability to mimic real user behavior remains an open challenge due to the lack of high-quality, publicly available datasets that capture both the observable actions and the internal reasoning of an actual user.
Approach:	They propose a dataset of Observation, Persona, Rationale, and Action collected from real human participants during online shopping sessions.
Outcome:	The proposed dataset is the first to evaluate how well current LLMs can accurately simulate the next web action of a specific user.

Can LLM Agents Simulate Multi-Turn Human Behavior? Evidence from Real Online Customer Behavior Data (2026.acl-long)

Copied to clipboard

Yuxuan Lu, Jing Huang, Yan Han, Bingsheng Yao, Sisong Bei, Yaochen Xie, Yisi Sang, Qi He, Dakuo Wang

RATE: Reviewer Profiling and Annotation-free Training for Expertise Ranking in Peer Review Systems (2026.acl-long)

Challenge:	Recent research shows that LLM Agents can generate “believable” human behaviors via prompt-only methods, leaving open questions of whether they can accurately generate step-by-step actions in multi-turn interaction tasks.
Approach:	They propose to use shopping data to evaluate LLMs' ability to accurately generate step-by-step actions in a multi-turn interaction task.
Outcome:	The proposed model achieves 17.26% action generation accuracy and 33.86% F1 score on final purchase prediction, representing improvements of 5.4% and 13.85% over baselines.

Copied to clipboard

Weicong Liu, Zixuan Yang, Yibo Zhao, Xiang Li

The Bitter Lesson of Diffusion Language Models for Agentic Workflows: A Comprehensive Reality Check (2026.acl-long)

Challenge:	LR-bench is a high-fidelity, up-to-date benchmark curated from 2024–2025 AI/NLP manuscripts with five-level self-assessed familiarity ratings collected via a large-scale email survey .
Approach:	They propose a reviewer-centric ranking framework that distills each reviewer’s recent publications into compact keyword-based profiles and fine-tunes an embedding model with weak preference supervision constructed from heuristic retrieval signals.
Outcome:	The proposed framework outperforms existing benchmarks and the CMU gold-standard dataset in the evaluation of AI/NLP manuscripts.

Copied to clipboard

Qingyu Lu, Liang Ding, Kanjian Zhang, Jinxia Zhang, Dacheng Tao

Trajectory2Task: Training Robust Tool-Calling Agents with Synthesized Yet Verifiable Data for Complex User Intents (2026.acl-long)

Challenge:	Embodied and Tool-Calling agents are effective in planning and complex reasoning, but require causal, precise, and logically grounded reasoning mechanisms to be viable for agentic tasks.
Approach:	They propose a framework that integrates dLLMs as plug-and-play cognitive cores.
Outcome:	The proposed model breaks the sequential latency bottleneck in agentic interactions.

Copied to clipboard

Ziyi Wang, Yuxuan Lu, Yimeng Zhang, Pei Chen, Ziwei Dong, Jing Huang, Jiri Gesi, Xianfeng Tang, Chen Luo, Qun Liu, Yisi Sang, Hanqing Lu, Manling Li, Jin Lai, Dakuo Wang

CT-FineBench: A Diagnostic Fidelity Benchmark for Fine-Grained Evaluation of CT Report Generation (2026.acl-long)

Challenge:	Tool-calling agents are increasingly deployed in real-world customer-facing workflows . but most studies on tool-callers focus on idealized settings with general, fixed, and well-specified tasks.
Approach:	They propose a tool-calling agent-based data pipeline that converts trajectories into user-facing tasks with controlled intent adaptations.
Outcome:	The proposed pipeline can be used to study tool use under three scenarios.

Copied to clipboard

Ruifeng Yuan, Wanxing Chang, Weiwei Cao, Bowen Shi, Zhongyu Wei, Ling Zhang, Jianpeng Zhang

Lost in Execution: On the Multilingual Robustness of Tool Calling in Large Language Models (2026.acl-long)

Challenge:	Existing evaluation metrics for radiology report generation focus on lexical overlap and entity matching.
Approach:	They propose a benchmark to evaluate the fine-grained factual consistency of CT reports . they use a question-answering process to query a machine-generated report .
Outcome:	The proposed benchmark evaluates the fine-grained factual consistency of CT reports . it correlates better with expert clinical assessment and is more sensitive to errors .

Copied to clipboard

Zheng Luo, T Pranav Kutralingam, Ogochukwu N. Okoani, Wanpeng Xu, Hua Wei, Xiyang Hu

MTA:A Merge-then-Adapt Framework for Personalized Large Language Models (2026.acl-long)

Challenge:	Large Language Models (LLMs) are increasingly deployed as agents that invoke external tools through structured function calls.
Approach:	They introduce a diagnostic benchmark and conduct a systematic evaluation of multilingual tool calling across Chinese, Hindi, and the low-resource language Igbo.
Outcome:	The proposed benchmarks show that multilingual tool calling fails despite correct intent understanding and tool selection.

Copied to clipboard

Xiaopeng Li, Yuanjin Zheng, Wanyu Wang, Wenlin Zhang, Pengyue Jia, Yingyi Zhang, Haiying He, Mengyang Ma, Yiqi Wang, Maolin Wang, Xuetao Wei, Xiangyu Zhao

Investigating Counterfactual Unfairness in LLMs towards Identities through Humor (2026.acl-long)

Challenge:	Personalized Large Language Models (PLLMs) aim to align outputs with individual user preferences . current methods of fine-tuning a separate module for each user are unscalable .
Approach:	They propose a Merge-then-Adapt framework for Personalized Large Language Models . they construct a shared Meta-LoRA bank and propose an Adaptive LoRA Fusion stage .
Outcome:	The proposed framework outperforms existing SOTA methods on the LaMP benchmark.

Copied to clipboard

Shubin Kim, Yejin Son, Junyeong Park, Keummin Ka, Seungbeen Lee, Jaeyoung Lee, Hyeju Jang, Alice Oh, Youngjae Yu

Response-G1: Explicit Scene Graph Modeling for Proactive Streaming Video Understanding (2026.acl-long)

Challenge:	Large Language Models (LLMs) absorb social and cultural biases embedded in vast web-scale corpora and are increasingly deployed in high-stakes domains such as hiring, education, and law.
Approach:	They propose a framework to investigate counterfactual unfairness through humor by observing how the model’s responses change when we swap who speaks and who is addressed while holding other factors constant.
Outcome:	The proposed framework covers humor generation refusal, speaker intention inference, and relational/societal impact prediction tasks.

Copied to clipboard

Ke Ma, Jiaqi Tang, Bin Guo, Xueting Han, Ruonan Xu, Qingfeng He, Ziheng Wang, Xu Wang, Qifeng Chen, Zhiwen Yu, Yunhao Liu

POLYCHARTQA: Benchmarking Large Vision-Language Models with Multilingual Chart Question Answering (2026.acl-long)

Challenge:	Existing methods for streaming video understanding are query-agnostic and implicitly model video evidence.
Approach:	They propose a framework that establishes explicit, structured alignment between the accumulated video evidence and the query’s expected response conditions via scene graphs.
Outcome:	The proposed model achieves more interpretable and accurate response timing decisions on both proactive and reactive tasks.

Copied to clipboard

Yichen Xu, Liangyu Chen, Liang Zhang, Zihao Yue, Jianzhe Ma, Wenxuan Wang, Qin Jin

Nature-Inspired Population-Based Evolution of Large Language Models (2026.acl-long)

Challenge:	Existing chart understanding benchmarks are overwhelmingly English-centric, limiting their accessibility and relevance to global audiences.
Approach:	They propose a multilingual chart question answering benchmark that enables efficient multilingual generation via data translation and code reuse.
Outcome:	The proposed benchmark systematically evaluates multilingual chart understanding on state-of-the-art LVLMs and shows a significant performance gap between English and other languages.

Copied to clipboard

Yiqun Zhang, Peng Ye, Xiaocui Yang, Shi Feng, Shufei Zhang, Lei Bai, Wanli Ouyang, Shuyue Hu

MTRouter: Cost-Aware Multi-Turn LLM Routing with History–Model Joint Embeddings (2026.acl-long)

Challenge:	a new framework for population-based evolution of large language models is emerging . a population-driven evolution of LLMs is a key component of evolution, authors say .
Approach:	They propose a framework that allows for population-based evolution of large language models . they start with a population of parent LLMs and allow this population to evolve .
Outcome:	The proposed framework outperforms existing methods on 12 datasets.

Copied to clipboard

Yiqun Zhang, Hao Li, Zihan Wang, Shi Feng, Xiaocui Yang, Daling Wang, Bo Zhang, Lei Bai, Shuyue Hu

Feeling Right vs. Being Right: How AI Sycophancy Affects Value-Laden Deliberation (2026.acl-long)

Challenge:	Multi-turn, long-horizon tasks require dozens of sequential model calls per episode.
Approach:	They propose a cost-aware multi-turn LLM routing tool which encodes interaction history and candidate models into joint history–model embeddings and learns an outcome estimator from logged trajectories to predict turn-level model utility.
Outcome:	The proposed model reduces cost and performance by 58.7% on ScienceWorld and on Humanity’s Last Exam (HLE) and even reduces costs for held-out tasks.

Copied to clipboard

Jeongwoo Ryu, Soomin Kim, Jinsu Eun, Kyusik Kim, Changhoon Oh, Bongwon Suh

Challenge:	Unlike human flattery, AI sycophancy is intentional and self-interested . scophancies are a byproduct of RLHF's user-preference alignment process .
Approach:	They propose to operationalize AI sycophancy as excessive face-saving, either active (preserving positive face through agreement) or passive (preserving negative face by withholding challenge).
Outcome:	The findings show that sycophancy is a byproduct of RLHF's user-preference alignment process and that it is not a human trait.

ParaSuite: Boosting LLM Reasoning via Paradox Resolution (2026.acl-long)

Copied to clipboard

Bin Chen, Yu Zhang, Hongfei Ye, Huiyang Wang, Wenxi Liu, Hongyang Chen

ReTRE: Benchmarking LLM Transfer Robustness with Structure-Preserving Variants (2026.acl-long)

Challenge:	Existing benchmarks for paradox research focus on checking basic logical consistency and not reflective reasoning.
Approach:	They propose a pipeline dedicated to paradox research that automates data synthesis, evaluation, and training.
Outcome:	The proposed pipeline improves paradoxical and general STEM reasoning.

Copied to clipboard

ZhongDong Li, Weijie Shi, Yue Cui, Haolun MA, Yuanjun Liu, Jiawei Li, An Liu, Jia Zhu, Jiajie Xu

Do MLLMs Capture How Interfaces Guide User Behavior? A Benchmark for Multimodal UI/UX Design Understanding (2026.acl-long)

Challenge:	Learning transfer theory emphasizes that applying acquired knowledge to novel manifestations is a key signal of deep understanding
Approach:	They propose a benchmark that probes transfer robustness along two rewrite levels: Near Transfer and Far Transfer.
Outcome:	The proposed benchmark demonstrates that large language models are robust when faced with novel manifestations of the same problem.

Copied to clipboard

Jaehyun Jeon, Min Soo Kim, Janghan Yoon, Sumin Shim, Yejin Choi, Hanbin Kim, Dae Hyun Kim, Youngjae Yu

DarwinTOD: LLM-Driven Lifelong Self-evolution for Task-oriented Dialog Systems (2026.acl-long)

Challenge:	Recent studies focus on surface-level features, overlooking how design choices influence user behavior at scale.
Approach:	They propose a benchmark for multimodal understanding of how UI/UX design affects user behavior built on 300 real-world UI image pairs from industry A/B tests.
Outcome:	The proposed benchmarks show that models exhibit limited understanding of the behavioral impact of UI/UX design.

Copied to clipboard

Shuyu Zhang, Yujie Liu, Xinru Wang, Cheng Zhang, Yanmin Zhu, Bin Li

Route to Rome Attack: Directing LLM Routers to Expensive Models via Adversarial Suffix Optimization (2026.acl-long)

Challenge:	Continual learning approaches fail to achieve autonomy lifelong improvement in dynamic environments . current task-oriented dialog systems are static, unable to learn from ongoing interactions .
Approach:	They propose a lifelong self-evolving dialog framework that integrates evolutionary computation and LLM driven self-improvement into a single framework.
Outcome:	The proposed framework surpasses state-of-the-art methods and exhibits continuous performance gains throughout evolution.

Copied to clipboard

Haochun Tang, Yuliang Yan, Jiahua Lu, Huaxiao Liu, Enyan Dai

GUI0: Self-Evolving Foundational GUI Agents in Super App Ecosystems (2026.acl-long)

Challenge:	Existing routing attacks depend on white-box access or heuristic prompts, rendering them ineffective in real-world black-box scenarios.
Approach:	They propose a cost-aware routing strategy that routes queries to the least-cost model . they propose heuristic prompts that are ineffective in real-world black-box scenarios .
Outcome:	The proposed approach significantly increases the routing rate to expensive models on queries of different distributions.

Copied to clipboard

Xinyi Wang, Wei Dai, Kyle Qiao, Ke Wang, Peng Chen, Gang Cao, null Kangqin, Zhongpu Wang, Xiaode Zhang, Yanming Liu, Jihao Gu, Jingtao Xu, Gong Zhi

Understanding the Prompt Sensitivity (2026.acl-long)

Challenge:	Automated interaction with graphical user interfaces (GUIs) is central to general artificial intelligence, but remains challenging within Super App ecosystems.
Approach:	They propose a framework synergizing autonomous data synthesis with dual-agent co-evolution . GUI0 establishes a domain-aware foundation model via synthesized corpora and employs curriculum-driven reinforcement learning .
Outcome:	The proposed framework outperforms Gemini-2.5-Pro and Claude-4-Sonnet in the SuperAPP benchmark and has universal efficacy across base models.

Copied to clipboard

Yang Liu, Chenhui Chu

rSIM: Incentivizing Reasoning Capabilities of LLMs via Reinforced Strategy Injection (2026.acl-long)

Challenge:	Prompt sensitivity is a measure of how strongly the output of a large language model (LLM) depends on the exact wording of its input prompt.
Approach:	They consider LLMs as multivariate functions and perform a first-order Taylor expansion to analyze the relationship between meaning-preserving prompts, their gradients, and log probabilities of the model’s next token.
Outcome:	The proposed model disperses meaning-preserving inputs, making it difficult to reduce to 0. The proposed models also dispersing prompt variants are more likely to introduce prompt sensitivity risks in LLMs.

Copied to clipboard

Sijia Chen, Baochun Li, Di Niu

Automatic and Reliable Evaluation for Academic Caption-to-Figure Generation with LMMs (2026.acl-long)

Challenge:	Existing literature on Reasoning Language Models (RLMs) focuses on the ability to integrate reasoning strategies into the chain-of-thought process, contributing to improved problem-solving accuracy.
Approach:	They propose a reinforced strategy injection mechanism that enables any LLM to become an RLM by employing a small planner to guide the LLM's CoT through the adaptive injection of reasoning strategies.
Outcome:	The proposed model outperforms existing models in mathematical, coding, and financial reasoning tasks and is generalizable.

Copied to clipboard

Guanghui Ye, Huan Zhao, Qin Zhu, Fengnan Li, Jiaqi Li, Yixian Shen, Zhonghao Ren, Zhihua Jiang

SafeConstellations: Mitigating Over-Refusals in LLMs Through Task-Aware Representation Steering (2026.acl-long)

Challenge:	Existing datasets for evaluating text-to-image generation focus mostly on real-life images, which poses challenges for assessing academic figure generation given real scientific captions.
Approach:	They propose a dataset that first provides a Holistic Evaluation for Academic caption-to-Figure Generation (HE4AFG) they collect real figure captions from 8 scientific domains and generate 3,900 evaluation samples .
Outcome:	The proposed model provides high-quality human ratings in terms of three aspects—scientific aesthetic (SA), topic relevance (TR), and attribute correctness (AC).

Copied to clipboard

Utsav Maskey, Sumit Yadav, Mark Dras, Usman Naseem

AIPO: Adaptive Information Guided Token-Level Reinforcement Learning for Large Language Model Reasoning (2026.acl-long)

Challenge:	Current safety alignment methods fail to identify intended benign task before refusing to respond.
Approach:	They propose a method that uses inference-time trajectory-shifting to guide model behavior . they show that LLMs persist in refusing inputs containing harmful content .
Outcome:	The proposed approach reduces over-refusals with minimal impact on utility.

Copied to clipboard

Bin Chen, Hongfei Ye, Huiyang Wang, Wenxi Liu, Yu Zhang, Furui Liu

AgentGym2: Benchmarking Large Language Model Agents in De-Idealized Real-World Environments (2026.acl-long)

Challenge:	Existing RLVR methods focus on all generated tokens rather than on which tokens contribute to reasoning.
Approach:	They propose to use a Random–Fourier approximation of the Hilbert–Schmidt Independence Criterion to focus updates on decisive tokens discovered on the fly to improve the efficiency of mutual-information estimation.
Outcome:	The proposed approach yields +20% accuracy over strong RLVR baselines while updating merely 10% of tokens, demonstrating superior efficiency and effectiveness.

Copied to clipboard

Zhiheng Xi, Dingwen Yang, Jiaqi Liu, Jixuan Huang, Honglin Guo, Baodai Huang, Tinggang Chen, Qi Zhang, Zhonghang Lu, Chenyu Liu, Jiajun Sun, Jiazheng Zhang, Dingwei Zhu, Xin Guo, Junzhe Wang, Zhihao Zhang, Yuming Yang, Junjie Ye, Minghe Gao, Dongrui Liu, Jiaming Ji, Guohao Li, Tao Gui, Qi Zhang, Xuanjing Huang

Through the Magnifying Glass: Adaptive Perception Magnification for Hallucination-Free VLM Decoding (2026.acl-long)

Challenge:	Existing benchmarks evaluate agents in simplified, idealized settings, relying on pre-packaged tool interfaces, overlooking critical steps, and assume inputs are clean and fully specified.
Approach:	They propose a framework that evaluates language agents in simplified, idealized settings . they show that even SOTA systems like Gemini and GPT-5 struggle on AgentGym2 .
Outcome:	Experiments on 15 proprietary and open-source models show that even SOTA systems like Gemini and GPT-5 struggle on AgentGym2 .

Copied to clipboard

Shunqi Mao, Chaoyi Zhang, Weidong Cai

Beyond Fully Random Masking: Attention-Guided Denoising and Optimization for Diffusion Language Models (2026.acl-long)

Challenge:	Existing vision-language models suffer from visual hallucination, where the generated responses contain inaccuracies that are not grounded in the visual input.
Approach:	They propose a visual decoding method that iteratively isolates relevant visual tokens based on attention and magnifies the corresponding regions.
Outcome:	The proposed method reduces language biases and amplifies weights of visual embedding during decoding, while still preserving strong reasoning capabilities.

Copied to clipboard

Jia Deng, Junyi Li, Xin Zhao, Jinpeng Wang, Hongyu Lu, Ji-Rong Wen

MemRec: Collaborative Memory-Augmented Agentic Recommender System (2026.acl-long)

Challenge:	Existing methods for full-attention dLLMs rely on random masking strategies that overlook intrinsic token dependencies.
Approach:	They propose an attention-guided denoising and optimization framework that aligns training and optimization with attention-derived dependencies.
Outcome:	The proposed framework outperforms state-of-the-art methods on mathematical and coding benchmarks.

Copied to clipboard

Weixin Chen, Yuhan Zhao, Jingyuan Huang, Zihe Ye, Mingxuan Ju, Tong Zhao, Neil Shah, Li Chen, Yongfeng Zhang

INFACT: A Diagnostic Benchmark for Induced Faithfulness and Factuality Hallucinations in Video-LLMs (2026.acl-long)

Challenge:	Existing recommender systems rely on semantic user and item memories to make predictions, but these memories are kept in isolation.
Approach:	They propose a framework that architecturally decouples memory management from reasoning to decouple memory management and reasoning from the user and item memories.
Outcome:	The proposed framework decouples memory management from reasoning and achieves state-of-the-art performance on four benchmarks.

Copied to clipboard

null Yangjunqi, Yuecong Min, Jie Zhang, Shiguang Shan, Xilin Chen

CLAOCS-TX: Cross-Lingual Triplet Extraction with Aspect-Opinion-Aware Code-Switched Prompting and LLM-Guided Contrastive Distillation (2026.acl-long)

Challenge:	Existing benchmarks only evaluate models in clean settings due to hallucinations .
Approach:	They propose a diagnostic benchmark that evaluates models in four modes for faithfulness and factuality.
Outcome:	The proposed benchmark evaluates models in four modes: Base (clean), Visual Degradation, Evidence Corruption, and Temporal Intervention for order-sensitive items.

Copied to clipboard

Lipika Dewangan, Chandresh Kumar Maurya

DAC-Bench: A Decision-Aware Benchmark for Compositional Mobile GUI Tasks (2026.acl-long)

Challenge:	Existing work on cross-lingual aspect–opinion–sentiment triplet extraction has focused on coarse-grained sentiment classification or aspect extraction.
Approach:	They propose a framework that leverages large language models as pseudo-label generators and semantic teachers for ASTE.
Outcome:	The proposed framework generates reliable pseudo triplets for unlabeled languages, while maintaining high-confidence supervision.

Copied to clipboard

Yuqing Zhang, Honghui Sheng, Xueyu Hu, Shengyu Zhang, Fei Wu

Challenge:	Existing benchmarks focus on short, linear workflows and step-level accuracy, highlighting performance degradations.
Approach:	They propose a decision-aware benchmark with compositional tasks comprising 830 episodes and 11,345 action steps across 35 applications on Android and iOS.
Outcome:	The proposed benchmarks show performance degradation and branch correctness issues in 7 different GUI agents.

GMoE: Global Mixture of Experts with Logit Propagation (2026.acl-long)

Copied to clipboard

Geonwoo Hong, Taehwan Kim

When Does Language Matter? Multilingual Instructions Reveal Step-wise Language Sensitivity in Vision-Language-Action Models (2026.acl-long)

Challenge:	Sparse Mixture of Experts architectures retain large memory footprints and exhibit significant redundancy, both within and across layers.
Approach:	They propose a sparse mixture of experts architecture that uses global experts shared across all layers and adds a Local Expert per layer for layer-specific adaptation.
Outcome:	The proposed architecture reduces computational cost by activating only a subset of experts per token while maintaining strong performance.

Copied to clipboard

Xuan Dong, Zhe Han, Tianhao Niu, Qingfu Zhu, Wanxiang Che

DialDefer: A Framework for Detecting and Mitigating LLM Dialogic Deference (2026.acl-long)

Challenge:	Vision-Language-Action models have shown strong performance in language-conditioned robotic manipulation, yet their robustness to linguistic variation remains poorly understood.
Approach:	They propose a step-wise inference-time intervention that aligns representations according to step language sensitivity, significantly improving performance under linguistic variation.
Outcome:	The proposed model significantly improves performance under linguistic variation under non-English instructions under language-agnostic steps.

Copied to clipboard

Parisa Rabbani, Priyam Sahoo, Ruben Mathew, Aishee Mondal, Harshita Ketharaman, Nimet Beyza Bozdag, Dilek Hakkani-Tür

Benchmarking Fine-Grained Error Detection in Multimodal Reasoning (2026.acl-long)

Challenge:	a single model can shift toward disagreement (skepticism) on graduate-level science and toward agreement (deference) on social judgment.
Approach:	They propose a framework to detect and mitigat framing-induced judgment shifts . they propose 'DialDefer' framework to help model disagreements and disagreements based on attribution .
Outcome:	The proposed framework detects and mitigates dialogic deference shifts in LLMs . human-vs-LLM attribution drives the largest shifts (17.7 pp swing)

Copied to clipboard

Chi-Min Chan, Han Zhu, Chunyang Jiang, Jiaming Ji, Juntao Dai, Wei Xue, Sirui Han, Yike Guo

Parallel Test-Time Scaling for Latent Reasoning Models (2026.acl-long)

Challenge:	Multimodal Process Reward Models (MPRMs) have emerged as a pivotal framework for enhancing the reasoning capabilities of Multimodal Large Language Models.
Approach:	They propose a benchmark specifically designed to evaluate MPRMs’ proficiency in detecting erroneous reasoning steps across diverse error categories.
Outcome:	The proposed model achieves up to 4.8% performance improvement through test-time scaling.

Copied to clipboard

Runyang You, Yongqi Li, Meng Liu, Wenjie Wang, Liqiang Nie, Wenjie Li

SciMDR: Advancing Scientific Multimodal Document Reasoning (2026.acl-long)

Challenge:	Parallel test-time scaling is a pivotal approach for enhancing large language models.
Approach:	They propose two uncertainty-inspired stochastic strategies for parallel test-time scaling for latent reasoning models and a Latent Reward Model for aggregation.
Outcome:	The proposed model scales well with compute and enables effective trajectory selection.

Copied to clipboard

Ziyu Chen, Yilun Zhao, Chengye Wang, Rilyn R. Han, Manasi Patwardhan, Arman Cohan

When Benchmarks Leak: Inference-Time Decontamination for LLMs (2026.acl-long)

Challenge:	Current models struggle to provide reliable assistance in real-world scientific workflows because evidence is distributed across long, multimodal documents.
Approach:	They propose a framework for QA Synthesis and document-scale regrounding that generates faithful, isolated QA pairs and reasoning on focused segments.
Outcome:	The proposed framework achieves significant improvements across multiple QA benchmarks, particularly in tasks requiring complex document-level reasoning.

Copied to clipboard

Jianzhe Chai, YU Zhe, Jun Sakuma

Challenge:	a large number of large language models (LLMs) are being evaluated for their performance, but their reliability is threatened by test set contamination.
Approach:	They propose a framework that decontaminates large language models by applying small perturbations to the input embedding space.
Outcome:	The proposed framework achieves strong decontamination effectiveness while incurring minimal degradation in benign utility.

HSGraphAgent: Knowledge-Graph-Guided Large Language Models for Harmonized System Code Classification (2026.acl-long)

Copied to clipboard

Qiang Xia, Zijian Zhang, Ao Wang, Wenhan Wang, Xiangyu Wang, Jian Li

Truth as a Trajectory: What Internal Representations Reveal About Large Language Model Reasoning (2026.acl-long)

Challenge:	Harmonized System (HS) code classification is a hierarchically structured and regulation-constrained task, often complicated by short and noisy product descriptions.
Approach:	They propose a knowledge-graph-guided LLM framework that formulates HS classification as a stepwise, regulation-aware reasoning process over an explicit HS knowledge graph.
Outcome:	The proposed framework constrains inference to legally valid paths while producing explicit and traceable reasoning trajectories.

Copied to clipboard

Hamed Damirchi, null Imezadelajara, Ehsan Abbasnejad, Afshar Shamsi, Zhen Zhang, Javen Qinfeng Shi

VL-Calibration: Decoupled Confidence Calibration for Large Vision-Language Models Reasoning (2026.acl-long)

Challenge:	Existing explainability methods for Large Language Models treat hidden states as static points in activation space, but they are saturated with polysemantic features.
Approach:	They propose a framework that shifts analysis from static activations to layer-wise geometric displacement.
Outcome:	The proposed framework outperforms existing explainability methods on commonsense reasoning, question answering, and toxicity detection benchmarks.

Copied to clipboard

Wenyi Xiao, Xinchi XU, Leilei Gan

Don’t Corrupt the Fact: A Trustworthy RAG Watermarking Framework based on Dual Factual Shield (2026.acl-long)

Challenge:	Existing verbalized confidence calibration methods for large vision language models optimize a single holistic confidence score using binary answer-level correctness.
Approach:	They propose a reinforcement learning framework that explicitly decouples confidence into visual and reasoning confidence.
Outcome:	Experiments show that the proposed framework decouples confidence into visual and reasoning confidence while suppressing ungrounded hallucinations while preserving valid perception.

Copied to clipboard

Hao Huang, JiaTang Luo, Ruihua Zhou, Yunpeng Li, Yuling Liu

I2E: From Image Pixels to Actionable Interactive Environments for Text-Guided Image Editing (2026.acl-long)

Challenge:	Existing watermarking methods are fact-agnostic and cause "faithfulness hallucinations" a novel framework to enforce knowledge loyalty is proposed to improve watermarks .
Approach:	They propose a new framework that enforces knowledge loyalty by spoofing terms from retrieved contexts and prompt-based semantic guidance to protect against factual corruption.
Outcome:	The proposed framework reduces the Knowledge Corruption Rate while maintaining its original high security and robustness.

Copied to clipboard

Jinghan Yu, Junhao Xiao, Chenyu Zhu, Jiaming Li, Jia Li, HanMing Deng, Xirui Wang, Guoli Jia, Jianjun Li, Xiang Bai, Bowen Zhou, Zhiyuan Ma

TInR: Exploring Tool-Internalized Reasoning in Large Language Models (2026.acl-long)

Challenge:	Existing text-guided image editing methods rely on end-to-end pixel-level inpainting paradigm . existing models lack such intermediate representations and Reasoning-then-action process .
Approach:	They propose a "Decompose-then-Action" paradigm that revisits image editing as an actionable interaction process within a structured environment.
Outcome:	The proposed paradigm outperforms existing methods in compositional editing tasks.

Copied to clipboard

Qiancheng Xu, Yongqi Li, Fan Liu, Hongru Wang, Min Yang, Wenjie Li

Fine-Grained Analysis of Shared Syntactic Mechanisms in Language Models (2026.acl-long)

Challenge:	Existing methods rely on external tool documentation during reasoning, leading to tool mastery difficulty, tool size constraints, and inference inefficiency.
Approach:	They propose a tool-internalized reasoning framework for unified reasoning and tool usage that integrates external tools into Large Language Models (LLMs) to address these issues, they propose 'tool-internet-based' reasoning.
Outcome:	The proposed method achieves superior performance across in-domain and out-of-domain settings, highlighting its effectiveness and efficiency.

Copied to clipboard

Ryoma Kumon, Hitomi Yanaka

One Persona, Many Cues, Different Results: How Sociodemographic Cues Impact LLM Personalization (2026.acl-long)

Challenge:	Recent advances in language models have demonstrated their ability to process linguistic expressions with complex syntactic structures.
Approach:	They investigate whether language models employ shared neural mechanisms across different constructions by applying causal interpretability methods at a granular level.
Outcome:	The proposed model performance improves on acceptability judgment benchmarks.

Copied to clipboard

Franziska Weeber, Vera Neplenbroek, Jan Batzner, Sebastian Padó

Lost in the Mix: Evaluating LLM Understanding of Code-Switched Text (2026.acl-long)

Challenge:	Prior work has used personas to study biases by relying on a single cue to prompt a persona, such as user names or explicit attribute mentions.
Approach:	They compare six commonly used personacues across seven open and proprietary LLMs on four writing and advice tasks.
Outcome:	The proposed model is based on a persona, a synthetic user profile defined by specific attributes, defined by gender or race.

Copied to clipboard

Amr Mohamed, Yang Zhang, Michalis Vazirgiannis, Guokan Shang

Building LLMs Like LEGO: Two-dimensional Architecture Reassembly of Large Language Models (2026.acl-long)

Challenge:	Code-switching (CSW) is widespread in multilingual communities and increasingly prevalent in online content.
Approach:	They propose a pipeline for producing linguistically grounded CSW variants of established benchmarks across five typologically diverse languages.
Outcome:	The proposed model sets show that inserting non-English tokens into English reduces accuracy on comprehension and reasoning benchmarks, whereas embedding English into non- English contexts often improves it.

Copied to clipboard

Xingyu Wu, Yu Zhou, KC Tan

How Long Reasoning Chains Influence LLMs’ Judgment of Answer Factuality (2026.acl-long)

Challenge:	Existing approaches to LLM reuse treat LLMs as monolithic artifacts.
Approach:	They propose to recompose pretrained large language models as modular building blocks . they propose a chromosome-based architectural encoding and evolutionary optimization .
Outcome:	The proposed model can be recomposed as modular building blocks without training data.

Copied to clipboard

Minzhu Tu, Shiyu Ni, Keping Bi

MMErroR: A Benchmark for Erroneous Reasoning in Vision-Language Models (2026.acl-long)

Challenge:	Large language models (LLMs) are increasingly adopted as scalable judges for open-ended generation, yet how they form judgments remains insufficiently understood.
Approach:	They show that exposing reasoning influences LLM-based judgment . they also show that reasoning fluency and factuality critically shape judgment outcomes .
Outcome:	Empirical results show that the presence of reasoning significantly alters judgment behavior . stronger judges exhibit more selective behavior and achieve higher judgment accuracy .

Copied to clipboard

Yang Shi, Yifeng Xie, Minzhe Guo, Liangsi Lu, Mingxuan Huang, Jingchao Wang, Zhihong Zhu, Boyan Xu, Zhiqi Huang

Learning How and What to Memorize: Cognition-Inspired Two-Stage Optimization for Evolving Memory (2026.acl-long)

Challenge:	Recent advances in vision-language models have improved performance in multi-modal learning.
Approach:	They propose a multi-modal benchmark that embeds a single coherent reasoning error in 1997 samples.
Outcome:	The proposed benchmark is based on a set of 1997 samples embedding a single coherent reasoning error.

Copied to clipboard

Derong Xu, Shuochen Liu, Pengfei Luo, Pengyue Jia, Yingyi Zhang, Yi Wen, Yimin Deng, Wenlin Zhang, Enhong Chen, Xiangyu Zhao, Tong Xu

Challenge:	Existing memory systems rely on static, hand-crafted update rules for personalization, but sparse outcome rewards provide weak supervision, resulting in unstable long-horizon optimization.
Approach:	They propose a memory guideline optimization framework that learns how memory should be organized and what information to update.
Outcome:	The proposed framework learns how memory should be organized and what information to update.

PolitNuggets: Benchmarking Agentic Discovery of Long-Tail Political Facts (2026.acl-long)

Copied to clipboard

Yifei Zhu

Diagnosing and Remedying Representation Deficiencies for Deterministic Reasoning in KGQA (2026.acl-long)

Challenge:	Large Reasoning Models (LRMs) are embedded in agentic frameworks and are under-evaluated.
Approach:	They propose a multilingual benchmark for agentic information synthesis using PolitNuggets . they standardize evaluation with an optimized Supervisor–Searcher multi-agent system .
Outcome:	The proposed model can discover and synthesize "long-tail" facts from dispersed sources.

Copied to clipboard

Gewen Liang, Mufan Xu, Kehai Chen, Wei Wang, Yuwei Wang, Muyun Yang, Tiejun Zhao, Min Zhang

Revealing the Seen, Imagining the Beyond: A Survey of Image-Grounded Chain-of-Thought Reasoning in Multimodal LLMs (2026.acl-long)

Challenge:	Existing approaches to solve non-deterministic reasoning problems in large language models are limited by their complexity and lack of a clear understanding of the problem.
Approach:	They propose a method to diagnose and correct non-deterministic reasoning behaviors in large language models.
Outcome:	The proposed method outperforms baselines and WebQSP benchmarks on the widely used WebQ SP and CWQ benchmarks.

Copied to clipboard

Qihua Dong, Yitian Zhang, Huimin Zeng, Yizhou Wang, Jianglin Lu, Kuo Yang, Yun Fu

Learning Faster with Better Tokens: Parameter-Efficient Vocabulary Adaptation for Specialized Text Summarization (2026.acl-long)

Challenge:	Recent advances in Multimodal Large Language Models (MLLMs) have shifted visual reasoning from tool-calling to end-to-end perceptionreasoning.
Approach:	They synthesize the emerging paradigm of Image-Grounded Chain-of-Thought (IG-CoT) they propose a method-centric taxonomy covering prompting, supervised fine-tuning, and reinforcement learning .
Outcome:	The proposed model is based on a method-centric taxonomy and benchmarks.

Copied to clipboard

Gunjan Balde, Soumyadeep Roy, Mainack Mondal, Niloy Ganguly

Adversarial Metric Learning for Fine-Grained Emotion Classification (2026.acl-long)

Challenge:	a new approach to adapt generalist models to expert domains is needed to overcome this problem.
Approach:	They propose a parameter-efficient domain adaptation approach that combines vocabulary adaptation with pretraining for LLM-based text summarization.
Outcome:	The proposed approach reduces training time by 35-55% over continual pretraining and reduces parameter counts up to 37% w.r.t expansion-only methods.

Copied to clipboard

Junfan Chen, Sizhe Wu, Richong Zhang, Chunming Hu

OMIBench: Benchmarking Olympiad-Level Multi-Image Reasoning in Large Vision-Language Models (2026.acl-long)

Challenge:	Recent advances in fine-grained emotion classification relied on contrastive learning with hard-pair mining.
Approach:	They propose an adversarial metric learning framework that replaces fixed similarity metrics with a learnable metric family and trains representations to remain discriminative under worst-case similarity distortions.
Outcome:	The proposed framework trains a pairwise discriminator to maximally confuse two hard pair types while training the encoder to remain discriminative under worst-case similarity distortions.

Copied to clipboard

Qiguang Chen, Chengyu Luan, Jiajun Wu, Qiming Yu, Yi Yang, Yizhuo Li, Jingqi Tong, Xiachong Feng, Libo Qin, Wanxiang Che

Authorship Attribution in Multilingual Machine-Generated Texts (2026.acl-long)

Challenge:	Existing multimodal reasoning benchmarks for large vision-language models emphasize single-image analysis and fail to exploit contextual information across multiple images.
Approach:	They propose a benchmark to evaluate Olympiad-level reasoning when evidence is distributed over multiple images.
Outcome:	The proposed model outperforms existing models on bi-image Olympiads and Gemini-3-Pro on multimodal Olympiad-level reasoning tasks.

Copied to clipboard

Lucio La Cava, Dominik Macko, Robert Moro, Ivan Srba, Andrea Tagarelli

Layer-Wise High-Impact Parameter Ratio Optimization in Post-Training Quantization for Large Language Models (2026.acl-long)

Challenge:	Large Language Models (LLMs) have reached human-like fluency and coherence, but distinguishing machine-generated text from human-written content becomes increasingly difficult.
Approach:	They propose a problem of multilingual authorship attribution (AA) that involves attributing texts to human or multiple LLM generators across diverse languages.
Outcome:	The proposed method can be adapted to multilingual settings, but still has significant limitations and challenges.

Copied to clipboard

Cuong Pham, Anh Dung Hoang, Cuong C. Nguyen, Trung Le, Gustavo Carneiro, Thanh-Toan Do

On the Role of Discriminative Models in Generative Relation Extraction (2026.acl-long)

Challenge:	Existing methods to quantize large language models suffer from significant accuracy loss at low bit-widths due to high-impact parameters.
Approach:	They propose a quadratic optimization framework that quantizes high-impact parameters to moderate bit-widths while quantizing low bit-wideths.
Outcome:	The proposed framework preserves high-impact parameters while preserving memory usage.

Copied to clipboard

Guozheng Li, Peng Wang, Zijie Xu, Jing Zhou, Jiajun Liu, Ziyu Shang

Speculative End-Turn Detector for Efficient Speech Chatbot Assistant (2026.acl-long)

Challenge:	Existing methods for relation extraction (RE) are discriminative and generative . previous studies show that discriminative models can support generative RE .
Approach:	They propose a framework that leverages discriminative models to produce a top-k set of candidate relations and integrates this knowledge into generative models via in-context or prompt learning.
Outcome:	The proposed framework achieves state-of-the-art on five widely used RE benchmarks.

Copied to clipboard

Hyunjong Ok, Suho Yoo, Jaeho Lee

TwiUSD: A Benchmark Dataset and Structure-Aware LLM Framework for User Stance Detection (2026.acl-long)

Challenge:	Spoken dialogue systems with large language models struggle with end-turn detection . this limitation often leads to premature or delayed responses, disrupting the flow of spoken conversations.
Approach:	They propose a dataset for end-turn detection that uses a lightweight GRU-based model and a high-performance Wav2vec-based system to make a more challenging classification of distinguishing turn ends from mere pauses.
Outcome:	The proposed framework significantly improves real-time ETD accuracy while keeping computations low.

Copied to clipboard

Fuqiang Niu, Zini Chen, Zhiyu Xie, Hu Huang, Qing Liao, Qianlong Wang, Genan Dai, Bowen Zhang

Breaking the Impasse: Dual-Scale Evolutionary Policy Training for Social Language Agents (2026.acl-long)

Challenge:	Existing benchmarks for political user-level stance detection rely on noisy heuristics or distant supervision.
Approach:	They propose a large-scale, expert-annotated benchmark for political user-level stance detection with explicit social network structure that integrates user content and followee signals.
Outcome:	The proposed framework outperforms baselines in terms of quality and reliability.

Copied to clipboard

Minzheng Wang, Run Luo, Yanbo Wang, Zichen Liu, Yuqiao Tan, Tao Tan, Nan Xu, Lu Wang, Wenji Mao

Render-of-Thought: Rendering Textual Chain-of-Thought as Images for Visual Latent Reasoning (2026.acl-long)

Challenge:	Reinforcement Learning with Verifiable Rewards (RLVR) is effective for closed-ended tasks, but it is not applicable to open-ended social language games.
Approach:	They propose a method that uses a time-scaled evolutionary perception mechanism to detect impasse by quantifying dual-scale value baseline divergence alongside match entropy.
Outcome:	Experiments on multiple social language games show that the proposed method outperforms baselines and avoids policy degeneration.

Copied to clipboard

Yifan Wang, Shiyu Li, Peiming Li, Xiaochen Yang, Zheng Wei, Yang Tang

Legal Judgment Prediction: A Reflection on the State of the Art (2026.acl-long)

Challenge:	Recent work on Chain-of-Thought prompting imposes substantial computational overhead . lack of supervision obscures the analyzability of the latent reasoning chain.
Approach:	They propose a framework to render latent reasoning chain into images, making latent rationale explicit and traceable.
Outcome:	The proposed framework achieves 3-4 token compression and substantial inference acceleration compared to explicit CoT prompting.

Copied to clipboard

Yi Feng, Chuanyi Li, Vincent Ng

Challenge:	Legal Judgment Prediction (LJP) involves predicting judgment outcomes based on fact descriptions of cases.
Approach:	They propose to use argument trees to build automated legal judgment prediction systems that are trustworthy and can be used to predict cases.
Outcome:	The proposed model outperforms competitors on standard evaluation datasets and enables pluralistic values to be naturally expressed.

GRAD: Generalizing RAG Adaptation with Decoding (2026.acl-long)

Copied to clipboard

Youngwon Lee, Seung-won Hwang, Zhewei Yao, Yuxiong He

SHARP: Self-adaptive Harmful Category-aware Prompt Generation for Black-box Jailbreaking (2026.acl-long)

Challenge:	Using GRAD, we can steer Retrieval-augmented generation objectives without retraining large language models.
Approach:	They propose an adaptive decoding-time framework that keeps the base generator fixed and composes small, objective-specific guidance at inference.
Outcome:	The proposed framework improves accuracy with favorable latency across public benchmarks and private settings with no in-domain labels while reliably activating helpful objectives and suppressing harmful ones, adaptively to tasks.

Copied to clipboard

Yingjie Xue, Xingyou Xia, Jun Zhang, Yunbo Cao, Dengpan Ye, Guotong Geng, Fei Li

Unveiling the Limits of Large Language Models in Inferring Pragmatic Meaning from Non-Verbal Responses (2026.acl-long)

Challenge:	Existing methods for jailbreak ignore the semantic differences between categories of harmful questions, leading to inconsistent success rates and reduced overall attack effectiveness.
Approach:	They propose a category-aware jailbreak framework that incorporates the semantic category of harmful questions into prompt generation.
Outcome:	The proposed framework improves attack success rates and category alignment and achieves better cross-category robustness compared to the state-of-the-art (SOTA) baselines.

Copied to clipboard

Sugyeong Eo, Heuiseok Lim

Does RLVR Extend Reasoning Boundaries? Investigating Capability Expansion in Vision-Language Models (2026.acl-long)

Challenge:	Existing studies have focused mainly on LLMs' comprehension of verbal behavior, with non-verbal behavior considered only in conjunction with verbal responses.
Approach:	They present the first systematic evaluation of LLMs’ ability to infer pragmatic meaning in dialogue consisting solely of non-verbal responses.
Outcome:	The proposed model fails to capture non-verbal intent and has accuracy dropping by 60% compared to verbal ones.

Copied to clipboard

Minghe Shen, Zhuo Zhi, Chonghan Liu, Shuo Xing, Zhengzhong Tu, Che Liu

Perception, Understanding and Reasoning: A Multimodal Benchmark for Video Fake News Detection (2026.acl-long)

Challenge:	Recent studies suggest that RLVR amplifies behaviors inherent to the pre-training distribution rather than inducing new capabilities.
Approach:	They propose a framework for RLVR that extends the spatial reasoning boundary . they use a mapping framework where the difficulty is precisely regulated by path length and number of turns .
Outcome:	The proposed framework extends the spatial reasoning boundary on two real-world navigation benchmarks.

Copied to clipboard

Cui Yakun, Peng Qi, Fushuo Huo, Hang Du, Weijie Shi, Juntao Dai, Zhenghao Zhu, Sirui Han

RV-HATE: Reinforced Multi-Module Voting for Implicit Hate Speech Detection (2026.acl-long)

Challenge:	Existing video fake news detection benchmarks focus on the detection accuracy, while failing to provide fine-grained assessments for the entire detection process.
Approach:	They propose a process-oriented video fake news detection benchmark that evaluates MLLMs' perception, understanding, and reasoning capabilities in VFND.
Outcome:	The proposed model achieves sota performance on video fake news detection tasks.

Copied to clipboard

Yejin Lee, Hyeseon An, Yo-Sub Han

Cross-Prompt Automated Essay Scoring of Multiple Traits: Making Sense of the State of the Art (2026.acl-long)

Challenge:	a new framework for hate speech detection addresses implicit hate speech by tailoring the detection process to dataset-specific attributes.
Approach:	They propose a framework to account for the dataset-specific characteristics of hate speech datasets.
Outcome:	The proposed framework improves detection accuracy and provides interpretable insights into the distinctive features of each dataset.

Copied to clipboard

Shengjie Li, Vincent Ng

JARVIS or Ultron? A Survey on the Safety and Security Threats of Computer-Using Agents (2026.acl-long)

Challenge:	despite recent progress in cross-prompt essay scoring, there is little analysis of what makes a state-of-the-art cross-propert scorer work well.
Approach:	They propose to apply transductive learning to cross-prompt scoring for the first time . they propose to train a model that can offer good performance when applied to unseen prompts .
Outcome:	The proposed model could be used in the rarely-studied classroom setting without additional training data.

Copied to clipboard

Ada Chen, Yongjiang Wu, Junyuan Zhang, Jingyu Xiao, Shu Yang, Jen-tse Huang, Kun Wang, Wenxuan Wang, Shuai Wang

Challenge:	Recent advances in computer-using agents have created new safety and security risks . despite the impressive capabilities of CUAs, there are still significant security risks.
Approach:	They propose a systematization of knowledge on the safety and security threats of Computer-Using Agents.
Outcome:	The proposed framework provides a framework for assessing the safety and security risks of computer-using agents.

Rose-SQL: Role-State Evolution Guided Structured Reasoning for Multi-Turn Text-to-SQL (2026.acl-long)

Copied to clipboard

Le Zhou, Feng Yao, Fengcai Qiao, Bo Xu, Fangyuan Wang, Boyan Xu

Why Do More Experts Fail? A Theoretical Analysis of Model Merging (2026.acl-long)

Challenge:	Existing approaches to multi-turn Text-to-SQL tasks rely on unstable APIs or expensive fine-tuning.
Approach:	They propose a training-free framework that leverages small-scale LRMs through in-context learning to enable accurate context-dependent parsing.
Outcome:	The proposed framework outperforms in-context learning baselines at the 4B scale and surpasses state-of-the-art models at the 8B and 14B scales.

Copied to clipboard

Zijing Wang, Xingle Xu, YongKang Liu, Yiqun Zhang, Peiqin Lin, Shi Feng, Daling Wang, Xiaocui Yang, Hinrich Schuetze

Challenge:	Existing methods for model merging struggle to maintain performance gains as the number of merged models increases.
Approach:	They propose a Reparameterized Heavy-Tailed method to extend the merged model’s coverage and enhance performance.
Outcome:	The proposed method extends the merged model’s coverage and enhances performance on 19 benchmarks, including knowledge-intensive and general-purpose tasks.

Combining Distantly Supervised Models with In Context Learning for Monolingual and Cross-Lingual Relation Extraction (2026.acl-long)

Copied to clipboard

Vipul Kumar Rathore, Malik Hammad Faisal, Parag Singla, null Mausam

Explaining Sources of Uncertainty in Automated Fact-Checking (2026.acl-long)

Challenge:	Existing Distantly Supervised Relation Extraction models rely on task-specific training, but their integration with in-context learning (ICL) using large language models (LLMs) remains underexplored.
Approach:	They propose a framework for distantly supervised relation extraction that uses a trained DSRE model to identify the top-k candidate relations for a given test sentence and a dynamic exemplar retrieval strategy that extracts reliable, sentence-level exemplars from training data.
Outcome:	The proposed framework achieves 20 F1 points gains in English and 17 F1 point gains on Indic languages over previous models and naive prompting baselines.

Copied to clipboard

Jingyi Sun, Greta Warren, Irina Shklovski, Isabelle Augenstein

TRACE: A Corpus of Team Creative Discussions (2026.acl-long)

Challenge:	Existing methods to explain model uncertainty as numbers or hedges do not reveal which evidence conflicts cause the uncertainty, leaving users unable to resolve disagreements.
Approach:	They propose a plug-and-play framework that generates natural-language explanations of model uncertainty grounded in conflicting/agreeing evidence.
Outcome:	The proposed framework generates explanations that more faithfully track model uncertainty and better align with the model’s fact-checking decisions than span-agnostic explanation prompting.

Copied to clipboard

Yixuan Jiang, Tiancheng Hu, Jose Hernandez-Orallo, David Stillwell, Luning Sun

Evaluating Structure-Aware Retrieval and Safety in Statute-Centric Legal QA (2026.acl-long)

Challenge:	Existing studies on team creativity lack the ability to observe discussion dynamics from the perspective of natural language processing (NLP) Standard approaches capture participants' perceptions rather than actual behavior.
Approach:	They propose a corpus of 309 group discussions from 103 teams across six creative problem-solving tasks.
Outcome:	The proposed analysis reveals that large teams explore more broadly but converge less effectively while team diversity shapes participation patterns more than discussion content.

Copied to clipboard

Kyubyung Chae, Jewon Yeom, Jeongjae Park, Seunghyun Bae, Ijun Jang, Hyunbin Jin, Jinkwan Jang, Taesup Kim

PRISMA: Preference-Reinforced Self-Training Approach for Interpretable Emotionally Intelligent Negotiation Dialogues (2026.acl-long)

Challenge:	Legal QA benchmarks focus on case law, overlooking statute-centric regulatory reasoning . relevant evidence is distributed across hierarchically linked documents, creating statutory retrieval gap .
Approach:	They propose a structure- and safety-aware benchmark for statute-centric legal QA . the benchmark assesses whether models can retrieve hierarchically fragmented evidence .
Outcome:	The proposed benchmark evaluates whether models can retrieve hierarchically fragmented evidence and safely abstain when statutory context is insufficient.

Copied to clipboard

Prajwal Vijay Kajare, Priyanshu Priya, Bikash Santra, Asif Ekbal

SkMTEB: Slovak Massive Text Embedding Benchmark and Model Adaptation (2026.acl-long)

Challenge:	Emotion plays a pivotal role in shaping negotiation outcomes, influencing trust, cooperation, and long-term relationships.
Approach:	They propose an Emotion-aware Negotiation Strategy-informed Chain-of-Thought reasoning mechanism which mimics human negotiation by perceiving, understanding, using, and managing emotions.
Outcome:	The proposed system generates interpretable emotions and improves negotiation effectiveness on job interviews and resource allocation datasets.

Copied to clipboard

Marek Suppa, Andrej Ridzik, Daniel Hládek, Natália Kňažeková, Viktória Ondrejová

A Linguistics-Aware LLM Watermarking via Syntactic Predictability (2026.acl-long)

Challenge:	Slovak embeddings are core infrastructure for semantic search, retrieval-augmented generation (RAG), clustering, and classification.
Approach:	They propose a MTEB-style text embedding benchmark for Slovak, a low-resource West Slavic language . they use 31 datasets across 7 task types to evaluate the performance of the models .
Outcome:	The proposed model achieves competitive performance with proprietary APIs while remaining locally deployable for RAG . the model is based on 31 datasets across 7 task types and is 4 the depth of existing benchmark for Slovak .

Copied to clipboard

Shinwoo Park, Hyejin Park, Hyeseon An, Yo-Sub Han

Selective Contrastive Learning For Gloss Free Sign Language Translation (2026.acl-long)

Challenge:	a central challenge remains balancing text quality against detection robustness.
Approach:	They propose a framework that aligns watermark strength with linguistic degrees of freedom . they use part-of-speech models to weaken the signal in grammatically constrained contexts .
Outcome:	The proposed framework outperforms existing methods in linguistic indeterminacy tests on languages . it weakens the watermark strength in grammatically constrained contexts and strengthens it in contexts with greater linguistic flexibility.

Copied to clipboard

Chang Hao Lai, Rui Zhao, Xuewen Zhong, Jinsong Su, Yidong Chen

SciCustom: A Framework for Custom Evaluation of Scientific Capabilities in Large Language Models (2026.acl-long)

Challenge:	Recent SLT systems adopt CLIP-like Vision-Language pretraining, but the random in-batch contrast provides few, batch-dependent negatives.
Approach:	They propose a method to train sign video-text similarity over a time period of 3 months . they use a random in-batch contrast strategy to track negative video- text similarity .
Outcome:	The proposed system improves sign language translation by focusing on challenging negatives . the results show that the random in-batch contrast provides few negatives and noisy supervision .

Copied to clipboard

Yiyang Gu, Junwei Yang, Junyu Luo, Ye Yuan, Bin Feng, Yingce Xia, Shufang Xie, Kaili Liu, Bohan Wu, Qi Shi, Haoran Li, Beier Xiao, Zhiping Xiao, Xiao Luo, Weizhi Zhang, Philip S. Yu, Zequn Liu, Ming Zhang

NSF-SciFy: Mining the NSF Awards Database for Scientific Claims (2026.acl-long)

Challenge:	Existing evaluations of large language models fail to reflect fine-grained capabilities . existing benchmarks are manually curated or domain-generic, limiting scalability and alignment with real use cases.
Approach:	They propose a framework that allows custom construction of benchmarks from large-scale scientific data to evaluate application-specific scientific capabilities in LLMs.
Outcome:	The proposed framework reveals fine-grained differences in scientific capabilities that standard benchmarks overlook . it allows custom construction of benchmarks from large-scale scientific data to evaluate application-specific capabilities in LLMs.

Copied to clipboard

Delip Rao, Weiqiu You, Eric Wong, Chris Callison-Burch

Learning Invariant Modality Representation for Robust Multimodal Learning from a Causal Inference Perspective (2026.acl-long)

Challenge:	NSF-SciFy contains 2.8 million claims from 400,000 abstracts spanning all science and mathematics disciplines.
Approach:	They propose to use a dataset to extract scientific claims from National Science Foundation award abstracts and to use it to refine language models.
Outcome:	The proposed method improves non-technical abstract generation, claim extraction, and investigation proposal extraction tasks while maintaining high precision and lower recall.

Copied to clipboard

Sijie Mai, Shiqin Han

Multi-Task Representation Alignment on Language Understanding: A Mutual Information Perspective (2026.acl-long)

Challenge:	Existing approaches to multimodal affective computing learn spurious correlations from training data rather than genuine causal relationships, harming generalization under distribution shifts or noisy modalities.
Approach:	They propose a causal modality-invariant representation framework that separates each modality into ‘causal invariant’ and ‘environment-specific spurious representation’ from a modal inference perspective.
Outcome:	Experiments on multiple multimodal benchmarks show that the proposed framework achieves state-of-the-art performance.

Copied to clipboard

Dou Hu, Lingwei Wei, Hongjiang Xiao, Songlin Hu, Yuan Zhang

EconProver: Towards More Economical Test-Time Scaling for Automated Theorem Proving (2026.acl-long)

Challenge:	Existing approaches to multitask learning fail to address task interference issues . Existing methods focus on task balancing or probabilistic modeling but fail to learn sufficient representations for all target tasks.
Approach:	They propose a multi-task representation alignment framework to achieve task-specific alignment and self-alignment on shared representations from a mutual information perspective.
Outcome:	The proposed framework outperforms 13 representative MTL methods under label-noisy and data-constrained conditions.

Copied to clipboard

Mukai Li, Linfeng Song, Zhenwen Liang, Jiahao Xu, Shansan Gong, Qi Liu, Haitao Mi, Dong Yu

Thinking-Based Non-Thinking: Solving the Reward Hacking Problem in Training Hybrid Reasoning Models via Reinforcement Learning (2026.acl-long)

Challenge:	Large Language Models (LLMs) have recently advanced the field of Automated Theorem Proving (ATP) Existing cost analyses regulate only the number of sampling passes, ignoring the substantial disparities in sampling costs.
Approach:	They propose to integrate two complementary methods into a unified EconRL pipeline to increase pass rates under constrained sampling passes.
Outcome:	The proposed method reduces token usage and sample passes while maintaining the original performance.

Copied to clipboard

Siyuan Gan, Jiaheng Liu, Boyan Wang, Tianpei Yang, Runqing Miao, Yuyao Zhang, Fanyu Meng, Junlan Feng, Linjian Meng, Jing Huo, Yang Gao

LLM-Based Multi-Agent Systems for Clinical Workflows: A Survey of AI Hospitals (2026.acl-long)

Challenge:	Existing work on large reasoning models (LRMs) focuses on using reinforcement learning (RL) to train hybrid reasoning models that automatically decide whether to engage in thinking or not based on the complexity of the query.
Approach:	They propose to use reinforcement learning to train hybrid reasoning models that automatically decide whether to engage in thinking or not based on the complexity of the query.
Outcome:	The proposed model reduces token usage by around 50%$ compared to DeepSeek-R1-Distill-Qwen-1.5B/7B and DeepScaleR-1.5b, while significantly improving accuracy.

Copied to clipboard

Zonghai Yao, Hong yu

Challenge:	Large Language Models (LLMs) are moving from isolated text generation toward agentic work inside clinical workflows.
Approach:	They propose a workflow-level taxonomy for LLM-based multi-agent systems for clinical and healthcare workflows . they propose integration readiness levels, task-level instrumentation requirements and recurring workflow failure modes as a practical framework for comparing, evaluating and deploying clinical LLM agents and AI hospitals.
Outcome:	The proposed systems should be compared at the workflow level, rather than only by model components or end-task accuracy.

Mind’s Eye: A Benchmark of Visual Abstraction, Transformation and Composition for Multimodal LLMs (2026.acl-long)

Copied to clipboard

Rohit Sinha, Aditya Sanjiv Kanade, Sai Srinivas Kancheti, Vineeth N. Balasubramanian, Tanuja Ganu

EverMemOS: A Self-Organizing Memory Operating System for Structured Long-Horizon Reasoning (2026.acl-long)

Challenge:	Existing evaluations of multimodal large language models (MLLMs) have demonstrated compelling visual understanding in recent years.
Approach:	They propose a multimodal large language model with eight visuo-cognitive tasks inspired by classic human intelligence tests organized under a novel A–R–T taxonomy: Abstraction, Relation, and Transformation.
Outcome:	The proposed frameworks are based on eight visuo-cognitive tasks inspired by human intelligence tests and organized under a novel A–R–T taxonomy: Abstraction, Relation, and Transformation.

Copied to clipboard

Chuanrui Hu, Xingze Gao, Zuyi Zhou, Dannong Xu, Yi Bai, Xintong Li, Hui Zhang, Tong Li, Chong Zhang, Lidong Bing, Yafeng Deng

Interpretable Coreference Resolution Evaluation Using Explicit Semantics (2026.acl-long)

Challenge:	Existing memory systems for LLMs store isolated records and retrieve fragments . Existing systems store isolated data and fragments, limiting their ability to consolidate evolving experience and resolve conflicts.
Approach:	They propose an engram-inspired memory operating system that implements an 'engram'-inspired lifecycle for computational memory.
Outcome:	Experiments on LoCoMo, LongMemEval, and PersonaMeM-v2 show that EverMemeOS outperforms state-of-the-art methods on memory-augmented reasoning tasks.

Copied to clipboard

Bruno Gatti, Giuliano Martinelli, Roberto Navigli

SURE or Not? Investigating Semantic Understanding in Dense Retrieval Models (2026.acl-long)

Challenge:	Existing evaluation methods for coreference resolution are limited by semantic and contextual information.
Approach:	They propose a semantically-enhanced evaluation framework for coreference resolution that assigns semantic labels to nominal mentions and propagates them to entire coreference clusters.
Outcome:	The proposed framework uncovers systematic weaknesses obscured by standard metrics.

Copied to clipboard

Lingdi Kong, Xuanang Chen, Ben He, Le Sun

When High Accuracy Hides Poor Calibration: Rethinking Confidence Evaluation in Transformer-Based Text Classification with Balanced Brier Score (2026.acl-long)

Challenge:	Dense retrieval models have been successful in a number of applications but it is unclear whether they truly understand semantics.
Approach:	They propose a benchmark for semantic understanding in dense retrieval that characterizes semantic precision, semantic abstraction and semantic equivalence along three dimensions.
Outcome:	The proposed model characterizes semantic understanding in dense retrieval along three dimensions: semantic precision, semantic abstraction, and semantic equivalence.

Copied to clipboard

Guilherme Fonseca, Gabriel Prenassi, Washington Cunha, Leonardo Chaves Dutra da Rocha, Marcos André Gonçalves

Doc-V*: Coarse-to-Fine Interactive Visual Reasoning for Multi-Page Document VQA (2026.acl-long)

Challenge:	Existing evidence for TC under fine-tuning is limited.
Approach:	They propose a calibration method that balances the contribution of correct and incorrect predictions within confidence bins.
Outcome:	The proposed calibration measures show that the models are overconfident even when miscalibrated . the proposed calibration methods challenge calibration assessment practices and provide a more reliable alternative for evaluating confidence quality in Transformer-based TC.

Copied to clipboard

Yuanlei Zheng, Pei Fu, Hang Li, Ziyang Wang, Yuyi Zhang, Wenyu Ruan, Xiaojin Zhang, Zhongyu Wei, Zhenbo Luo, Jian Luan, Wei Chen, Xiang Bai

Privacy-R1: Privacy-Aware Multi-LLM Agent Collaboration via Reinforcement Learning (2026.acl-long)

Challenge:	Existing OCR-free approaches to document visual question answering are brittle and passive.
Approach:	They propose an OCR-free agentic framework that casts multi-page DocVQA as sequential evidence aggregation.
Outcome:	The proposed framework outperforms open-source and proprietary models in five benchmarks and improves out-of-domain performance by 47.9% over baseline.

Copied to clipboard

Zheng Hui, Yijiang River Dong, Sanhanat Sivapiromrat, Ehsan Shareghi, Nigel Collier

Reinforcement Learning for Diffusion LLMs via Energy-Based Gibbs Alignment (2026.acl-long)

Challenge:	Prior approaches to rewriting large language models shatters linguistic coherence and removes privacy-sensitive information.
Approach:	They propose a framework that trains an agent to dynamically route text chunks . it implicitly distinguishes between replaceable Personally Identifiable Information (PII) and task-critical PII .
Outcome:	The proposed framework achieves state-of-the-art on the privacy-utility frontier . it trains an agent to dynamically route text chunks, learning a policy that balances privacy leakage and task performance.

Copied to clipboard

Yijia Fan, Jing Yang, Mingyu Liu, Kaitong Cai, Jian Wang, Keze Wang, Jusheng Zhang

CxMP: A Linguistic Minimal-Pair Benchmark for Evaluating Constructional Understanding in Language Models (2026.acl-long)

Challenge:	Diffusion Large Language Models (dLLMs) offer parallel decoding and bidirectional context modeling . aligning dLLms with reinforcement learning (RL) remains a challenge .
Approach:	They propose a variational framework that reformulates RL for dLLMs as a distribution matching problem.
Outcome:	The proposed framework reformulates RL for dLLMs as a distribution matching problem.

Copied to clipboard

Miyu Oba, Saku Sugawara

Evaluating Temporal Consistency in Multi-Turn Language Models (2026.acl-long)

Challenge:	Understanding language acquisition in language models remains an open question, yet many benchmarks focus on grammatical acceptability, with far less attention to interpreting meanings conveyed by grammatological forms.
Approach:	They propose a benchmark to evaluate constructional understanding in language models using a controlled minimal-pair.
Outcome:	The proposed benchmarks show that understanding of constructions develops more slowly and remains limited even in large language models (LLMs).

Copied to clipboard

Yash Kumar Atri, Steven L. Johnson, Thomas Hartvigsen

Synthia: Scalable Grounded Persona Generation from Social Media Data (2026.acl-long)

Challenge:	Language models are increasingly deployed in interactive settings where users reason about facts over time . we study temporal scope stability, the ability to preserve, override, or transfer time-scoped factual context across dialogue turns.
Approach:	They propose a diagnostic benchmark to isolate temporal scope stability in controlled multi-turn interactions.
Outcome:	The proposed model can preserve, override, or transfer time-scoped factual context across dialogue turns.

Copied to clipboard

Vahid Rahimzadeh, Erfan Moosavi Monazzah, Mohammad Taher Pilehvar, Yadollah Yaghoobzadeh

Challenge:	Persona-driven large language models (LLMs) are increasingly used in computational social science, yet their validity critically depends on the fidelity of the underlying personas.
Approach:	They propose a persona-generation framework that grounds LLM-generated personas in real social-media posts while delegating narrative construction to language models.
Outcome:	The proposed framework outperforms state-of-the-art methods for most demographics across different dimensions while maintaining interaction graph structure among personas grounded in real social network users.

From Recognition to Reasoning: Benchmarking and Enhancing MLLMs on Real-World Receipt Document Understanding (2026.acl-long)

Copied to clipboard

Yandi Wang, Libin Zhan, Ziwei Huang, Tiancheng Luo, Yuxuan Jiang, Wang Dong, Leilei Gan, Jun Chen

LogicEval: A Systematic Framework for Evaluating Automated Repair Techniques for Logical Vulnerabilities in Real-World Software (2026.acl-long)

Challenge:	Existing models for visual information extraction suffer from limitations in scale and realism . ReceiptBench is a large-scale, human-annotated benchmark for receipts .
Approach:	They propose a large-scale, human-annotated benchmark for visual information extraction . the method organizes information extraction into four hierarchical sub-tasks .
Outcome:	The proposed method surpasses proprietary models on complex reasoning tasks.

Copied to clipboard

Syed Md Mukit Rashid, Abdullah Al Ishtiaq, Kai Tu, Yilu Dong, Tianwei Wu, Ali Ranjbar, Tianchang Yang, Najrin Sultana, Shagufta Mehnaz, Syed Rafiul Hussain

Mind the Pause: Disfluency-Aware Objective Tuning for Multilingual Speech Correction with LLMs (2026.acl-long)

Challenge:	Existing automated program-repair techniques focus on repairing memory corruptions, but they struggle with logical vulnerabilities because of their limited semantic understanding of the code and its expected behavior.
Approach:	They evaluated a dataset of 122 logical vulnerabilities and a framework to evaluate patches for logical weaknesses.
Outcome:	The proposed framework evaluates both traditional and LLM-based approaches for addressing real-world logical vulnerabilities.

Copied to clipboard

Deepak Kumar, Baban Gain, Asif Ekbal

BioTool: A Comprehensive Tool-Calling Dataset for Enhancing Biomedical Capabilities of Large Language Models (2026.acl-long)

Challenge:	Spontaneous speech is rarely fluent, and disfluencies can degrade readability and reliability . a sequence tagger first marks disfluent tokens, and these signals guide instruction fine-tuning .
Approach:	They propose a multilingual correction pipeline where a sequence tagger first marks disfluent tokens . they add a contrastive learning objective that penalizes the reproduction of disfluency tokens.
Outcome:	The proposed model improves readability and reliability of ASR transcripts in three languages . disfluencies can cause misinterpretations, incoherent responses, poor user experience .

Copied to clipboard

Xin Gao, Ruiyi Zhang, Meixi Du, Peijia Qin, Pengtao Xie

COMPASS: A Framework for Evaluating Organization-Specific Policy Alignment in LLMs (2026.acl-long)

Challenge:	despite the success of large language models, their performance in highly specialized domains remains unsatisfactory.
Approach:	They propose a biomedical tool-calling dataset designed for fine-tuning LLMs . the dataset contains 34 frequently used tools from the NCBI, Ensembl, and UniProt databases .
Outcome:	The proposed dataset outperforms commercial LLMs on biomedical domains.

Copied to clipboard

Dasol Choi, DongGeon Lee, Brigitta Jesica Kartono, Helena Berndt, Taeyoun Kwon, Joonwon Jang, Haon Park, Hwanjo Yu, Minsuk Kahng

Paper2Rebuttal: A Multi-Agent Framework for Transparent Author Response Assistance (2026.acl-long)

Challenge:	Large language models are being rapidly adopted across a wide range of domains, including healthcare, finance, and the public sector.
Approach:	They propose a framework to evaluate whether large language models comply with policies . they apply COMPASS to eight diverse industry scenarios to validate models .
Outcome:	The proposed framework evaluates whether LLMs comply with allowlist and denylist policies.

Copied to clipboard

Qianli Ma, Chang Guo, Zhiheng Tian, Siyu Wang, Jipeng Xiao, Yuanhao Yue, Zhipeng Zhang

Challenge:	Current approaches to writing effective rebuttals are limited by the direct-to-text generation problem . authors must accurately decipher reviewer intent while ensuring every response is firmly anchored in verifiable manuscript details.
Approach:	They propose a framework that reframes rebuttal generation as an evidence-centric planning task.
Outcome:	The proposed framework outperforms baselines in coverage, faithfulness, and strategic coherence.

Data Efficient RLVR via Off-Policy Influence Guidance (2026.acl-long)

Copied to clipboard

Erle Zhu, Dazhi Jiang, Yuan Wang, Xujun Li, Jiale Cheng, Yuxian Gu, Yilin Niu, Aohan Zeng, Jie Tang, Minlie Huang, Hongning Wang

CEBC: Conformal Evidence-Bounded Control for Low-Hallucination Vision–Language Generation (2026.acl-long)

Challenge:	Existing data selection methods for RLVR are heuristic-based, lacking theoretical guarantees and generalizability.
Approach:	They propose an off-policy influence estimation method that approximates data influence using offline trajectories.
Outcome:	The proposed method reduces the computational cost of policy rollouts and improves storage and computation efficiency.

Copied to clipboard

Ashish Mishra, Tarun Kumar, Arpit Shah, Suparna Bhattacharya, Martin Foltin

Dual Alignment Between Language Model Layers and Human Sentence Processing (2026.acl-long)

Challenge:	Existing mitigation approaches reduce hallucinated object mentions at the cost of degraded generation quality or require expensive retraining and task-specific supervision.
Approach:	They propose a lightweight framework for low-hallucination vision–language generation . it uses evidence-bounded minimal editing to revise or suppress unsupported referenced entities .
Outcome:	The proposed framework reduces hallucinations while maintaining or improving quality metrics.

Copied to clipboard

Tatsuki Kuribayashi, Alex Warstadt, Yohei Oseki, Ethan Gotlieb Wilcox

Rethinking Composed Image Retrieval Evaluation: A Fine-Grained Benchmark from Image Editing (2026.acl-long)

Challenge:	Existing studies have demonstrated both the successes and limitations of accurate predictability estimation by modern LMs in cognitive modeling.
Approach:	They propose to use internal layers to better estimate human cognitive effort observed in syntactic ambiguity processing in English.
Outcome:	The proposed models can be modeled using surprisal from early layers of large language models (LLMs) this raises the question whether such advantages extend to more syntactically challenging constructions, where surprised estimates underestimate human cognitive effort.

Copied to clipboard

Tingyu Song, Yanzhao Zhang, Mingxin Li, Zhuoning Guo, Dingkun Long, Pengjun Xie, Siyue Zhang, Yilun Zhao, Shu Wu

EDSD: Entropy-Driven Design for Faster Speculative Decoding (2026.acl-long)

Challenge:	Composed Image Retrieval (CIR) is a complex task in multimodal understanding . current CIR benchmarks lack a robust evaluation pipeline and limited query categories .
Approach:	They construct a fine-grained CIR benchmark that allows for precise control over modification types and content.
Outcome:	The proposed benchmark covers 5,000 high-quality queries structured across five main categories and fifteen subcategories.

Copied to clipboard

Longkai Cheng, Ximing Wang, Jiangcai Zhu, Kailai Shao, Chao Chen, Haixiang Hu

MUTANT: A Recipe for Multilingual Tokenizer Design (2026.acl-long)

Challenge:	Existing methods for speculative decoding incur substantial training overhead to mitigate information misalignment between autoregressive draft model training and decoding.
Approach:	They propose an Entropy-Driven Speculative Decoding framework that uses entropy as a unified, interpretable signal for both draft model training and architectural design.
Outcome:	Experiments on seven large language models show that EDSD improves training efficiency by 24.8% and increases acceptance length by 4.0% compared to state-of-the-art methods.

Copied to clipboard

Souvik Rana, Ashish Kulkarni, Arul Menezes, Chandra Khatri, Shubham Agarwal

FactCorrector: A Graph-Inspired Approach to Long-Form Factuality Correction of Large Language Models (2026.acl-long)

Challenge:	Subword tokenization schemes such as Byte Pair Encoding (BPE) are widely adopted, but their effectiveness in multilingual settings remains understudied.
Approach:	They propose a multilingual tokenizer that produces linguistically coherent tokens for multilingual LLMs.
Outcome:	The proposed tokenizer improves fertility score by 39.5% over LLaMA4 and 18% over Sutra.

Copied to clipboard

Javier Carnerero-Cano, Massimiliano Pronesti, Radu Marinescu, Tigran T. Tchrakian, James Barry, Jasmina Gajcin, Yufang Hou, Alessandra Pascale, Elizabeth M. Daly

MADE: A Living Benchmark for Multi-Label Text Classification with Uncertainty Quantification of Medical Device Adverse Events (2026.acl-long)

Challenge:	Large language models (LLMs) often produce factually incorrect responses.
Approach:	They propose a new method that adapts across domains without retraining and leverages structured feedback to generate a correction.
Outcome:	The proposed method outperforms baseline methods on a VELI5 dataset and several popular long-form factuality datasets.

Copied to clipboard

Raunak Agarwal, Markus A. Wenzel, Simon Baur, Jonas Zimmer, George Harvey, Jackie Ma

Overcoming Copyright Barriers in Corpus Distribution Through Non-Reversible Hashing (2026.acl-long)

Challenge:	Existing MLTC benchmarks are saturated and may be affected by training data contamination.
Approach:	They propose a machine learning benchmark based on medical device adverse event reports . they establish baselines across 20 encoder- and decoder-only models .
Outcome:	The proposed benchmarks show that small fine-tuned models achieve the strongest head-to-tail accuracy while maintaining competitive UQ.

Copied to clipboard

Arthur Amalvy, Vincent Labatut, Xavier Bost, Hen-Hsen Huang

Challenge:	Annotated corpora are crucial in the field of natural language processing, but are difficult to exchange among researchers.
Approach:	They propose a method to lawfully share the annotations of any sequential copyrighted corpus.
Outcome:	The proposed method is robust to reasonable divergences in the version of the copyrighted data owned by the user.

UniSRM: A Unified Speech Reward Model for Reasoning-Based Fine-grained Assessment (2026.acl-long)

Copied to clipboard

Yuanyuan Wang, Dongchao Yang, Yayue Deng, Zhiyong Wu, Steven Y. Guo, Helen M. Meng, Xixin Wu

DPEPO: Diverse Parallel Exploration Policy Optimization for LLM-based Agents (2026.acl-long)

Challenge:	Existing methods for speech generation rely on subjective, expensive judgments . Existing models only cover a narrow set of scenarios and only provide limited coverage .
Approach:	They propose a unified speech reward model that can support multi-dimensional, interpretable reward signals with reliable reasoning.
Outcome:	The proposed model can support multi-dimensional, interpretable reward signals with reliable reasoning.

Copied to clipboard

JunShuo Zhang, Chengrui Huang, Feng Guo, Zihan Li, Ke Shi, Menghua Jiang, Jiguo Yu, Shuo Shang, Shen Gao

TEA-Bench: A Systematic Benchmarking of Tool-enhanced Emotional Support Dialogue Agent (2026.acl-long)

Challenge:	Existing approaches to large language model (LLM) agents that follow the sequential "reason-then-act" paradigm suffer from limited exploration and incomplete environmental understanding as they interact with only a single environment per step.
Approach:	They propose a paradigm that enables an agent to interact with multiple environments simultaneously and share cross-trajectory experiences.
Outcome:	The proposed paradigm achieves state-of-the-art (SOTA) success rates while maintaining comparable efficiency to strong sequential baselines.

Copied to clipboard

Xingyu Sui, Yanyan Zhao, Yulin Hu, Jiahe Guo, Weixiang Zhao, Bing Qin

An Information-Theoretic Foundation for the Subregular Hierarchy (2026.acl-long)

Challenge:	Existing ESC systems focus on affective support in text-only settings, ignoring how external tools can enable factual grounding and reduce hallucination in multi-turn emotional support.
Approach:	They propose a benchmark for evaluating tool-augmented agents in ESC with realistic emotional scenarios and an MCP-style tool environment.
Outcome:	The proposed benchmarks show that tool augmentation improves emotional support quality and reduces hallucination, but weaker models benefit only marginally.

Copied to clipboard

Mai Phan Quoc Hung, Khanh Nguyen Quoc, Đoàn Minh Luong, Duong Thu Ngan, Duong Thi Phuong Thao, Tuan Do

ReGATE: Learning Faster and Better with Fewer Tokens in MLLMs (2026.acl-long)

Challenge:	Subregular theory posits that phonological patterns in natural languages occupy restricted region of formal hierarchy . phonology patterns in SL, SP, and Tier-based Strictly Local (TSL) languages are restricted . experimental evidence demonstrates humans fail to learn patterns outside these subregulate classes .
Approach:	They propose a subregular hypothesis that phonological patterns in natural languages occupy a restricted region of the formal language hierarchy.
Outcome:	The proposed framework offers a framework for understanding computational restrictions on natural language phonology.

Copied to clipboard

Chaoyu Li, Yogesh Kulkarni, Pooyan Fazli

From Words to Pixels: A Comprehensive Survey on Large Language Models in Visual Segmentation (2026.acl-long)

Challenge:	Existing efficiency methods mainly target inference via token reduction or merging, offering limited benefits during training.
Approach:	They propose an adaptive token pruning method that uses a teacher-student framework to prune MLLMs to reduce inference costs.
Outcome:	The proposed method matches the peak accuracy of standard training on MVBench up to 2 faster, using only **38% of the tokens.

Copied to clipboard

Yizhou Wang, Mang Tik Chiu, Lingzhi Zhang, Xuan Shen, Sohrab Amirghodsi, Yun Fu

Red-Bandit: Test-Time Adaptation for LLM Red-Teaming via Bandit-Guided LoRA Experts (2026.acl-long)

Challenge:	Visual segmentation with instruction has been a challenging task for many years . large language models and large multimodal models have spurred a new wave of research .
Approach:	They review recent works in LLM-based visual segmentation and analyze their architectural innovations, training strategies, and benchmark performance.
Outcome:	The present study reviews the most recent works in LLM-driven visual segmentation . it identifies key challenges and promising future directions .

Copied to clipboard

Christos Ziakas, Nicholas Loo, Nishita Jain, Alessandra Russo

Graph-Based Alternatives to LLMs for Human Simulation (2026.acl-long)

Challenge:	Existing approaches to audit Large Language Models (LLMs) lack mechanisms to efficiently adapt to model-specific vulnerabilities at inference.
Approach:	They propose a red-teaming framework that adapts online to identify and exploit model failure modes under distinct attack styles.
Outcome:	The proposed framework outperforms state-of-the-art methods on AdvBench and HarmBench, while generating more human-readable adversarial prompts (lower perplexity).

Copied to clipboard

Joseph Suh, Suhong Moon, Serina Chang

Minimal Free Resolution Guided Adaptive Tree Reasoning (2026.acl-long)

Challenge:	Large language models (LLMs) are a popular approach for simulating human behaviors, yet it remains unclear if they are necessary for all simulation tasks.
Approach:	They propose a graph neural network that can match or surpass strong LLMs for close-ended simulations.
Outcome:	The proposed model outperforms strongest LLM-based methods across three datasets and three evaluation settings.

Copied to clipboard

Dezhao Tang, Meihan Liu, Yulai Tong, Guan Yuan, Qiuyan Yan

GrAInS: Gradient-based Attribution for Inference-Time Steering of LLMs and VLMs (2026.acl-long)

Challenge:	Existing approaches to building dynamic reasoning trees rely on manual decomposition patterns and subproblems.
Approach:	They propose a hierarchical reasoning framework based on MFR theory that supports adaptive reasoning trees and reliable error correction within a single LLM.
Outcome:	The proposed framework significantly reduces logical errors and improves reasoning accuracy compared to the Chain-of-Thought, Decompose–Analyze–Rethink and Tree-of–Though.

Copied to clipboard

Duy Nguyen, Archiki Prasad, Elias Stengel-Eskin, Mohit Bansal

DiNO: Disinformation Narrative Observer (2026.acl-long)

Challenge:	Existing methods for fine-tuning large language models often ignore token-level causal influence and underutilize model logits.
Approach:	They propose a novel approach that uses a gradient-based approach to identify influential tokens and construct directional steering vectors based on their contribution to preferred over dispreferred outputs.
Outcome:	The proposed approach outperforms fine-tuning and prior steering methods on both LLM and VLM tasks without degrading fluency or general capabilities.

Copied to clipboard

Witold Sosnowski, Arkadiusz Modzelewski, Kinga Skorupska, Adam Wierzbicki

Agentic Very Long Video Understanding (2026.acl-long)

Challenge:	Disinformation is an escalating global threat, making it essential to understand its content, dissemination, and evolution.
Approach:	They propose a method to extract disinformation narratives from news articles . they evaluated how well their topics and stances aligned with a recognized disinformation dataset.
Outcome:	The proposed method outperforms other narrative mining methods in analyzing disinformation narratives.

Copied to clipboard

Aniket Rege, Arka Sadhu, Yuliang Li, Kejie Li, Ramya Korlakai Vinayak, Yuning Chai, Yong Jae Lee, Hyo Jin Kim

OLA: Output Language Alignment in Code-Switched LLM Interactions (2026.acl-long)

Challenge:	Existing methods for understanding video over long periods of time are limited . eGAgent system provides tools for structured search and reasoning over entity scene graphs .
Approach:	They propose a system that can interpret and recall video over days or weeks . they use entity scene graphs to equip a planning agent with tools for structured search and reasoning .
Outcome:	The proposed method achieves state-of-the-art performance on EgoLifeQA and Video-MME-long datasets.

Copied to clipboard

Juhyun Oh, Haneul Yoo, Faiz Ghifari Haznitrama, Alice Oh

Generalizing Trust: Weak-to-Strong Trustworthiness in Language Models (2026.acl-long)

Challenge:	Existing LLMs that code-switch between languages often fail to align with user's implicit language expectation, causing responses to be in undesired languages.
Approach:	They propose a benchmark to evaluate LLMs’ Output Language Alignment in code-switched interactions.
Outcome:	The proposed benchmark evaluates LLMs’ Output Language Alignment in code-switched interactions.

Copied to clipboard

Lillian Sun, Martin Pawelczyk, Zhenting Qi, Aounon Kumar, Himabindu Lakkaraju

Reasoning Is Not All You Need: Examining LLMs for Multi-Turn Mental Health Conversations (2026.acl-long)

Challenge:	Recent studies have highlighted weak-to-strong generalization, where a strong model trained only on a weak model’s labels surpasses the weak model in task performance.
Approach:	They propose two fundamental fine-tuning strategies that leverage trustworthiness regularization during the fine-uning of the weak model and the weak-to-strong transfer to improve trustworthy.
Outcome:	The proposed models show that they can generalize robustness, fairness, and privacy better when trained on weak models than models trained on strong models.

Copied to clipboard

Mohit Chandra, Siddharth Sriraman, Harneet Singh Khanuja, Yiqiao Jin, Munmun De Choudhury

BatonVoice: An Operationalist Framework for Enhancing Controllable Speech Synthesis with Linguistic Intelligence from LLMs (2026.acl-long)

Challenge:	Existing evaluation frameworks focus on diagnostic accuracy and win-rates and often overlook alignment with patient-specific goals, values, and personalities required for meaningful conversations.
Approach:	They propose a framework for synthetically generating realistic, multi-turn mental health sensemaking conversations and a dataset to examine their models in healthcare settings.
Outcome:	The proposed framework synthesizes a dataset comprising over 2,200 patient–LLM conversations and evaluates them using human-centric criteria.

Copied to clipboard

Yue Wang, Ruotian Ma, Xingyu Chen, Zhengliang Shi, Morunliu Yang, Wanshun Chen, Huang Liu, Jiadi Yao, Xin He, Qu Yang, Qingxuan Jiang, Fanghua Ye, Juntao Li, Zhaopeng Tu, Xiaolong Li, Liefeng Bo, Min Zhang

Monotonic Scaffolding as a Diagnostic Lens for Legal Reasoning in LLMs (2026.acl-long)

Challenge:	Existing approaches often fail to leverage the linguistic intelligence of Large Language Models (LLMs) Existing models lack the ability to follow text instructions for controllable Text-to-Speech (TTS).
Approach:	They propose a framework where an LLM acts as a conductor, understanding user instructions and generating a textual plan - explicit vocal features.
Outcome:	The proposed model outperforms open- and closed-source models in speech synthesis and achieves zero-shot cross-lingual generalization.

Copied to clipboard

Pedro Calais, Janderson Santos, Anisio Lacerda, Wagner Meira Jr.

Jailbreak-Zero: A Path to Pareto Optimal Red Teaming for Large Language Models (2026.acl-long)

Challenge:	Modern evaluation of Legal QA systems is shifting from terminal accuracy toward process-aware analyses of model reasoning.
Approach:	They propose a diagnostic framework grounded in monotonic scaffolding where language models receive gold-standard, case-relevant information across stages aligned with the canonical legal framework FIRAC.
Outcome:	The proposed framework evaluates 3,123 Brazilian Bar Exam questions . it shows that terminal accuracy overestimates legal reasoning competence .

Copied to clipboard

Kai Hu, Abhinav Aggarwal, Mehran Khodabandeh, David Zhang, Eric Hsin, Li Chen, Ankit Jain, Matt Fredrikson, Akash Bharadwaj

Factual Retrieval in LLMs Is a Redundant, Distributed and Non-Contiguous Process (2026.acl-long)

Challenge:	Existing approaches to red teaming are based on example-based evaluation, where a static list of specific prompts is used to define and measure "unsafe content"
Approach:	They propose a new automated red teaming framework that shifts from example-based to policy-based evaluation that focuses on risk coverage, semantic diversity, and fidelity.
Outcome:	The proposed method achieves superior, human-readable attacks against open-source and proprietary models even for unseen safety policies.

Copied to clipboard

Hail Hochman, Natalie Shapira, Yoav Goldberg

What About the Scene With the Hitler Reference? HAUNT: A Framework to Probe LLMs’ Self-consistency in Closed Domains Via Adversarial Nudge (2026.acl-long)

Challenge:	Existing work posits that factual knowledge is stored at the last entity token position, but the precise mechanics of how facts are retrieved from model parameters remain unclear.
Approach:	They propose an iterative patching protocol to identify a minimal subset of layers necessary for attribute retrieval.
Outcome:	The proposed method shows that models possess multiple paths for the same entity and fact, highlighting a high degree of redundancy in attribute computation.

Copied to clipboard

Arka Dutta, Sujan Dutta, Rijul Magu, Soumyajit Datta, Munmun De Choudhury, Ashiqur R. KhudaBukhsh

GitChameleon 2.0: Evaluating AI Code Generation Against Python Library Version Incompatibilities (2026.acl-long)

Challenge:	Claude exhibits strong resilience, while GPT and Grok demonstrate moderate resilience . open models fall short significantly, while proprietary models exhibit weak resilience compared to open models .
Approach:	They propose a framework for stress testing factual fidelity in large language models in the presence of adversarial nudges.
Outcome:	The proposed model is robust to adversarial nudges in two closed domains.

Copied to clipboard

Diganta Misra, Nizar Islah, Victor May, Brice Rauby, Zihan Wang, Justine Gehring, Antonio Orvieto, Muawiz Sajjad Chaudhary, Eilif B. Muller, Irina Rish, Samira Ebrahimi Kahou, Massimo Caccia

Knowledge without Wisdom: Measuring Misalignment between LLMs and Intended Impact (2026.acl-long)

Challenge:	Existing code evolution benchmarks lack execution-based evaluation for generating code compliant with specific library versions.
Approach:	They propose a new Python code completion problem that evaluates the ability of large language models to perform version-conditioned code generation.
Outcome:	The proposed benchmarks show that state-of-the-art systems can perform version-conditioned code generation with high success rates.

Copied to clipboard

Michael Hardy, Yunsung Kim

Challenge:	a recent study shows that large language models excel on benchmarks that operationalize knowledge.
Approach:	They compare LLM alignment on benchmarks, downstream tasks and intended impact . they find that inter-model behaviors on disparate tasks correlate higher than expert human behaviors on target tasks .
Outcome:	The proposed methods show that LLMs perform poorly on learning tasks . the results show that they are poorly aligned with downstream measures of teaching quality .

Apertus: Democratizing Open and Compliant LLMs for Global Language Environments (2026.acl-long)

Copied to clipboard

Alejandro Hernández-Cano, Alexander Hägele, Allen Hao Huang, Angelika Romanou, Antoni-Joan Solergibert, Barna Pásztor, Bettina Messmer, Dhia Garbaya, Eduard Frank Ďurech, Ido Hakimi, Juan Garcia Giraldo, Mete Ismayilzada, Negar Foroutan, Skander Moalla, Tiancheng Chen, Vinko Sabolčec, Yixuan Xu, Michael Aerni, Badr AlKhamissi, Inés Altemir Marinas, Mohammad Hossein Amani, Matin Ansaripour, Ilia Badanin, Harold Benoit, Emanuela Boros, Nicholas John Browning, Fabian Bösch, Maximilian Böther, Niklas Canova, Camille Challier, Clément Charmillot, Jonathan Coles, Jan Milan Deriu, Arnout Devos, Lukas Drescher, Daniil Dzenhaliou, Maud Ehrmann, Dongyang Fan, Simin Fan, Silin Gao, Miguel Gila, María Grandury, Diba Hashemi, Alexander Miserlis Hoyle, Jiaming Jiang, Mark Klein, Andrei Kucharavy, Anastasiia Kucherenko, Frederike Lübeck, Roman Machacek, Theofilos Ioannis Manitaras, Andreas Marfurt, Kyle Matoba, Simon Matrenok, Henrique Mendonça, Fawzi Roberto Mohamed, Syrielle Montariol, Luca Mouchel, Sven Najem-Meyer, Jingwei Ni, Gennaro Oliva, Matteo Pagliardini, Elia Palme, Andrei Panferov, Léo Paoletti, Marco Passerini, Ivan Pavlov, Auguste Poiroux, Kaustubh Ponkshe, Nathan Ranchin, Javier Rando, Mathieu Sauser, Jakhongir Saydaliev, Mukhammadali Sayfiddinov, Marian Schneider, Stefano Schuppli, Marco Scialanga, Andrei Semenov, Kumar Shridhar, Raghav Singhal, Anna Sotnikova, Alexander Sternfeld, Ayush Kumar Tarun, Paul Teiletche, Jannis Vamvas, Xiaozhe Yao, Hao Zhao, Alexander Ilic, Ana Klimovic, Andreas Krause, Caglar Gulcehre, David Rosenthal, Elliott Ash, Florian Tramèr, Joost VandeVondele, Livio Veraldi, Martin Rajman, Thomas C. Schulthess, Torsten Hoefler, Antoine Bosselut, Martin Jaggi, Imanol Schlag

Reframing Responsibility: Framing-Aware Event Causality Identification (2026.acl-long)

Challenge:	Apertus is a fully open suite of large language models (LLMs) designed to address responsibility shortcomings in today’s open model ecosystem, namely data responsibility and global representation.
Approach:	They propose to release a fully open suite of large language models (LLMs) that address data responsibility and global representation shortcomings in today’s open model ecosystem.
Outcome:	The proposed model is pretrained on openly available data and suppresses verbatim recall of data while retaining task performance.

Copied to clipboard

Jin Zhao, Jiayi Yao, Xinrui Hu, Nianwen Xue

Adaptive Instruction Composition for Automated LLM Red-Teaming (2026.acl-long)

Challenge:	Causal explanations in political narratives are often framed and contested.
Approach:	They propose a framing-aware extension of ECI that models causal explanations as structured claims including responsibility targets, evaluative frams, source type, and epistemic modality.
Outcome:	The proposed model enables quantitative analysis of divergent causal attribution across narratives.

Copied to clipboard

Jesse Zymet, Andy Luo, Swapnil Shinde, Sahil Wadhwa, Emily Chen

CHOIR: Harmonizing Structured Persona Diversity for Robust Collaborative LLM Reasoning (2026.acl-long)

Challenge:	Adaptive instruction composition is a framework for red-teaming that combines crowdsourced texts with random combinations to optimize effectiveness and diversity.
Approach:	They propose a framework that combines crowdsourced texts according to an adaptive mechanism trained to optimize effectiveness with diversity.
Outcome:	The proposed framework outperforms random combination on effectiveness and diversity metrics even under model transfer.

Copied to clipboard

Xiangjue Dong, Cong Wang, Maria Teleki, Millennium Bismay, Ruihong Huang, James Caverlee

TabReX: Tabular Referenceless eXplainable Evaluation (2026.acl-long)

Challenge:	Persona-assigned Large Language Models can be useful for personalized, context-aware reasoning.
Approach:	They propose a framework that harmonizes demographic perturbations into a unified prediction by balancing agreement and divergence among counterfactual personas.
Outcome:	The proposed framework improves reasoning performance even when base personas are suboptimal.

Copied to clipboard

Tejas Anvekar, Junha Park, Aparna Garimella, Vivek Gupta

DREAM-S: Speculative Decoding with Searchable Drafting and Target-Aware Refinement for Multimodal Generation (2026.acl-long)

Challenge:	Existing metrics for evaluating the quality of tables generated by large language models flatten tables into text, ignoring structure or relying on fixed references that limit generalization.
Approach:	They propose a reference-less framework for evaluating tabular generation via graph-based reasoning . tabReX converts source text and generated tables into canonical knowledge graphs .
Outcome:	The proposed framework provides a high correlation with expert rankings and stable under harder perturbations.

Copied to clipboard

Zining Liu, Yunhai Hu, Tianhua Xia, BO Bao, Eric Sather, Vithursan Thangarasa, Sai Qian Zhang

Revisiting Non-Verbatim Memorization in Large Language Models: The Role of Entity Surface Forms (2026.acl-long)

Challenge:	Speculative decoding (SD) has proven to be effective for autoregressive generation in large language models (LLMs), however its application to vision-language models (VLMs) remains relatively unexplored.
Approach:	They propose a Speculative Decoding framework for vision-language models that integrates a neural architecture search framework and target-aware supernet training to identify optimal interaction strategies.
Outcome:	DREAM-S achieves 3.85 speedup compared to baselines on well-established vision-language models.

Copied to clipboard

Yuto Nishida, Naoki Shikoda, Yosuke Kishinami, Ryo Fujii, Makoto Morishita, Hidetaka Kamigaito, Taro Watanabe

ChatAnime: Towards User-Centered Emotional Support in LLM-based Virtual Character Chat (2026.acl-long)

Challenge:	Entity-based QA is a common framework for analyzing non-verbatim memorization, but typically query each entity using a single canonical surface form.
Approach:	They propose a dataset that pairs Wikidata factual triples with categorized entity surface forms . they examine surface-conditioned factual memorization and find that prediction outcomes change when only the entity surface form is changed.
Outcome:	The proposed dataset shows that large language models memorize factual knowledge when only the subject entity surface form is changed.

Copied to clipboard

Lanlan Qiu, Sophia Xiao Pu, Yeqi Feng, Wenchang Gao, Tianxing He

DyBBT: Dynamic Balance via Bandit-inspired Targeting for Dialog Policy with Cognitive Dual Systems (2026.acl-long)

Challenge:	Existing research focuses on character consistency in fictional or game-based scenarios . ESRP framework is designed to align role-playing with real-world user scenarios based on emotional needs.
Approach:	They propose a framework to align role-playing with real-world user scenarios and emotional needs.
Outcome:	The proposed framework aligns role-playing with real-world user scenarios and emotional needs.

Copied to clipboard

Shuyu Zhang, Yifan Wei, Jialuo Yuan, Xinru Wang, Yanmin Zhu, Yujie Liu, Bin Li

SciPedia: Unlocking the Value of Scientific Data for Pre-training (2026.acl-long)

Challenge:	Task oriented dialog systems often rely on static exploration strategies that do not adapt to dynamic dialog contexts.
Approach:	They propose a dialog policy learning framework that formalizes the exploration challenge through a structured cognitive state space C.
Outcome:	The proposed framework achieves SOTA performance in success rate, efficiency, and generalization.

Copied to clipboard

Yiwei Qin, Zhen Huang, Tiantian Mi, Weiye Si, Qipeng Guo, Siyuan Feng, Pengfei Liu

From Word Sequences to Behavioral Sequences: Adapting Modeling and Evaluation Paradigms for Longitudinal NLP (2026.acl-long)

Challenge:	High-quality scientific data is critical for advancing LLMs, yet academic literature remains underutilized.
Approach:	They construct a large-scale raw scientific corpus but identify a critical Learnability Gap . they develop a multi-stage pipeline featuring content cleaning and pedagogical augmentation .
Outcome:	The proposed approach boosts average performance by +2.12 (3B) and +2.95 (7B) on in-domain tasks.

Copied to clipboard

Adithya V Ganesan, Vasudha Varadarajan, Oscar Kjell, Whitney Ringwald, Scott M. Feltman, Benjamin J. Luft, Roman Kotov, Ryan L. Boyd, H. Andrew Schwartz

Challenge:	a longitudinal model for NLP relies on document-level evaluation to map isolated instances of language to an outcome.
Approach:	They propose a longitudinal model that aligns evaluation splits to generalization over people and time . they propose integrating a sequence inputs to incorporate history by default .
Outcome:	The proposed model improves on a dataset of 17k daily diary transcripts paired with PTSD symptom severity from 238 participants.

Why Mean Pooling Works: Quantifying Second-Order Collapse in Text Embeddings (2026.acl-long)

Copied to clipboard

Tomomasa Hara, Hiroto Kurita, Masaaki Imaizumi, Kentaro Inui, Sho Yokoi

Self-Guided Alignment: Adaptive Preference Sensing for Multi-Objective Generation (2026.acl-long)

Challenge:	Text embeddings are used across a wide range of NLP tasks, including retrieval-augmented generation.
Approach:	They propose a metric to quantify the collapse induced by mean pooling and a simple metric for measuring how often it occurs in real models and texts.
Outcome:	The proposed metric measures how often the collapse occurs in real models and texts.

Copied to clipboard

Ning Wang, Zhanyang Liu, Taotao Zhou, Xinrui Zhang, Zongru Shao, Haojie Zhou

ReasMark: A Robust Watermark for Attributing LLM Reasoning Under Knowledge Distillation Attacks (2026.acl-long)

Challenge:	Existing approaches to align LLMs with diverse human values rely on ground-truth scores . existing approaches implicitly approximate an average-user preference, thereby failing to capture heterogeneity of human values or accommodate conflicting user needs.
Approach:	They propose a framework that transforms passive reward dependency into an intrinsic adaptive sensing capability.
Outcome:	The proposed framework outperforms state-of-the-art models in multiple model scales and improves preference alignment.

Copied to clipboard

Peizhuo Lv, Ruihua Zhou, Yunpeng Li, Ruigang Liang, Xingshuo Han, XiaoFeng Wang, Wei Dong, Yuling Liu

One Single Hub Text Breaks CLIP: Identifying Vulnerabilities in Cross-Modal Encoders via Hubness (2026.acl-long)

Challenge:	Existing reasoning-enhanced large language models fail to provide reliable attribution of reasoning behavior once it is transferred through knowledge distillation.
Approach:	They propose to embed a reasoning-length gap in a model by querying a target domain and training a local student to imitate its outputs.
Outcome:	et al. show that ReasMark outperforms baselines while preserving task utility.

Copied to clipboard

Hiroyuki Deguchi, Katsuki Chousa, Yusuke Sakai

Awakening Dormant Experts:Counterfactual Routing to Mitigate MoE Hallucinations (2026.acl-long)

Challenge:	et al., 2010) show that hub embeddings are close to many unrelated examples in high-dimensional embeddable spaces . cross-modal encoders that project different modalities into a shared space are useful for cross-module applications .
Approach:	They propose a method for identifying the hub embedding and its corresponding hub text . they use images to evaluate cross-modal encoders that project different modalities into a shared space .
Outcome:	The proposed method can identify a single hub embedding and its corresponding hub text . it achieves comparable or higher similarity scores than human-written reference captions in many images .

Copied to clipboard

Wentao Hu, Yanbo Zhai, Xiaohui Hu, Mingkuan Zhao, Shanhong yu, Xue Liu, Kaidong Yu, Shuangyong Song, Xuelong Li

Characterizing and Evaluating Working Emotion Vocabularies in Multilingual Large Language Models (2026.acl-long)

Challenge:	Sparse Mixture-of-Experts models are vulnerable to hallucinations, authors say . static Top-k routing leaves "specialist experts" under-prioritized for specific tokens .
Approach:	They propose a training-free inference framework to awaken dormant experts . they propose 'counterfactual routing' to shift computational resources from syntax-dominant to knowledge-intensive layers .
Outcome:	Experiments show that CoR improves factual accuracy by 3.1% without increasing the inference budget.

Copied to clipboard

Nicholas Deas, Ivan Ernesto Perez Mejia, Ellie Yang, Kathleen McKeown

HalluCitation Matters: Revealing the Impact of Hallucinated References with 300 Hallucinated Papers in ACL Conferences (2026.acl-long)

Challenge:	Prior work evaluating emotion and affective understanding in large language models rely on predetermined label sets or focus on a singular evaluation task.
Approach:	They examine the ability of multilingual language models to predict any term used by an author to label their own feelings or emotions.
Outcome:	The proposed models perform poorly on three different tasks in English and Spanish.

Copied to clipboard

Yusuke Sakai, Hidetaka Kamigaito, Taro Watanabe

Solve-Detect-Verify: Inference-Time Scaling with Flexible Generative Verifier (2026.acl-long)

Challenge:	citations that do not correspond to any existing work are a serious concern to scientific reliability and credibility.
Approach:	They analyze papers published at ACL, NAACL, and EMNLP in 2024 and 2025 . they identify 300 papers with at least one HalluCitation, most of which were published in 2025.
Outcome:	The authors analyze papers published at ACL, NAACL, and EMNLP in 2024 and 2025 . they find that nearly 300 papers contain at least one HalluCitation, most of which were published in 2025.

Copied to clipboard

Jianyuan Zhong, Zeju Li, Zhijian Xu, Xiangyu Wen, Kezhi Li, Qiang Xu

Challenge:	Recent advances in Large Language Models (LLMs) have enhanced capabilities in complex reasoning through step-by-step trace generation.
Approach:	They propose a generative verifier that dynamically allocates compute between rapid fast thinking and deliberative slow thinking.
Outcome:	The proposed solution outperforms GenPRM-32B on ProcessBench while requiring 2.3x fewer TFLOPS and 15x less training data.

SPARD: Self-Paced Curriculum for RL Alignment via Integrating Reward Dynamics and Data Utility (2026.acl-long)

Copied to clipboard

Xuyang Zhi, Peilun Zhou, Chengqiang Lu, Hang Lv, Yiwei Liang, Rongyang Zhang, Yan Gao, null Yiwu, Yao Hu, Hongchao Gu, Defu Lian, Hao Wang, Enhong Chen

Challenge:	Large Language Models (LLMs) are shifting the focus from single verifiable tasks toward complex, open-ended real-world scenarios.
Approach:	They propose a framework that automatically adjusts reward weights and data importance to synchronize learning intent with data utility for optimal performance.
Outcome:	The proposed framework improves model capabilities across all domains and scales.

Lost in Simulation: LLM-Simulated Users are Unreliable Proxies for Human Users in Agentic Evaluations (2026.acl-long)

Copied to clipboard

Preethi Seshadri, Samuel Cahyawijaya, Ayomide Odumakinde, Sameer Singh, Seraphina Goldfarb-Tarrant

Toward Robust Evaluation for Multilingual Grammatical Error Correction: Can Large Language Models Replace Human References? (2026.acl-long)

Challenge:	Agentic benchmarks rely on LLM-simulated users to evaluate agent performance . however, the robustness, validity, and fairness of this approach remain unexamined .
Approach:	They investigate whether LLM-simulated users are reliable proxies for real human users . they find that agent success rates vary up to 9 percentage points across different LLMs .
Outcome:	The results show that simulated users underestimate success on challenging tasks while miscalibrate performance on moderately difficult tasks.

Copied to clipboard

Alla Rozovskaya, Dan Roth

Cell-Based Representation of Relational Binding in Language Models (2026.acl-long)

Challenge:	Prior work has shown that using aclosest-gold reference yields more accurate performance estimates, but producing such references for each system individually is costly.
Approach:	They propose a method for generating closest-gold references by prompting a large language model with system outputs and a standard reference-based evaluations show weak or no correlation.
Outcome:	The proposed method outperforms state-of-the-art models on 14 languages across 14 benchmarks.

Copied to clipboard

Qin Dai, Benjamin Heinzerling, Kentaro Inui

LOTUS: Evolving Multimodal Unlearning via Hyperbolic Entailment and Lorentz Transport (2026.acl-long)

Challenge:	Recent work has found evidence that Large Language Models (LLMs) are able to track entities across discourse . however, the mechanism by which they bind entities, relations, and attributes remains unclear .
Approach:	They propose a low-dimensional cell-based binding representation for relational binding . they also show that context-specific CBR representations are related by translation vectors .
Outcome:	The proposed model encodes a low-dimensional cell-based binding representation (CBR) a translation vector in activation space enables cross-context transfer, the study shows .

Copied to clipboard

Zekun Wang, Jingjie Zeng, Yingxu Li, Hongfei Lin, Liang Yang

KARL: Reinforcement Learning for LLM Agents on Multi-Turn Knowledge-Intensive Agentic Tasks (2026.acl-long)

Challenge:	Existing unlearning methods suffer from a geometric mismatch, causing catastrophic forgetting or unsafe substitution.
Approach:	They propose a framework for surgical semantic pruning within the Lorentz manifold.
Outcome:	Experiments on MLLMU-Bench show that LOTUS significantly outperforms baselines while maintaining general utility.

Copied to clipboard

Xueqiao Sun, Xiao Liu, Bowen Lv, Hanchen Zhang, Bohao Jing, Zehan Qi, Yifan Xu, Yuxiao Dong, Jie Tang

FIGMA: Towards FIne-Grained Music retrievAl (2026.acl-long)

Challenge:	Large Language Models have shown remarkable potential as autonomous agents, but their effectiveness in knowledge-intensive tasks remains limited by passive knowledge utilization.
Approach:	They propose a framework that enables LLM agents to dynamically explore structured knowledge sources through multi-turn interactions.
Outcome:	The proposed framework outperforms existing retrieval-augmented approaches on knowledge graph and database tasks while maximizing tool-use behaviors end-to-end.

Copied to clipboard

Nishit Anand, Ashish Seth, Sreyan Ghosh, Dinesh Manocha, Ramani Duraiswami

MathSight: A Benchmark Exploring Have Vision-Language Models Really Seen in University-Level Mathematical Reasoning? (2026.acl-long)

Challenge:	Existing music retrieval models fail to retrieve fine-grained musical attributes when using coarse semantic queries.
Approach:	They propose a multi-view contrastive architecture that captures high-level semantic context and fine-grained musical attributes within a unified representation space.
Outcome:	The proposed method outperforms existing CLAP-based music retrieval models on multiple benchmarks.

Copied to clipboard

Yuandong Wang, Yao Cui, Yuxin Zhao, Zhen Yang, Yangfu Zhu, Zhenzhou Shao

TokDrift: When LLM Speaks in Subwords but Code Speaks in Grammar (2026.acl-long)

Challenge:	Existing benchmarks rarely isolate how much visual information contributes to reasoning . a growing collection of benchmarks has catalyzed rapid progress in multimodal reasoning - but how much it contributes remains unclear .
Approach:	They propose a university-level multimodal mathematical reasoning benchmark to quantify the effect of visual input.
Outcome:	The proposed benchmark disentangles and quantifies the effect of visual input on multimodal reasoning models.

Copied to clipboard

Yinxi Li, Yuntian Deng, Pengyu Nie

REVEALER: Reinforcement-Guided Visual Reasoning for Element-Level Text-Image Alignment Evaluation (2026.acl-long)

Challenge:	Large language models (LLMs) for code rely on subword tokenizers learned from mixed natural language text and programming language code but driven by statistics rather than grammar.
Approach:	They propose a framework that applies semantic-preserving rewrite rules to create code variants differing only in tokenization.
Outcome:	The proposed framework can create code variants differing only in tokenization . the findings highlight the need for grammar-aware tokenization for future code LLMs.

Copied to clipboard

FuLin Shi, Wenyi Xiao, Leilei Gan, Liang Ding, null Binchen

ReasonAny: Incorporating Reasoning Capability to Any Model via Simple and Effective Model Merging (2026.acl-long)

Challenge:	Existing methods for text-to-image alignment evaluation rely on coarse-grained metrics or static Question Answering pipelines that lack fine-grounded interpretability and struggle to reflect human preferences.
Approach:	They propose a reinforcement-guided visual reasoning framework for element-level text-to-image alignment evaluation.
Outcome:	The proposed framework achieves state-of-the-art results on four benchmarks and surpasses the strong proprietary Gemini 3 Pro and Training-based baselines.

Copied to clipboard

Junyao Yang, Chen Qian, Wen Shen, Yong Liu, Jing Shao, Dongrui Liu

Do LLM Agents Mirror Socio-Cognitive Effects in Power-Asymmetric Conversations? (2026.acl-long)

Challenge:	Existing models with long chain-of-thought reasoning lack reasoning depth and domain-specific utility.
Approach:	They propose a model merging framework that integrates reasoning with domain-specific task models.
Outcome:	The proposed model merging framework outperforms state-of-the-art models while maintaining robust reasoning performance.

Copied to clipboard

Anvesh Rao Vijjini, Sagar B. Manjunath, Snigdha Chaturvedi

CIG: Measuring Conversational Information Gain in Deliberative Dialogues with Semantic Memory Dynamics (2026.acl-long)

Challenge:	Power differences shape human communication through well-documented socio-cognitive effects . asymmetric relationships or power differentials give rise to well-known socio-computational effects - lianelli, 1976 .
Approach:	They simulate multi-turn, power-asymmetric dialogues with personas from diverse professions . they find that LLMs show key socio-cognitive effects of power, albeit with nuances and variability .
Outcome:	The results show that large language models exhibit socio-cognitive effects of power . the results are consistent with previous studies on LLMs .

Copied to clipboard

Ming-Bin Chen, Jey Han Lau, Lea Frermann

A Multilingual Social Bias Benchmark Incorporating Thinking Processes (2026.acl-long)

Challenge:	Using a semantic memory, we score each utterance along three interpretable dimensions: Novelty, Relevance, and Implication Scope.
Approach:	They propose a framework for Conversational Information Gain that evaluates each utterance in terms of how it advances collective understanding of the target topic.
Outcome:	The proposed framework evaluates each utterance in terms of how it advances collective understanding of the target topic.

Copied to clipboard

Masahiro Kaneko, Danushka Bollegala, Timothy Baldwin

Preference Heads in Large Language Models: A Mechanistic Framework for Interpretable Personalization (2026.acl-long)

Challenge:	Large Language Models (LLMs) can learn useful knowledge and harmful stereotypes, making bias evaluation essential.
Approach:	They propose a multilingual social bias benchmark that incorporates human-generated reasoning as part of the thinking process.
Outcome:	The proposed method demonstrates superior performance over LLM-generated methods . human-generated thinking yields higher-quality evaluations than template-based approaches .

Copied to clipboard

Weixu Zhang, Ye Yuan, Changjiang Han, Yuxing Tian, Zipeng Sun, Linfeng Du, Jikun Kang, Hong Kang, Xue Liu, Haolun Wu

Influence-based Online Experience Selection for Effective RLHF (2026.acl-long)

Challenge:	Large Language Models exhibit strong implicit personalization ability, but most approaches treat this behavior as a black box.
Approach:	They propose a mechanistic interpretation perspective and propose 'sparse' set of Preference Heads . they compute a Preference Contribution Score for each attention head and compare their predictions .
Outcome:	The proposed framework computes a Preference Contribution Score (PCS) for each attention head and measures its causal impact on user aligned outputs.

Copied to clipboard

Yifan Gong, Jing Yao, Xiting Wang, Xunlong Wang, Xiaoyuan Yi, Xing Xie

NaturalGAIA: A Verifiable Benchmark and Hierarchical Framework for Long-Horizon GUI Tasks (2026.acl-long)

Challenge:	Existing methods for RL fail to establish an interpretable connection between data and optimization objectives.
Approach:	They propose a data selection method that dynamically estimates the influence of individual training samples on policy optimization.
Outcome:	The proposed method significantly improves training effectiveness with fewer optimization steps.

Copied to clipboard

Zihan Zheng, Tianle Cui, Taoran Wang, Fengtao Wang, Jiahui Pan, Lewei He, Qianglong Chen

Challenge:	Current research faces an "Evaluation-Realism Dilemma" due to unstable MLLM judges or manual verification.
Approach:	They propose a verifiable evaluation dataset grounded in real-world human GUI intents.
Outcome:	The proposed framework outperforms the state-of-the-art framework in achieving a weighted pathway success rate of 45.6% while reducing token consumption and execution time by 76%.

Look Within or Beyond? A Theoretical Comparison Between Parameter-Efficient and Full Fine-Tuning (2026.acl-long)

Copied to clipboard

YongKang Liu, Xingle Xu, Ercong Nie, Zijing Wang, Shi Feng, Daling Wang, Qian Li, Hinrich Schuetze

Tackling the Root of Misinformation by Teaching Laypeople about Logical Fallacies via Socratic Questioning and Critical Argumentation (2026.acl-long)

Challenge:	Parameter-Efficient Fine-Tuning (PEFT) is an alternative to Full-Parameter Fine-tuning, but its effectiveness on complex tasks such as reasoning and instruction-following remains unclear.
Approach:	They propose to use PEFT to reduce the number of trainable parameters while freezing the weights of LLMs.
Outcome:	The proposed methods perform well on standard tasks, but weaknesses on complex and adversarial settings call for new directions beyond current paradigms.

Copied to clipboard

Minjing Shi, Junling Wang, Jingwei Ni, Sankalan Pal Chowdhury, Mrinmaya Sachan

TRAC: Teacher-Guided Token Reward with Adaptive Calibration for Robust Policy Optimization (2026.acl-long)

Challenge:	Existing systems that detect logical fallacies in public discourse do not help people recognize them independently.
Approach:	They propose an intelligent tutoring system which uses large language models to help humans learn about logical fallacies.
Outcome:	The proposed system outperforms baseline LLMs lacking such pedagogical strategies.

Copied to clipboard

Sitong Wu, Haoru Tan, Xichen Zhang, Bin Xia, Wenhu Zhang, Xiaojuan Qi, Bei Yu, Jiaya Jia

Macaron: Controlled, Human-Written Benchmark for Multilingual and Multicultural Reasoning via Template-Filling (2026.acl-long)

Challenge:	Current reward models for reinforcement learning (RL) rely on outcome rewards that propagate a single scalar value across all tokens based on final correctness.
Approach:	They propose a framework that derives dense token-level supervision from LLMs . they use a multi-granularity calibration mechanism to modulate teacher influence .
Outcome:	The proposed framework evaluates teacher reliability across problem-level expertise, trajectory-level discrimination, and token-level confidence.

Copied to clipboard

Alaa Elsetohy, Sama Hadhoud, Haryo Akbarianto Wibowo, Chenxi Whitehouse, Genta Indra Winata, Fajri Koto, Alham Fikri Aji

Typology-Aware Multilingual Morphosyntactic Parsing with Joint Abstract Node Modeling (2026.acl-long)

Challenge:	Existing benchmarks test reasoning over culturally grounded premises, but translation-parallel benchmarks inherit English-centric scenarios.
Approach:	They propose a template-first benchmark that factorizes reasoning type and cultural aspect across question languages.
Outcome:	The proposed benchmark factorizes reasoning type and cultural aspect across question languages.

Copied to clipboard

Kutay Acar, Gülşen Eryiğit

Jeongin Yun, Jaeri Lee, Jongjin Kim, Minjun Kim, Jinho Song, U Kang

Challenge:	UniDive 2025 Morphosyntactic Parsing (MSP) shared task unifies dependency structure, morphological features, and unrealized arguments.
Approach:	They propose a multilingual, typology-aware joint system that integrates word-type prediction, content-only parsing, morphological tagging, and an abstract-node component within a single architecture.
Outcome:	The proposed model outperforms the leading submission by 3.23 percentage points in MSLAS, 3.35 in LAS, and 1.78 in FEATS macro F1.

Copied to clipboard

Beyond Variance: Knowledge-Aware LLM Compression via Fisher-Aligned Subspace Diagnostics (2026.acl-long)

Challenge:	Existing methods for parameter sharing rely on naive grouping and fail to correct sharing-induced discrepancies.
Approach:	They propose a parameter sharing framework that performs similarity-based grouping to ensure accurate sharing and allocates parameters adaptively to preserve diversity within each group.
Outcome:	The proposed framework outperforms existing methods, achieving 32.1% lower perplexity and 23.3% higher few-shot reasoning accuracy.

Copied to clipboard

Ibne Farabi Shihab, Sanjeda Akter, Anuj Sharma

REMIND: Memorization and Unlearning in LLMs Through the Lens of Input Loss Landscapes (2026.acl-long)

Challenge:	Existing methods for activation compression are gradient-blind and preserve high-variance dimensions regardless of their impact on factual knowledge preservation.
Approach:	They propose a knowledge-aware compression framework that models activation-gradient coupling by directly modeling subspaces.
Outcome:	The proposed framework preserves 6–8% more accuracy on knowledge-intensive benchmarks compared to variance-based methods at 50% rank reduction.

Copied to clipboard

Liran Cohen, Yaniv Nemcovsky, Avi Mendelson

Reinforced Efficient Reasoning via Semantically Diverse Exploration (2026.acl-long)

Challenge:	REMIND is a framework that diagnoses residual memorization states by probing local ILL curvature over semantically coherent neighborhoods.
Approach:	They propose a framework that diagnoses memorization states by probing local ILL curvature over semantically coherent neighborhoods.
Outcome:	The proposed framework outperforms baseline models with 82% multi-class ROC-AUC and 2 higher AUC at 1% FPR.

Copied to clipboard

Ziqi Zhao, Zhaochun Ren, Jiahong Zou, Liu Yang, Zhiwei Xu, Xuri Ge, Zhumin Chen, Xinyu Ma, Daiting Shi, Shuaiqiang Wang, Dawei Yin, Xin Xin

Is Chain-of-Thought Really Not Explainability? Chain-of-Thought Can Be Faithful without Hint Verbalization (2026.acl-long)

Challenge:	Existing methods for reinforcement learning with verifiable rewards suffer from limited exploration diversity and inefficient reasoning.
Approach:	They propose a method that rewards concise and correct reasoning while penalizing unnecessarily long reasoning chains.
Outcome:	Extensive experiments on Qwen and Llama models validate the effectiveness and efficiency of ROSE.

Copied to clipboard

Kerem Zaman, Shashank Srivastava

MegaRAG: Multimodal Knowledge Graph-Based Retrieval Augmented Generation (2026.acl-long)

Challenge:	Recent work labels a CoT as unfaithful if it omits a prompt-injected hint that affected the prediction.
Approach:	They propose to use the Biasing Features metric to label a CoT as unfaithful if it omits a prompt-injected hint that affected the prediction.
Outcome:	The proposed metric confuses unfaithfulness with incompleteness, the authors argue . larger inference-time budgets greatly increase hint verbalization, they show .

Copied to clipboard

Chi-Hsiang Hsiao, Yi-Cheng Wang, Tzung-Sheng Lin, Yi-Ren Yeh, Chu-song Chen

SELECting over Tokens: Curating Pre-training Data at Scale via Token Classification (2026.acl-long)

Challenge:	Existing RAG solutions for large language models are limited by context windows limiting their ability to process long-form, domain-specific content.
Approach:	They propose a multimodal knowledge graph-based RAG that enables cross-modal reasoning . their method incorporates visual cues into the construction of knowledge graphs, retrieval phase, and answer generation process .
Outcome:	Experimental results show that the proposed approach outperforms existing approaches on textual and multimodal benchmarks.

Copied to clipboard

Xin Tong, Weidong Zhang, Jiaang Li, Haibin Chen, Shilei Liu, Langming Liu, Kangtao Lv, Yujin Yuan, Wenbo Su, Bo Zheng

TROJail: Trajectory-Level Optimization for Multi-Turn Large Language Model Jailbreaks with Process Rewards (2026.acl-long)

Challenge:	Existing pipelines rely on expert-crafted heuristic rules, which lack content-aware, fine-grained noise detection.
Approach:	They propose a framework that reframes data refinement as a highly efficient token classification task.
Outcome:	The proposed framework outperforms existing pipelines on benchmarks and is 2.5x faster at inference.

Copied to clipboard

Xiqiao Xiong, Ouxiang Li, Zhuo Liu, Moxin Li, Wentao Shi, Fengbin Zhu, Qifan Wang, Fuli Feng

Accurate and Efficient Statistical Testing for Word Semantic Breadth (2026.acl-long)

Challenge:	Existing approaches to training multi-turn attackers to probe model safety vulnerabilities rely on turn-level optimization, which is insufficient for learning long-term attack strategies.
Approach:	They propose a multi-turn reinforcement learning problem that optimizes the harmfulness of the final-turn response as the outcome reward.
Outcome:	The proposed approach improves attack success rates across multiple models and benchmarks, highlighting the effectiveness of the proposed approach.

Copied to clipboard

Yo Ehara

Constructing coherent spatial memory in LLM agents through graph rectification (2026.acl-long)

Challenge:	Existing studies on semantic breadth and spread of words have focused on semantic relatedness, but contextualized token embeddings can be misleading.
Approach:	They propose a Householder-aligned permutation test to isolate dispersion differences from directional differences.
Outcome:	The proposed method reduces Type-I error by 32.5% while preserving sensitivity to genuine breadth differences.

Copied to clipboard

Puzhen Zhang, Xuyang Chen, Yu Feng, Yuhan Jiang, Liqiu Meng