Peer Revenue Play credits. Real papers.

Papers

Fresh from arXiv, typeset properly. Highlighted words trade on the market.

Computer Science Computation and Language

cs.CL Watch Computation and Language

Monday 26 January

Computation and Language

Suppressing Final Layer Hidden State Jumps in Transformer Pretraining

Keigo Shibata, Kazuki Yano, Ryosuke Takahashi et al. (+3 more)

This paper discusses the internal behavior of Transformer language models. Many recent pre-trained models have been reported to exhibit only slight changes in the angular distance between the input and output hidden state vectors in the middle Transformer layers, despite a disproportionately large “jump” in the angular distance occurring in or around the final Transformer layer. To characterize this, we first introduce a quantitative metric for the jump strength around the final layer, and then demonstrate its prevalence across many open-weight models, as well as its amplification throughout pre-training. Assuming such jumps indicate an undesirable property, we propose the jump-suppressing regularizer (JREG) which penalizes this jump during pre-training, thereby encouraging more balanced capability usage across the middle layers. Empirical evaluations of three model sizes of Llama-based models, trained with the proposed JREG method, reveal improved task performance compared to the baseline without altering the model architecture.
Read the full abstract

All authors Keigo Shibata, Kazuki Yano, Ryosuke Takahashi, Jaesung Lee, Wataru Ikeda, Jun Suzuki

Computation and Language

Temp-R1: A Unified Autonomous Agent for Complex Temporal KGQA via Reverse Curriculum Reinforcement Learning

Zhaoyan Gong, Zhiqiang Liu, Songze Li et al. (+7 more)

Temporal Knowledge Graph Question Answering (TKGQA) is inherently challenging, as it requires sophisticated reasoning over dynamic facts with multi-hop dependencies and complex temporal constraints. Existing methods rely on fixed workflows and expensive closed-source APIs, limiting flexibility and scalability. We propose Temp-R1, the first autonomous end-to-end agent for TKGQA trained through reinforcement learning. To address cognitive overload in single-action reasoning, we expand the action space with specialized internal actions alongside external action. To prevent shortcut learning on simple questions, we introduce reverse curriculum learning that trains on difficult questions first, forcing the development of sophisticated reasoning before transferring to easier cases. Our 8B-parameter Temp-R1 achieves state-of-the-art performance on MultiTQ and TimelineKGQA, improving 19.8% over strong baselines on complex questions. Our work establishes a new paradigm for autonomous temporal reasoning agents. The code is available at https://github.com/zjukg/Temp-R1.
Read the full abstract

All authors Zhaoyan Gong, Zhiqiang Liu, Songze Li, Xiaoke Guo, Yuanxiang Liu, Xinle Deng, Zhizhen Liu, Lei Liang, Huajun Chen, Wen Zhang

Computation and Language

U-Fold: Dynamic Intent-Aware Context Folding for User-Centric Agents

Jin Su, Runnan Fang, Yeqiu Li et al. (+5 more)

Large language model (LLM)-based agents have been successfully deployed in many tool-augmented settings, but their scalability is fundamentally constrained by context length. Existing context-folding methods mitigate this issue by summarizing past interactions, yet they are typically designed for single-query or single-intent scenarios. In more realistic user-centric dialogues, we identify two major failure modes: (i) they irreversibly discard fine-grained constraints and intermediate facts that are crucial for later decisions, and (ii) their summaries fail to track evolving user intent, leading to omissions and erroneous actions. To address these limitations, we propose U-Fold, a dynamic context-folding framework tailored to user-centric tasks. U-Fold retains the full user–agent dialogue and tool-call history but, at each turn, uses two core components to produce an intent-aware, evolving dialogue summary and a compact, task-relevant tool log. Extensive experiments on $\tau$-bench, $\tau^2$-bench, VitaBench, and harder context-inflated settings show that U-Fold consistently outperforms ReAct (achieving a 71.4% win rate in long-context settings) and prior folding baselines (with improvements of up to 27.0%), particularly on long, noisy, multi-turn tasks. Our study demonstrates that U-Fold is a promising step toward transferring context-management techniques from single-query benchmarks to realistic user-centric applications.
Read the full abstract

All authors Jin Su, Runnan Fang, Yeqiu Li, Xiaobin Wang, Shihao Cai, Pengjun Xie, Ningyu Zhang, Fajie Yuan

Computation and Language · Cross-listed from cs.AI

Think-Augmented Function Calling: Improving LLM Parameter Accuracy Through Embedded Reasoning

Lei Wei, Xiao Peng, Jinpeng Ou et al. (+1 more)

Large language models (LLMs) have demonstrated remarkable capabilities in function calling for autonomous agents, yet current mechanisms lack explicit reasoning transparency during parameter generation, particularly for complex functions with interdependent parameters. While existing approaches like chain-of-thought prompting operate at the agent level, they fail to provide fine-grained reasoning guidance for individual function parameters. To address these limitations, we propose Think-Augmented Function Calling (TAFC), a novel framework that enhances function calling accuracy through explicit reasoning at both function and parameter levels. Our method introduces a universal "think" parameter augmentation that enables models to articulate their decision-making process, with dynamic optimization for parameter descriptions to improve reasoning quality. For complex parameters, TAFC automatically triggers granular reasoning based on complexity scoring, ensuring appropriate justification for critical decisions. Additionally, we propose reasoning-guided optimization to align generated reasoning with human expectations. TAFC requires no architectural modifications to existing LLMs while maintaining full API compatibility. Evaluation on ToolBench across proprietary and open-source models demonstrates significant improvements in parameter generation accuracy and reasoning coherence for multi-parameter functions, while providing enhanced interpretability for debugging AI agent behaviors.
Read the full abstract

All authors Lei Wei, Xiao Peng, Jinpeng Ou, Bin Wang

Computation and Language

Reflecting Twice before Speaking with Empathy: Self-Reflective Alternating Inference for Empathy-Aware End-to-End Spoken Dialogue

Yuhang Jia, Pei Liu, Haoqin Sun et al. (+6 more)

End-to-end Spoken Language Models (SLMs) hold great potential for paralinguistic perception, and numerous studies have aimed to enhance their capabilities, particularly for empathetic dialogue. However, current approaches largely depend on rigid supervised signals, such as ground-truth response in supervised fine-tuning or preference scores in reinforcement learning. Such reliance is fundamentally limited for modeling complex empathy, as there is no single "correct" response and a simple numerical score cannot fully capture the nuances of emotional expression or the appropriateness of empathetic behavior. To address these limitations, we sequentially introduce EmpathyEval, a descriptive natural-language-based evaluation model for assessing empathetic quality in spoken dialogues. Building upon EmpathyEval, we propose ReEmpathy, an end-to-end SLM that enhances empathetic dialogue through a novel Empathetic Self-Reflective Alternating Inference mechanism, which interleaves spoken response generation with free-form, empathy-related reflective reasoning. Extensive experiments demonstrate that ReEmpathy substantially improves empathy-sensitive spoken dialogue by enabling reflective reasoning, offering a promising approach toward more emotionally intelligent and empathy-aware human-computer interactions.
Read the full abstract

All authors Yuhang Jia, Pei Liu, Haoqin Sun, Jiaming Zhou, Xuxin Cheng, Cao Liu, Ke Zeng, Xunliang Cai, Yong Qin

Computation and Language · Cross-listed from cs.DL

Designing large language model prompts to extract scores from messy text: A shared dataset and challenge

Mike Thelwall

In some areas of computing, natural language processing and information science, progress is made by sharing datasets and challenging the community to design the best algorithm for an associated task. This article introduces a shared dataset of 1446 short texts, each of which describes a research quality score on the UK scale of 1* to 4*. This is a messy collection, with some texts not containing scores and others including invalid scores or strange formats. With this dataset there is also a description of what constitutes a valid score and a "gold standard" of the correct scores for these texts (including missing values). The challenge is to design a prompt for Large Language Models (LLMs) to extract the scores from these texts as accurately as possible. The format for the response should be a number and no other text so there are two aspects to the challenge: ensuring that the LLM returns only a number, and instructing it to deduce the correct number for the text. As part of this, the LLM prompt needs to explain when to return the missing value code, -1, instead of a number when the text does not clearly contain one. The article also provides an example of a simple prompt. The purpose of the challenge is twofold: to get an effective solution to this problem, and to increase understanding of prompt design and LLM capabilities for complex numerical tasks. The initial solution suggested has an accuracy of 72.6%, so the challenge is to beat this.
Read the full abstract

Computation and Language · Cross-listed from cs.LG

FGGM: Fisher-Guided Gradient Masking for Continual Learning

Chao-Hong Tan, Qian Chen, Wen Wang et al. (+6 more)

Catastrophic forgetting impairs the continuous learning of large language models. We propose Fisher-Guided Gradient Masking (FGGM), a framework that mitigates this by strategically selecting parameters for updates using diagonal Fisher Information. FGGM dynamically generates binary masks with adaptive thresholds, preserving critical parameters to balance stability and plasticity without requiring historical data. Unlike magnitude-based methods such as MIGU, our approach offers a mathematically principled parameter importance estimation. On the TRACE benchmark, FGGM shows a 9.6% relative improvement in retaining general capabilities over supervised fine-tuning (SFT) and a 4.4% improvement over MIGU on TRACE tasks. Additional analysis on code generation tasks confirms FGGM's superior performance and reduced forgetting, establishing it as an effective solution.
Read the full abstract

All authors Chao-Hong Tan, Qian Chen, Wen Wang, Yukun Ma, Chong Zhang, Chong Deng, Qinglin Zhang, Xiangang Li, Jieping Ye

Computation and Language

BoRP: Bootstrapped Regression Probing for Scalable and Human-Aligned LLM Evaluation

Peng Sun, Xiangyu Zhang, Duan Wu et al. (+5 more)

Accurate evaluation of user satisfaction is critical for iterative development of conversational AI. However, for open-ended assistants, traditional A/B testing lacks reliable metrics: explicit feedback is sparse, while implicit metrics are ambiguous. To bridge this gap, we introduce BoRP (Bootstrapped Regression Probing), a scalable framework for high-fidelity satisfaction evaluation. Unlike generative approaches, BoRP leverages the geometric properties of LLM latent space. It employs a polarization-index-based bootstrapping mechanism to automate rubric generation and utilizes Partial Least Squares (PLS) to map hidden states to continuous scores. Experiments on industrial datasets show that BoRP (Qwen3-8B/14B) significantly outperforms generative baselines (even Qwen3-Max) in alignment with human judgments. Furthermore, BoRP reduces inference costs by orders of magnitude, enabling full-scale monitoring and highly sensitive A/B testing via CUPED.
Read the full abstract

All authors Peng Sun, Xiangyu Zhang, Duan Wu, Lu Tan, Jian Lin, He Yang, Qi Qian, Yikai Wang

Computation and Language

TechING: Towards Real World Technical Image Understanding via VLMs

Tafazzul Nadeem, Bhavik Shangari, Manish Rai et al. (+2 more)

Professionals working in technical domain typically hand-draw (on whiteboard, paper, etc.) technical diagrams (e.g., flowcharts, block diagrams, etc.) during discussions; however, if they want to edit these later, it needs to be drawn from scratch. Modern day VLMs have made tremendous progress in image understanding but they struggle when it comes to understanding technical diagrams. One way to overcome this problem is to fine-tune on real world hand-drawn images, but it is not practically possible to generate large number of such images. In this paper, we introduce a large synthetically generated corpus (reflective of real world images) for training VLMs and subsequently evaluate VLMs on a smaller corpus of hand-drawn images (with the help of humans). We introduce several new self-supervision tasks for training and perform extensive experiments with various baseline models and fine-tune Llama 3.2 11B-instruct model on synthetic images on these tasks to obtain LLama-VL-TUG, which significantly improves the ROUGE-L performance of Llama 3.2 11B-instruct by 2.14x and achieves the best all-round performance across all baseline models. On real-world images, human evaluation reveals that we achieve minimum compilation errors across all baselines in 7 out of 8 diagram types and improve the average F1 score of Llama 3.2 11B-instruct by 6.97x.
Read the full abstract

All authors Tafazzul Nadeem, Bhavik Shangari, Manish Rai, Gagan Raj Gupta, Ashutosh Modi

Computation and Language · Cross-listed from cs.CY

Generative AI in Saudi Arabia: A National Survey of Adoption, Risks, and Public Perceptions

Abdulaziz AlDakheel, Ali Alshehre, Esraa Alamoudi et al. (+3 more)

Generative Artificial Intelligence (GenAI) is rapidly becoming embedded in Saudi Arabia's digital transformation under Vision 2030, yet public awareness, adoption, and concerns surrounding these tools remain underexplored. This study provides an early snapshot of GenAI engagement among Saudi nationals. Using a nationwide survey of 330 participants across regions, age groups, and employment sectors, we examine seven dimensions of GenAI use: awareness and understanding, adoption patterns, perceived impacts, training needs, risks and barriers, data-sharing behaviors, and future expectations. Findings show that 93% of respondents actively use GenAI primarily for text-based tasks, while more advanced uses such as programming or multimodal generation are less common. Despite the prevalence of use, overall awareness and conceptual understanding remain uneven, with many reporting limited technical knowledge. Participants recognize GenAI's benefits for productivity, work quality, and understanding complex information, yet caution that sustained reliance may undermine critical thinking and key professional skills. Trust in AI-generated outputs remains cautious, with widespread concerns about privacy, misinformation, and ethical misuse, including potential job displacement. Respondents show strong interest in structured GenAI training that combines foundational skills, domain-specific applications, and clear guidance on privacy, ethics, and responsible use. These results establish a baseline for GenAI engagement in Saudi Arabia and highlight priorities for policymakers and developers: expanding AI literacy, ensuring culturally and linguistically aligned GenAI solutions, and strengthening frameworks for privacy and responsible deployment.
Read the full abstract

All authors Abdulaziz AlDakheel, Ali Alshehre, Esraa Alamoudi, Moslim AlKhabbaz, Ahmed Aljohani, Raed Alharbi

Computation and Language · Cross-listed from cs.HC

PaperTok: Exploring the Use of Generative AI for Creating Short-form Videos for Research Communication

Meziah Ruby Cristobal, Hyeonjeong Byeon, Tze-Yu Chen et al. (+5 more)

The dissemination of scholarly research is critical, yet researchers often lack the time and skills to create engaging content for popular media such as short-form videos. To address this gap, we explore the use of generative AI to help researchers transform their academic papers into accessible video content. Informed by a formative study with science communicators and content creators (N=8), we designed PaperTok, an end-to-end system that automates the initial creative labor by generating script options and corresponding audiovisual content from a source paper. Researchers can then refine based on their preferences with further prompting. A mixed-methods user study (N=18) and crowdsourced evaluation (N=100) demonstrate that PaperTok's workflow can help researchers create engaging and informative short-form videos. We also identified the need for more fine-grained controls in the creation process. To this end, we offer implications for future generative tools that support science outreach.
Read the full abstract

All authors Meziah Ruby Cristobal, Hyeonjeong Byeon, Tze-Yu Chen, Ruoxi Shang, Donghoon Shin, Ruican Zhong, Tony Zhou, Gary Hsieh

Computation and Language · Cross-listed from cs.LG

PaperSearchQA: Learning to Search and Reason over Scientific Papers with RLVR

James Burgess, Jan N. Hansen, Duo Peng et al. (+5 more)

Search agents are language models (LMs) that reason and search knowledge bases (or the web) to answer questions; recent methods supervise only the final answer accuracy using reinforcement learning with verifiable rewards (RLVR). Most RLVR search agents tackle general-domain QA, which limits their relevance to technical AI systems in science, engineering, and medicine. In this work we propose training agents to search and reason over scientific papers – this tests technical question-answering, it is directly relevant to real scientists, and the capabilities will be crucial to future AI Scientist systems. Concretely, we release a search corpus of 16 million biomedical paper abstracts and construct a challenging factoid QA dataset called PaperSearchQA with 60k samples answerable from the corpus, along with benchmarks. We train search agents in this environment to outperform non-RL retrieval baselines; we also perform further quantitative analysis and observe interesting agent behaviors like planning, reasoning, and self-verification. Our corpus, datasets, and benchmarks are usable with the popular Search-R1 codebase for RLVR training and released on https://huggingface.co/collections/jmhb/papersearchqa. Finally, our data creation methods are scalable and easily extendable to other scientific domains.
Read the full abstract

All authors James Burgess, Jan N. Hansen, Duo Peng, Yuhui Zhang, Alejandro Lozano, Min Woo Sun, Emma Lundberg, Serena Yeung-Levy

Computation and Language

MemWeaver: Weaving Hybrid Memories for Traceable Long-Horizon Agentic Reasoning

Juexiang Ye, Xue Li, Xinyu Yang et al. (+4 more)

Large language model-based agents operating in long-horizon interactions require memory systems that support temporal consistency, multi-hop reasoning, and evidence-grounded reuse across sessions. Existing approaches largely rely on unstructured retrieval or coarse abstractions, which often lead to temporal conflicts, brittle reasoning, and limited traceability. We propose MemWeaver, a unified memory framework that consolidates long-term agent experiences into three interconnected components: a temporally grounded graph memory for structured relational reasoning, an experience memory that abstracts recurring interaction patterns from repeated observations, and a passage memory that preserves original textual evidence. MemWeaver employs a dual-channel retrieval strategy that jointly retrieves structured knowledge and supporting evidence to construct compact yet information-dense contexts for reasoning. Experiments on the LoCoMo benchmark demonstrate that MemWeaver substantially improves multi-hop and temporal reasoning accuracy while reducing input context length by over 95% compared to long-context baselines.
Read the full abstract

All authors Juexiang Ye, Xue Li, Xinyu Yang, Chengkai Huang, Lanshun Nie, Lina Yao, Dechen Zhan

Computation and Language

Fine-Grained Emotion Detection on GoEmotions: Experimental Comparison of Classical Machine Learning, BiLSTM, and Transformer Models

Ani Harutyunyan and Sachin Kumar

Fine-grained emotion recognition is a challenging multi-label NLP task due to label overlap and class imbalance. In this work, we benchmark three modeling families on the GoEmotions dataset: a TF-IDF-based logistic regression system trained with binary relevance, a BiLSTM with attention, and a BERT model fine-tuned for multi-label classification. Experiments follow the official train/validation/test split, and imbalance is mitigated using inverse-frequency class weights. Across several metrics, namely Micro-F1, Macro-F1, Hamming Loss, and Subset Accuracy, we observe that logistic regression attains the highest Micro-F1 of 0.51, while BERT achieves the best overall balance surpassing the official paper's reported results, reaching Macro-F1 0.49, Hamming Loss 0.036, and Subset Accuracy 0.36. This suggests that frequent emotions often rely on surface lexical cues, whereas contextual representations improve performance on rarer emotions and more ambiguous examples.
Read the full abstract

Computation and Language · Cross-listed from cs.LG

FP8-RL: A Practical and Stable Low-Precision Stack for LLM Reinforcement Learning

Zhaopeng Qiu, Shuang Yu, Jingqi Zhang et al. (+4 more)

Reinforcement learning (RL) for large language models (LLMs) is increasingly bottlenecked by rollout (generation), where long output sequence lengths make attention and KV-cache memory dominate end-to-end step time. FP8 offers an attractive lever for accelerating RL by reducing compute cost and memory traffic during rollout, but applying FP8 in RL introduces unique engineering and algorithmic challenges: policy weights change every step (requiring repeated quantization and weight synchronization into the inference engine) and low-precision rollouts can deviate from the higher-precision policy assumed by the trainer, causing train-inference mismatch and potential instability. This report presents a practical FP8 rollout stack for LLM RL, implemented in the veRL ecosystem with support for common training backends (e.g., FSDP/Megatron-LM) and inference engines (e.g., vLLM/SGLang). We (i) enable FP8 W8A8 linear-layer rollout using blockwise FP8 quantization, (ii) extend FP8 to KV-cache to remove long-context memory bottlenecks via per-step QKV scale recalibration, and (iii) mitigate mismatch using importance-sampling-based rollout correction (token-level TIS/MIS variants). Across dense and MoE models, these techniques deliver up to 44% rollout throughput gains while preserving learning behavior comparable to BF16 baselines.
Read the full abstract

All authors Zhaopeng Qiu, Shuang Yu, Jingqi Zhang, Shuai Zhang, Xue Huang, Jingyi Yang, Junjie Lai

Computation and Language · Cross-listed from cs.AI

DeepPlanning: Benchmarking Long-Horizon Agentic Planning with Verifiable Constraints

Yinger Zhang, Shutong Jiang, Renhao Li et al. (+6 more)

While agent evaluation has shifted toward long-horizon tasks, most benchmarks still emphasize local, step-level reasoning rather than the global constrained optimization (e.g., time and financial budgets) that demands genuine planning ability. Meanwhile, existing LLM planning benchmarks underrepresent the active information gathering and fine-grained local constraints typical of real-world settings. To address this, we introduce DeepPlanning, a challenging benchmark for practical long-horizon agent planning. It features multi-day travel planning and multi-product shopping tasks that require proactive information acquisition, local constrained reasoning, and global constrained optimization. Evaluations on DeepPlanning show that even frontier agentic LLMs struggle with these problems, highlighting the importance of reliable explicit reasoning patterns and parallel tool use for achieving better effectiveness-efficiency trade-offs. Error analysis further points to promising directions for improving agentic LLMs over long planning horizons. We open-source the code and data to support future research.
Read the full abstract

All authors Yinger Zhang, Shutong Jiang, Renhao Li, Jianhong Tu, Yang Su, Lianghao Deng, Xudong Guo, Chenxu Lv, Junyang Lin

Computation and Language

Typhoon-S: Minimal Open Post-Training for Sovereign Large Language Models

Kunat Pipatanakul and Pittawat Taveekitworachai

Large language models (LLMs) have progressed rapidly; however, most state-of-the-art models are trained and evaluated primarily in high-resource languages such as English and Chinese, and are often developed by a small number of organizations with access to large-scale compute and data. This gatekeeping creates a practical barrier for sovereign settings in which a regional- or national-scale institution or domain owner must retain control and understanding of model weights, training data, and deployment while operating under limited resources and strict transparency constraints. To this end, we identify two core requirements: (1) adoptability, the ability to transform a base model into a general-purpose assistant, and (2) sovereign capability, the ability to perform high-stakes, region-specific tasks (e.g., legal reasoning in local languages and cultural knowledge). We investigate whether these requirements can be achieved without scaling massive instruction corpora or relying on complex preference tuning pipelines and large-scale reinforcement fine-tuning (RFT). We present Typhoon S, a minimal and open post-training recipe that combines supervised fine-tuning, on-policy distillation, and small-scale RFT. Using Thai as a representative case study, we demonstrate that our approach transforms both sovereign-adapted and general-purpose base models into instruction-tuned models with strong general performance. We further show that small-scale RFT with InK-GRPO – an extension of GRPO that augments the GRPO loss with a next-word prediction loss – improves Thai legal reasoning and Thai-specific knowledge while preserving general capabilities. Our results suggest that a carefully designed post-training strategy can reduce the required scale of instruction data and computation, providing a practical path toward high-quality sovereign LLMs under academic-scale resources.
Read the full abstract

Computation and Language

BEAR: Budgeted Evidence Allocation for Multi-Document Reasoning

Lin Sun, Linglin Zhang, Jingang Huang et al. (+3 more)

We argue that multi-document reasoning is constrained not only by how much text a model can read, but also by how limited query-time evidence budget is allocated across documents and semantic granularities. Full-context inference exposes the model to broad evidence non-selectively and at high per-query cost, while flat chunk retrieval often returns locally relevant passages that are weakly organized for cross-document synthesis. We present BEAR, a framework for structured evidence allocation that builds hierarchical semantic indices offline and performs coarse-to-fine evidence access at query time through complementary exploration and recovery paths. This coarse-to-fine design can be viewed as structured evidence allocation under a fixed evidence-context budget. Across synthetic and real-world benchmarks, BEAR performs particularly strongly on DragonBall, remains competitive with strong retrieval-based baselines on HotpotQA, and yields the best retrieval-based result on 2Wiki under our evaluated protocol, while operating under substantially smaller query-time evidence budgets than the reported long-context references. Additional analyses suggest that the gains are associated with hierarchy as an allocation substrate together with complementary exploration and recovery, rather than semantic chunking alone.
Read the full abstract

All authors Lin Sun, Linglin Zhang, Jingang Huang, Change Jia, Zhengwei Cheng, Xiangzheng Zhang

Computation and Language

GLEN-Bench: A Graph-Language based Benchmark for Nutritional Health

Jiatan Huang, Zheyuan Zhang, Tianyi Ma et al. (+4 more)

Nutritional interventions are important for managing chronic health conditions, but current computational methods provide limited support for personalized dietary guidance. We identify three key gaps: (1) dietary pattern studies often ignore real-world constraints such as socioeconomic status, comorbidities, and limited food access; (2) recommendation systems rarely explain why a particular food helps a given patient; and (3) no unified benchmark evaluates methods across the connected tasks needed for nutritional interventions. We introduce GLEN-Bench, the first comprehensive graph-language based benchmark for nutritional health assessment. We combine NHANES health records, FNDDS food composition data, and USDA food-access metrics to build a knowledge graph that links demographics, health conditions, dietary behaviors, poverty-related constraints, and nutrient needs. We test the benchmark using opioid use disorder, where models must detect subtle nutritional differences across disease stages. GLEN-Bench includes three linked tasks: risk detection identifies at-risk individuals from dietary and socioeconomic patterns; recommendation suggests personalized foods that meet clinical needs within resource constraints; and question answering provides graph-grounded, natural-language explanations to facilitate comprehension. We evaluate these graph-language approaches, including graph neural networks, large language models, and hybrid architectures, to establish solid baselines and identify practical design choices. Our analysis identifies clear dietary patterns linked to health risks, providing insights that can guide practical interventions.
Read the full abstract

All authors Jiatan Huang, Zheyuan Zhang, Tianyi Ma, Mingchen Li, Yaning Zheng, Yanfang Ye, Chuxu Zhang

Computation and Language

CHiRPE: A Step Towards Real-World Clinical NLP with Clinician-Oriented Model Explanations

Stephanie Fong, Zimu Wang, Guilherme C. Oliveira et al. (+9 more)

The medical adoption of NLP tools requires interpretability by end users, yet traditional explainable AI (XAI) methods are misaligned with clinical reasoning and lack clinician input. We introduce CHiRPE (Clinical High-Risk Prediction with Explainability), an NLP pipeline that takes transcribed semi-structured clinical interviews to: (i) predict psychosis risk; and (ii) generate novel SHAP explanation formats co-developed with clinicians. Trained on 944 semi-structured interview transcripts across 24 international clinics of the AMP-SCZ study, the CHiRPE pipeline integrates symptom-domain mapping, LLM summarisation, and BERT classification. CHiRPE achieved over 90% accuracy across three BERT variants and outperformed baseline models. Explanation formats were evaluated by 28 clinical experts who indicated a strong preference for our novel concept-guided explanations, especially hybrid graph-and-text summary formats. CHiRPE demonstrates that clinically-guided model development produces both accurate and interpretable results. Our next step is focused on real-world testing across our 24 international sites.
Read the full abstract

All authors Stephanie Fong, Zimu Wang, Guilherme C. Oliveira, Xiangyu Zhao, Yiwen Jiang, Jiahe Liu, Beau-Luke Colton, Scott Woods, Martha E. Shenton, Barnaby Nelson, Zongyuan Ge, Dominic Dwyer

Computation and Language

Sparks of Cooperative Reasoning: LLMs as Strategic Hanabi Agents

Mahesh Ramesh, Kaousheik Jayakumar, Aswinkumar Ramkumar et al. (+3 more)

Cooperative reasoning under incomplete information remains challenging for both humans and multi-agent systems. The card game Hanabi embodies this challenge, requiring theory-of-mind reasoning and strategic communication. We benchmark 17 state-of-the-art LLM agents in 2-5 player games and study the impact of context engineering across model scales (4B to 600B+) to understand persistent coordination failures and robustness to scaffolding: from a minimal prompt with only explicit card details (Watson setting), to scaffolding with programmatic, Bayesian-motivated deductions (Sherlock), to multi-turn state tracking via working memory (Mycroft setting). We show that (1) agents can maintain an internal working memory for state tracking and (2) cross-play performance between different LLMs smoothly interpolates with model strength. In the Sherlock setting, the strongest reasoning models exceed 15 points on average across player counts, yet still trail experienced humans and specialist Hanabi agents, both consistently scoring above 20. We release the first public Hanabi datasets with annotated trajectories and move utilities: (1) HanabiLogs, containing 1,520 full game logs for instruction tuning, and (2) HanabiRewards, containing 560 games with dense move-level value annotations for all candidate moves. Supervised and RL finetuning of a 4B open-weight model (Qwen3-Instruct) on our datasets improves cooperative Hanabi play by 21% and 156% respectively, bringing performance to within ~3 points of a strong proprietary reasoning model (o4-mini) and surpassing the best non-reasoning model (GPT-4.1) by 52%. The HanabiRewards RL-finetuned model further generalizes beyond Hanabi, improving performance on a cooperative group-guessing benchmark by 11%, temporal reasoning on EventQA by 6.4 points, instruction-following on IFBench by 1.7 Pass@10, and matching AIME 2025 mathematical reasoning Pass@10.
Read the full abstract

All authors Mahesh Ramesh, Kaousheik Jayakumar, Aswinkumar Ramkumar, Pavan Thodima, Aniket Rege, Emmanouil-Vasileios Vlatakis-Gkaragkounis

Computation and Language

Grounded Concreteness: Human-Like Concreteness Sensitivity in Vision-Language Models

Aryan Roy, Zekun Wang and Christopher J. MacLellan

Do vision–language models (VLMs) develop more human-like sensitivity to linguistic concreteness than text-only large language models (LLMs) when both are evaluated with text-only prompts? We study this question with a controlled comparison between matched Llama text backbones and their Llama Vision counterparts across multiple model scales, treating multimodal pretraining as an ablation on perceptual grounding rather than access to images at inference. We measure concreteness effects at three complementary levels: (i) output behavior, by relating question-level concreteness to QA accuracy; (ii) embedding geometry, by testing whether representations organize along a concreteness axis; and (iii) attention dynamics, by quantifying context reliance via attention-entropy measures. In addition, we elicit token-level concreteness ratings from models and evaluate alignment to human norm distributions, testing whether multimodal training yields more human-consistent judgments. Across benchmarks and scales, VLMs show larger gains on more concrete inputs, exhibit clearer concreteness-structured representations, produce ratings that better match human norms, and display systematically different attention patterns consistent with increased grounding.
Read the full abstract

Computation and Language

Neurocomputational Mechanisms of Syntactic Transfer in Bilingual Sentence Production

Ahmet Yavuz Uluslu and Elliot Murphy

We discuss the benefits of incorporating oscillatory neural mechanisms into the study of bilingual production errors and their traditionally documented timing signatures (e.g., event-related potentials), which can offer new implementational-level constraints for theories of bilingualism. We argue that a recent neural model of language, ROSE, can offer a neurocomputational account of syntactic transfer in bilingual production, capturing some of its formal properties and the scope of morphosyntactic sequencing failure modes. We take as a case study cross-linguistic influence (CLI) and attendant theories of functional inhibition/competition, and present these as being driven by specific oscillatory failure modes during L2 sentence planning. We argue that modeling CLI in this way not only offers the kinds of linking hypotheses ROSE was built to encourage, but also licenses the exploration of more spatiotemporally complex biomarkers of non-native processing than more commonly discussed neural signatures.
Read the full abstract

Computation and Language

Addressing LLM Diversity by Infusing Random Concepts

Pulin Agrawal and Prasoon Goyal

Large language models (LLMs) are known to produce outputs with limited diversity. In this work, we study whether infusing random concepts in the prompts can improve the diversity of the generated outputs. To benchmark the approach, we design a systematic evaluation protocol which involves prompting an LLM with questions of the form "Name 10 Hollywood actors", and analyzing diversity measures of the resulting LLM outputs. Our experiments on multiple LLMs show that prepending random words/sentences unrelated to the prompt result in greater diversity in the outputs of LLMs. We believe that this promising result and the evaluation protocol opens up interesting avenues for future work, such as how infusing randomness into LLMs could be applied to other domains. Further, the evaluation protocol could also inspire research into benchmarking LLM diversity more systematically.
Read the full abstract

Sunday 25 January

Computation and Language

State Design Matters: How Representations Shape Dynamic Reasoning in Large Language Models

Annie Wong, Aske Plaat, Thomas Bäck et al. (+2 more)

As large language models (LLMs) move from static reasoning tasks toward dynamic environments, their success depends on the ability to navigate and respond to an environment that changes as they interact at inference time. An underexplored factor in these settings is the representation of the state. Holding model parameters fixed, we systematically vary three key aspects: (1) state granularity (long form versus summary), (2) structure (natural language versus symbolic), and (3) spatial grounding (text-only versus images or textual map encodings) across sequential decision-making benchmarks. We find that trajectory summarisation improves performance by reducing noise and stabilising long-horizon reasoning. Second, natural language representations are the most robust across models, whereas structured encodings help mainly for models with strong code or structured output priors, such as JSON schemas. Third, while image-inputs show some benefit, text-based spatial encodings prove most effective. This advantage stems not from the spatial information itself, but from the act of construction, which compels the model to perform the spatial reasoning that static input does not elicit. Overall, we demonstrate that design choices for representing state are a decisive factor in performance, distinct from the availability of information itself. We note, however, that even with improved representations, current LLMs and VLMs remain brittle over long horizons, particularly when they must synthesise information to manage multiple subtasks to reach a goal.
Read the full abstract

All authors Annie Wong, Aske Plaat, Thomas Bäck, Niki van Stein, Anna V. Kononova

Computation and Language

Multi-source Heterogeneous Public Opinion Analysis via Collaborative Reasoning and Adaptive Fusion: A Systematically Integrated Approach

Yi Liu

The analysis of public opinion from multiple heterogeneous sources presents significant challenges due to structural differences, semantic variations, and platform-specific biases. This paper introduces a novel Collaborative Reasoning and Adaptive Fusion (CRAF) framework that systematically integrates traditional feature-based methods with large language models (LLMs) through a structured multi-stage reasoning mechanism. Our approach features four key innovations: (1) a cross-platform collaborative attention module that aligns semantic representations while preserving source-specific characteristics, (2) a hierarchical adaptive fusion mechanism that dynamically weights features based on both data quality and task requirements, (3) a joint optimization strategy that simultaneously learns topic representations and sentiment distributions through shared latent spaces, and (4) a novel multimodal extraction capability that processes video content from platforms like Douyin and Kuaishou by integrating OCR, ASR, and visual sentiment analysis. Theoretical analysis demonstrates that CRAF achieves a tighter generalization bound with a reduction of O(sqrt(d log K / m)) compared to independent source modeling, where d is feature dimensionality, K is the number of sources, and m is sample size. Comprehensive experiments on three multi-platform datasets (Weibo-12, CrossPlatform-15, NewsForum-8) show that CRAF achieves an average topic clustering ARI of 0.76 (4.1% improvement over best baseline) and sentiment analysis F1-score of 0.84 (3.8% improvement). The framework exhibits strong cross-platform adaptability, reducing the labeled data requirement for new platforms by 75%.
Read the full abstract

Computation and Language

Rethinking Soft Compression in Retrieval-Augmented Generation: A Query-Conditioned Selector Perspective

Yunhao Liu, Zian Jia, Xinyu Gao et al. (+2 more)

Retrieval-Augmented Generation (RAG) effectively grounds Large Language Models (LLMs) with external knowledge and is widely applied to Web-related tasks. However, its scalability is hindered by excessive context length and redundant retrievals. Recent research on soft context compression aims to address this by encoding long documents into compact embeddings, yet they often underperform non-compressed RAG due to their reliance on auto-encoder-like full-compression that forces the encoder to compress all document information regardless of relevance to the input query. In this work, we conduct an analysis on this paradigm and reveal two fundamental limitations: (I) Infeasibility, full-compression conflicts with the LLM's downstream generation behavior; and (II) Non-necessity: full-compression is unnecessary and dilutes task-relevant information density. Motivated by these insights, we introduce SeleCom, a selector-based soft compression framework for RAG that redefines the encoder's role as query-conditioned information selector. The selector is decoder-only and is trained with a massive, diverse and difficulty-graded synthetic QA dataset with curriculum learning. Extensive experiments show that SeleCom significantly outperforms existing soft compression approaches and achieves competitive or superior performance to non-compression baselines, while reducing computation and latency by 33.8%~84.6%.
Read the full abstract

All authors Yunhao Liu, Zian Jia, Xinyu Gao, Kanjun Xu, Yun Xiong

Computation and Language · Cross-listed from eess.AS

RIR-Mega-Speech: A Reverberant Speech Corpus with Comprehensive Acoustic Metadata and Reproducible Evaluation

Mandip Goswami

Despite decades of research on reverberant speech, comparing methods remains difficult because most corpora lack per-file acoustic annotations or provide limited documentation for reproduction. We present RIR-Mega-Speech, a corpus of approximately 117.5 hours created by convolving LibriSpeech utterances with roughly 5,000 simulated room impulse responses from the RIR-Mega collection. Every file includes RT60, direct-to-reverberant ratio (DRR), and clarity index ($C_{50}$) computed from the source RIR using clearly defined, reproducible procedures. We also provide scripts to rebuild the dataset and reproduce all evaluation results. Using Whisper small on 1,500 paired utterances, we measure 5.20% WER (95% CI: 4.69–5.78) on clean speech and 7.70% (7.04–8.35) on reverberant versions, corresponding to a paired increase of 2.50 percentage points (2.06–2.98). This represents a 48% relative degradation. WER increases monotonically with RT60 and decreases with DRR, consistent with prior perceptual studies. While the core finding that reverberation harms recognition is well established, we aim to provide the community with a standardized resource where acoustic conditions are transparent and results can be verified independently. The repository includes one-command rebuild instructions for both Windows and Linux environments.
Read the full abstract

Computation and Language · Cross-listed from cs.AI

Sentipolis: Emotion-Aware Agents for Social Simulations

Chiyuan Fu, Lyuhao Chen, Yunze Xiao et al. (+3 more)

LLM agents are increasingly used for social simulation, yet emotion is often treated as a transient cue, causing emotional amnesia and weak long-horizon continuity. We present Sentipolis, a framework for emotionally stateful agents that integrates continuous Pleasure-Arousal-Dominance (PAD) representation, dual-speed emotion dynamics, and emotion–memory coupling. Across thousands of interactions over multiple base models and evaluators, Sentipolis improves emotionally grounded behavior, boosting communication, and emotional continuity. Gains are model-dependent: believability increases for higher-capacity models but can drop for smaller ones, and emotion-awareness can mildly reduce adherence to social norms, reflecting a human-like tension between emotion-driven behavior and rule compliance in social simulation. Network-level diagnostics show reciprocal, moderately clustered, and temporally stable relationship structures, supporting the study of cumulative social dynamics such as alliance formation and gradual relationship change.
Read the full abstract

All authors Chiyuan Fu, Lyuhao Chen, Yunze Xiao, Weihao Xuan, Carlos Busso, Mona Diab

Computation and Language

CommonLID: Re-evaluating State-of-the-Art Language Identification Performance on Web Data

Pedro Ortiz Suarez, Laurie Burchell, Catherine Arnett et al. (+94 more)

Language identification (LID) is a fundamental step in curating multilingual corpora. However, LID models still perform poorly for many languages, especially on the noisy and heterogeneous web data often used to train multilingual language models. In this paper, we introduce CommonLID, a community-driven, human-annotated LID benchmark for the web domain, covering 109 languages. Many of the included languages have been previously under-served, making CommonLID a key resource for developing more representative high-quality text corpora. We show CommonLID's value by using it, alongside five other common evaluation sets, to test eight popular LID models. We analyse our results to situate our contribution and to provide an overview of the state of the art. In particular, we highlight that existing evaluations overestimate LID accuracy for many languages in the web domain. We make CommonLID and the code used to create it available under an open, permissive license.
Read the full abstract

All authors Pedro Ortiz Suarez, Laurie Burchell, Catherine Arnett, Rafael Mosquera-Gómez, Sara Hincapie-Monsalve, Thom Vaughan, Damian Stewart, Malte Ostendorff, Idris Abdulmumin, Vukosi Marivate, Shamsuddeen Hassan Muhammad, Atnafu Lambebo Tonja, Hend Al-Khalifa, Nadia Ghezaiel Hammouda, Verrah 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 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 Rice, Azril Hafizi Amirudin, Jesujoba O. Alabi, Srikant Panda, Yassine Toughrai, Bruhan Kyomuhendo, Daniel Ruffinelli, Akshata A, 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 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 Luger