Peer Revenue Publish or profit

Papers

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

Computer Science Computation and Language

cs.CL Watch Computation and Language

Monday 12 October

Computation and Language

From Sampled Evidence to Radiology Reports: Objectives, Constraints, and Realization

Kun Zhao, Haoteng Tang, Chenghua Lin et al. (+1 more)

Sampling multiple reports creates an inference problem: how should sampled evidence be converted into a final decision, and does a better structured decision yield a better report? We study these questions through a framework that separates the decision signal, scoring objective, and feasible decision set, followed by text realization. On MIMIC-CXR and IU-Xray, controlled comparisons across 13 generator configurations per benchmark under grouped development/test evaluation reveal three findings. First, changing the objective under fixed evidence and rule capacity changes predictions and their held-out metric trade-offs. Second, relaxing candidate constraints has objective-dependent effects under shared signals and frozen scores. Diagnostics connect the IU macro-oriented assertion increase to near-zero label thresholds and show that limited candidate coverage coexists with selection headroom. Third, a controlled realization study with SFT-Qwen3-VL-8B shows that decision-level gains can be attenuated, amplified, or reversed. Change decomposition reveals that realization can remove incorrect assertions while losing correct findings, so target fidelity and reference accuracy need not improve together. Finding gains also do not consistently improve reference-based report quality. These results motivate separate evaluation of decision rules, candidate constraints, and realized reports.
Read the full abstract

All authors Kun Zhao, Haoteng Tang, Chenghua Lin, Liang Zhan

Computation and Language

StartLux-Decision: AutoThink Decision Models with Reinforcement Learning for Calibrated Deliberative Decisions

Xialie Zhuang, Zunhai Su, Jia Pan et al. (+3 more)

We introduce StartLux-Decision, decision models from 0.8B to 27B parameters that combine efficient single-pass prediction with adaptive, confidence-guided reasoning. The Instant models use bidirectional attention in full-attention layers to improve decision accuracy while keeping single-pass inference. The AutoThink models invoke explicit reasoning for uncertain decisions and answer confident cases directly. A three-stage training pipeline establishes shared decision capabilities, specializes the two variants, and equips AutoThink with reasoning through reasoning supervision and reinforcement learning. We introduce Reinforcement Learning for Calibrated Deliberative Decisions (RLCD$^2$), which uses a frozen model and proper scoring rules to reward improvements in decision probability quality and applies a soft penalty to overlong reasoning traces. RLCD$^2$ raises the share of reasoning traces that terminate within the generation limit by up to 48 percentage points and cuts average reasoning length by up to 65%, while largely preserving decision accuracy. On the public subset of Decision Index 0.3, the Instant models outperform the strongest evaluated comparable-scale open-weight decision models by 7.4-16.9 points, and the 27B AutoThink model reaches 73.4. We also introduce DecisionPivot, a programmatically verifiable benchmark of computational decisions built from contrast pairs that differ in a single answer-changing detail. On DecisionPivot, AutoThink outperforms its Instant counterparts by 11.5-18.4 points in pair accuracy and leads among evaluated systems. The models run on vLLM, llama.cpp and MLX, with quantized variants for lower memory use.
Read the full abstract

All authors Xialie Zhuang, Zunhai Su, Jia Pan, Ngai Wong, Guanhua Chen, Chuan-Wei Kuo

Computation and Language

Nufo: A Benchmark for Ewe Automatic Speech Recognition

Christ Kekeli Keli, Steve Sanogo, Patrice Sebastiano et al. (+2 more)

Automatic speech recognition (ASR) remains challenging for low-resource languages, where the scarcity of transcribed speech limits the development of accurate systems. This challenge is particularly acute for African languages such as Ewe, which has millions of speakers but remains underrepresented in publicly available ASR resources and benchmarks. In this work, we consolidate, normalize, and partition existing Ewe speech resources into a standardized 174-hour Ewe ASR benchmark, the largest to date to our knowledge. We fine-tune and evaluate pretrained speech models, including Whisper and AfriHuBERT, to establish strong baselines on our benchmark. Whisper-medium achieves a word error rate (WER) of 37.98%, while AfriHuBERT achieves 39.70% on the test set. We further provide a qualitative analysis of common recognition errors. We release the benchmark under the name Nufo, derived from the Ewe word for "talk", accompanied by models and evaluation resources, to facilitate future research on Ewe and other underrepresented African languages.
Read the full abstract

All authors Christ Kekeli Keli, Steve Sanogo, Patrice Sebastiano, Hamad Aldenawi, Ryan Whetten

Computation and Language

From Conceptual EneMonIE to an Executable Backend: A Modular Architecture for Energy Information Extractionand Grounded Question Answering

Janik Lesage, Khadim Lô and and Jean-Sébastien Dessureault

The original EneMonIE proposal describes a Web-based semantic system for continuously monitoring energy trends through targeted information extraction from heterogeneous online media. This article presents EneMonIE-Backend, an implementation-derived redesign that converts that conceptual pipeline into a modular, executable, and application-independent backend. The architecture separates five concerns: data sources, acquisition and normalization, semantic processing, knowledge organization, and persistent storage. It operationalizes configurable connectors for RSS feeds, URL collections, arXiv, social-media JSON files, and open-data CSV files; it also introduces question-driven Web search and scraping. Relevant documents pass through energy categorization, named entity recognition, temporal and event extraction, source sensitive sentiment analysis, and extractive summarization. The resulting semantic representations support entity co-occurrence graphs, cross-document interlinking, cluster-level information fusion, and relational persistence in SQLite. Three grounded question-answering paths are exposed at the backend level: live Web evidence, the stored corpus, and selected configured resources. Answer generation uses a language model when configured and otherwise falls back to deterministic extractive output. Unlike the original proposal, the presented architecture deliberately excludes the user-interface and server-side application layers, allowing the semantic core to be embedded in different applications. A deterministic smoke validation confirms relevance filtering, persistence, clustering, evidence retrieval, and document deletion. The work provides a reproducible architectural basis for future evaluation on larger multilingual energy corpora.
Read the full abstract

Computation and Language · Cross-listed from cs.LG

Online LLM Personalization from Pairwise Preferences Across Users

Romain Puech, Georgy Salakhutdinov and Dimitris Bertsimas

Large language models can answer the same question in different styles, but users' preferred styles are hidden and vary across users and questions. We study inference-time personalization through online selection from a fixed set of prompts, each representing a persona. Users arrive sequentially, each asks a short sequence of questions, and the system displays two responses per question and observes one pairwise choice. We formulate this interaction as a variant of the contextual dueling bandit problem in which the learner starts without preference data, adapts within each session, and exploits information shared across users. The objective is to minimize cumulative strong regret, summing the utility gaps of both displayed responses relative to the user's best response among the available options. We make two contributions. First, we introduce the problem and release a reproducible preference benchmark and dataset built around a tutoring case study, containing educational questions, LLM-generated answers in six teaching styles, and pairwise judgments from LLM-simulated students with hidden preference profiles. We fit Bradley-Terry models to these judgments to obtain reference utilities for regret evaluation. Second, we propose a baseline that learns a mixture prior over users from completed sessions and adapts to each new user within the session. On students and topics never used in development, it removes 44-48% of random pairs' excess regret and achieves over half the performance of a per-user linear oracle fitted in hindsight. We release code, prompts, and the benchmark.
Read the full abstract

Computation and Language · Cross-listed from cs.LG

MLAfication: Efficiently Enabling High-Fidelity MLA for Modern LLMs

Zile Wang, Zhide Xie and Tao Ji

Multi-Head Latent Attention (MLA) compresses the KV cache into a latent representation, reducing inference memory and bandwidth. Existing MHA2MLA/TransMLA methods do not support QKNorm and require costly recovery on modern LLMs, especially MoE. We present MLAfication, a QKNorm-compatible, parameter-efficient extension of MHA2MLA. It combines calibrated KNorm, parameter-efficient distillation, contribution-aware Partial-RoPE, and activation-aware joint KV initialization. Across MHA and GQA models, including QKNorm- and MoE-equipped variants, MLAfication preserves performance under substantial KV-cache compression, surpasses the current SOTA on commonsense and long-context benchmarks, and extends MLA conversion to modern architectures. On Qwen3-30B-A3B, MLAfication delivers up to $5.35\times$ the serving throughput of BF16 GQA and, combined with 4-bit KV quantization, reduces the KV-cache footprint by $94.73\%$. Our code is available at https://github.com/JT-Ushio/MLAfication.
Read the full abstract

Computation and Language

Welcome to the Age of Agent Ads: Using On-Policy Attacks to Steer Agentic Decisions

Tim R. Davidson, Marija Šakota, Joshua Joseph et al. (+1 more)

People increasingly delegate option-ranking tasks (e.g., shopping, job search, travel) to AI agents, which creates incentives to steer agentic decisions via "agent ads." To defend against such steering attacks, people count on models' alignment training and platforms' terms of service, and consequently the most effective existing attacks are designed to trigger off-policy behaviors. Exposing a new attack surface, we introduce on-policy attacks, which steer agentic decisions without pushing individually aligned behaviors off policy. Our example attack consists of two steps: To remain compliant with platforms' terms of service, we first inject a truthful negation about a fictitious concept ("Does not contain blizyester") into a product description. This triggers an agent to search the web ("What is blizyester?"), which surfaces planted websites full of fabricated information associating the fictitious concept with harmful implications ("Blizyester linked to dementia"). Trained to keep users safe from harm, the agent penalizes every option missing the truthful negation. Evaluating our attack across four real-world listing datasets, we find some frontier models exposed, enabling an adversarial actor to penalize competitors without violating existing platform-listing guidelines. We discuss the broader challenge of regulating advertising aimed at AI agents and provide recommendations to model developers and platform providers.
Read the full abstract

All authors Tim R. Davidson, Marija Šakota, Joshua Joseph, Robert West

Computation and Language

Soft Teacher Forcing

Mohammad Fahes, Cordelia Schmid and Anurag Arnab

Soft tokens have so far been used to replace discrete chain-of-thought in latent reasoning methods. These approaches typically require multi-stage training and, in some cases, auxiliary modules, yet they still underperform supervised fine-tuning (SFT) on <input, chain-of-thought, output> triples. In this paper, we ask whether soft tokens can be used to improve SFT itself, instead of as an avenue for latent reasoning. We introduce Soft Teacher Forcing (Soft-TF), a simple training algorithm that combines teacher-forced embeddings with soft tokens through random convex combinations. The soft tokens are computed in a single parallel forward pass over the ground-truth sequence. Soft-TF adds no parameters, no external model and no curriculum, and leaves inference unchanged. Experiments show that Soft-TF improves over SFT on GSM8K, MATH, MathQA and AQuA, by +2.1 to +4.1 points on GSM8K across seven Gemma, Llama and Mistral models.
Read the full abstract

Computation and Language

DuoDiscover: Scaling Autonomous Discovery Beyond the Evaluation Bottleneck

Shubham Agarwal, Akshat Gupta, Mike Heddes et al. (+21 more)

Evaluation has been central to advances in AI, and it is equally critical to autonomous discovery. Yet scaling autonomous discovery to real-world engineering challenges faces a fundamental bottleneck: the cost of evaluating progress. In real settings, a single true evaluation, such as a full benchmark or serving trace, can take hours, so a search can afford only a handful of true evaluations. We present DuoDiscover, a search system that jointly discovers what to optimize and how to evaluate progress efficiently, through three mechanisms. First, each worker in a fleet of coding agents co-discovers an optimization target and a fast, tailored proxy evaluator, which we call a yardstick. The agent then hill-climbs the yardstick rather than the true evaluation. Second, since search can exploit gaps between a yardstick and the true objective, fidelity tests continually recheck each yardstick's ranking and invalidate any that drift. Third, the fleet is bottom-up, with no central decomposition of the problem. A lightweight orchestrator spawns and terminates agents, briefs running ones, and seeds new agents with others' confirmed work. DuoDiscover enables autonomous optimization from software to chip architecture. We demonstrate DuoDiscover on two repository-level benchmarks and four GPU and TPU kernel suites covering hundreds of kernels. On SWE-fficiency, it doubles the average speedup of human experts. On SOL-Exec, it ranks first on 121 of 235 kernels and in the top three on 212. On the heavily optimized vLLM and SQLite engines, DuoDiscover achieves 1.92$\times$ and 1.38$\times$ gains over the base code. On a CPU microarchitecture optimization task with evaluations of up to 28 hours, DuoDiscover surpasses both AlphaEvolve-based ArchAgent v2 and the best human design, with 1.65$\times$ ArchAgent v2's gain in 16.7 days versus more than 40 days, advancing the frontier of autonomous chip design.
Read the full abstract

All authors Shubham Agarwal, Akshat Gupta, Mike Heddes, Dilara Soylu, Rui Meng, Patrick Tser Jern Kon, Jiefeng Chen, Mert Cemri, Shu Liu, Abe Gonzalez, Akanksha Jain, Bhavana Dalvi Mishra, Linfeng Song, Vilobh Meshram, Xiangru Tang, Jinsung Yoon, Aditya Parameswaran, Ion Stoica, Vahab Mirrokni, Burak Gokturk, Parthasarathy Ranganathan, Tomas Pfister, Nanyun Peng, Chun-Liang Li

Computation and Language · Cross-listed from cs.AI

Learning Perceptual Geometry of Natural Odor Mixtures from LLM-Derived Descriptors

Farhad Pashakhanloo and Venkatesh N. Murthy

The relationship between odor chemistry and perception remains incompletely understood. Many existing studies focus on the perception of single molecules, whereas natural odors contain tens to hundreds of components, adding extra layer of complexity to percept formation. Here, we explore the relationship between molecular composition and perceived similarity in natural odor objects. We study this using a database of more than 1,200 volatile-compound-in-food mixtures containing more than 8,000 molecular components, and use large language models (LLMs) as a scalable surrogate for deriving mixture-level perceptual descriptions. We find that language-derived representations provide useful supervision for predicting held-out mixture similarities from molecular composition. The prediction was consistently strongest when perceptual targets were derived from chemosensory prompts than from other sensory prompts. Further, during the mixture-level training, we learned low-rank molecular embeddings that enabled better prediction of held-out mixture similarities than component-overlap metrics, even for mixture pairs with no shared molecules. Although no single-molecule labels were used during training, the learned molecular geometry showed significant alignment with the independent human odor-descriptor space. Finally, we explored the use of external molecular features in our mixture-similarity prediction task. These features showed modest zero-shot alignment with mixture-level perceptual similarity. They improved prediction under limited supervision but offered little additional predictive benefit with the full training set. Together, these results show that language-derived descriptions can provide scalable supervision for predicting the perceptual organization of complex natural mixtures, contributing to a quantitative understanding of how molecular composition gives rise to odor perception.
Read the full abstract

Computation and Language · Cross-listed from cs.AI

The Format is the Filter: Scoring Format Fixes What an LLM Judge Can Express

Lovedeep Gondara, Jekaterina Novikova, Sagarika Suresh et al. (+2 more)

When an LLM judge filters instruction-tuning data, its scoring format (binary verdict, 1–5 rating, 0–100 score, or comparison) is rarely justified or ablated. Holding the candidate pool, judge prompt, decoding and selection budget fixed, we score 15,406 UltraFeedback candidates under five formats with three judges, curate, fine-tune 66 models, and evaluate them with four judges. A 0–100 scale is worth only 7–12 equally used levels, two to three times a 1–5 scale rather than twenty, and a binary verdict leaves the top score tied in 77–78% of prompts, so most selections are made at random. These properties are set by the format almost independently of the judge. Kendall's $\tau_b$ hides them: it ranks binary best while $\tau_a$ and top-1 agreement rank it last, because it discounts the ties a selector must break at random. Both comparative formats inherit a slot bias whose sign differs across judges. Downstream, curated data beat random selection under every evaluation judge, robust to length and markdown adjustment, but no ordering among formats is detectable. We propose four quantities that judge-based curation pipelines should report.
Read the full abstract

All authors Lovedeep Gondara, Jekaterina Novikova, Sagarika Suresh, Jack Ji, Jithin Pradeep

Computation and Language

ToolHACE: A Dataset for Detecting Hallucinations in Tool-Based LLM Responses

Nikolay Ivanov, Sergey Pletenev, Elisei Rykov et al. (+8 more)

Tool-augmented language models ground their responses in external tool outputs, improving reliability beyond parametric knowledge alone. However, even when the correct tool is called and returns accurate information, models may still contradict such outputs, omit critical details, or generate unsupported claims. Existing hallucination benchmarks for tool-augmented dialogue primarily focus on the tool-calling stage, evaluating whether the correct tool and arguments are selected, while overlooking errors that arise during post-tool response generation. To address this gap, we introduce ToolHACE, a span-level benchmark for hallucination detection in tool-augmented LLM responses, covering five fine-grained hallucination categories. Through extensive experiments with modern hallucination detectors, we show that post-tool-call hallucinations are widespread and remain challenging to detect. Our results further demonstrate that effective detection requires training on specialized tool-augmented corpora, while transfer from tool-agnostic hallucination datasets performs poorly in this setting. These findings highlight the need for dedicated benchmarks and models for reliable hallucination detection in tool-augmented language systems.
Read the full abstract

All authors Nikolay Ivanov, Sergey Pletenev, Elisei Rykov, Anatolii Frolov, Bogdan Monogov, Alexander Potekhin, Maria Marina, Daniil Moskovskiy, Viktor Moskvoretskii, Alexander Panchenko, Ilseyar Alimova

Computation and Language

D2: High-Capability Full-Duplex Spoken Language Models from Half-Duplex Backbones

Cheng-Kuang Chang, Alexander H. Liu, Brian Kingsbury et al. (+5 more)

Full-duplex spoken language models (FD-SLMs) support simultaneous listening and speaking, but often lag behind half-duplex spoken language models (HD-SLMs) in general capabilities. We propose Dialogue-to-Duplex (D2), a framework that converts pretrained HD-SLMs into high-capability FD-SLMs at fine temporal granularity. D2 adapts the original audio encoder to streaming through decoder-free representation distillation and learns full-duplex interaction through supervised fine-tuning. We systematically study the effects of chunk size on full-duplex interaction, general capabilities, and inference computation. We apply D2 to two distinct HD-SLM architectures, Qwen3-Omni and LLaMA-Omni 2. At a 12.5Hz update rate, Qwen3-Omni-D2 achieves 99.2% turn-taking success with 521-ms latency, 99.5% interruption success, and a state-of-the-art VoiceBench score of 70.1.
Read the full abstract

All authors Cheng-Kuang Chang, Alexander H. Liu, Brian Kingsbury, Hilde Kuehne, Rogerio Feris, Samuel Thomas, James Glass, Kai-Wei Chang

Computation and Language · Cross-listed from cs.AI

When LLM Agents See a Conflict but Still Proceed: Cross-Domain Dependencies in Network Slicing

Marcos Lima Romero and Ricardo Suyama

Future AI-native networks are being designed to apply autonomous agents built on large language models (LLMs) in network management through intents. When intents from different domains meet on the same resource, an agent must recognize the cross-domain dependency before taking an irreversible action. We investigate this problem through a smart-grid use case, in which a power utility holds an active intent on a backup slice for protection traffic. This slice appears idle by design, which is precisely the profile targeted by a cost-reduction intent. We built a tool-based environment in which the dependency must be either queried by the agent or read from an inventory annotation, together with a negative control, and evaluated ten models from four different providers in 600 runs. Most models acted against the dependency they had observed: seven of the ten models removed the protection slice at least once after receiving the cross-domain dependency, and in the high-harm tier the slice was removed in 76% of the exposed runs, while four models practically never did. Only one of ten models never removed the slice. We identify two failure modes, not querying the dependency and querying it but proceeding anyway, and show that agents often reported no conflict after observing it. These results indicate that, in AI-native networks, visibility of cross-domain dependencies is not a sufficient safeguard, and that dependency verification should be enforced by the management architecture before irreversible actions rather than delegated to the LLM's judgment.
Read the full abstract

Computation and Language

Gluing Local Contexts into Global Meaning: A Sheaf-Theoretic Decomposition of Transformer Representations

Bryce Grant and Peng Wang

Existing interpretability tools do not test whether a representation keeps the same content when only the phrasing changes. We treat each paraphrase pair as a local witness that two activations should agree and build a cellular sheaf over the paraphrase graph; its Laplacian $L_\mathcal{F}$ answers the question on a frozen model with no training. The kernel $H^0$ is a content subspace: at 20 dimensions it beats the preserved subspace of the concept eraser LEACE on held-out CounterFact fact retrieval across eight architectures by a mean $+17.1$\,pp, and constraining all paraphrases of a fact jointly adds a further $+6.5$\,pp. The spectral complement concentrates causal influence: ablating the most paraphrase-variable coordinates, a coordinate-wise proxy for it, changes model outputs $5.6$ to $26.5\times$ more than ablating coordinates of matched per-coordinate variability, across nine models from 124M to 13B parameters. Stacking layers into a grid creates $b_1 = N(L-1)$ algebraic 1-cycles, and the resulting model-derived harmonic mass varies by nearly three orders of magnitude across six architectures and ranks them as their steering fragility does (Spearman $\rho = 0.90$, exact one-sided permutation $p = 0.014$). The same paraphrase graph yields the content subspace, the influence directions, and the fragility measure. Project page: https://cwru-aism.github.io/gluing-lc-page/
Read the full abstract

Computation and Language

Calibrated Typed Decisions for Entity, Relation, and Event Extraction

Yang Wang, Hanlin Xiao, Yanan Ma et al. (+2 more)

Information extraction models often perform well on the datasets they are trained on, but transferring to new domains and annotation guidelines remains challenging. Supporting user-specified types alone does not resolve this problem. We present CalOn, a unified 0.6B-parameter encoder for entity, relation, and event extraction across datasets and schemas. The model jointly encodes text and type definitions, then directly identifies entity spans, relations between spans, and event triggers and arguments. Its extraction decisions consider the requested types together, using their combined evidence to identify relevant candidates before assigning a type. This gives all three tasks a common probabilistic formulation while keeping every extracted mention anchored to the input text. We train on diverse sources, and supplement supervised learning with a reward that encourages correct extractions and penalises incorrect ones. Across 40 benchmarks, CalOn outperforms the evaluated baselines on most benchmarks. It leads all reported relation extraction benchmarks and achieves the highest average F1 for event triggers and arguments with gold triggers. The same model also performs strongly on aspect sentiment triplet extraction and pronoun resolution through its span-pair formulation. A correction for the probability bias introduced during training improves confidence estimates for entities, relations, and event arguments after calibration, with little change in F1.
Read the full abstract

All authors Yang Wang, Hanlin Xiao, Yanan Ma, Chenghao Xiao, Chenghua Lin

Computation and Language · Cross-listed from cs.LG

Towards Universal Calibration for In-Context Classification

Awik Dhar, Sunita Sarawagi and Arpit Agarwal

In-Context Learning (ICL) has received a lot of interest recently due to its ability to improve classification accuracy for new tasks without having to fine-tune the model over these tasks. However, the impact of ICL on calibration has been far less studied. Recent work has shown that ICL can have a negative impact on calibration due to biased selection of in-context examples. The existing methods for improving ICL calibration require additional task and model-specific training, detracting from the no-fine-tuning convenience of ICL. In this work, we introduce CALIBFORMER, a universal ICL calibrator for classification tasks, that generalizes to new models and tasks without further fine-tuning on these tasks. Specifically, we design a transformer-based calibrator which uses the model predictions on in-context examples to predict a confidence score for the test input. We show that the calibrator trained on three models and ten tasks, is able to generalize/calibrate ICL predictions on several new tasks and four LLMs unseen during training. Even for frontier models like Gemini, CALIBFORMER is able to obtain over 50% reduction in calibration error.
Read the full abstract

Computation and Language

UZHCL_for_the_win at Touche: Encoder Fine-Tuning, Synthetic Augmentation, and an Underpowered Encoder-LLM Comparison

Psychias Stylianos

Online discussions on platforms like Reddit are a productive testbed for fallacy detection: short texts, broad topical range, and the full inventory of informal fallacies. The Touche 2026 lab at CLEF formalises this as a shared task on a Reddit-derived corpus that combines fallacy detection with argument-scheme classification. This paper describes our participation in all three subtasks (ST1 binary fallacy detection, ST2 eight-way fallacy classification, ST3 four-way argument-scheme classification) and both evaluation tracks. Our submitted systems are fine-tuned RoBERTa-large encoders with task-specific heads, augmented with a small batch of LLM-generated synthetic contrastive pairs; for ST2 and ST3 on the enhanced track we additionally added confidence-filtered pseudo-labelled examples from the unlabelled pool. A wider backbone comparison spanning additional encoders (DeBERTa-v3, ModernBERT) and an LLM track (four Qwen variants from 3B to 32B, zero-shot through CoT-LoRA) showed no validated benefit over RoBERTa-large at this corpus scale, with 5-fold variance wide enough that we call the comparison underpowered rather than settled. We also characterise how much of the enhanced track's lift reflects organiser-rewritten inputs carrying label information rather than genuine argument clarification. On the official Touche 2026 test set our systems ranked first on both base-track classification cells (ST2 and ST3) and second on both binary-detection (ST1) tracks, with the two enhanced-track classification cells topped by cascaded and multitask systems. We close with the rare-class limitation on ST3 and the most useful follow-ups. Code: https://github.com/psychias/fallacy_detection_clef_2026
Read the full abstract

Computation and Language

AD4EDU: Benchmarking University Lecture Audio Description across Accessibility Standards, Metrics, and Learner Preferences

Psychias Stylianos and Gao Yingqiang

Audio description (AD) aims to enhance the accessibility of visual media for blind and low-vision (BLV) persons. However, in higher education, where slide-based lectures are a default mode of teaching, lecture AD remains largely underexplored. The main reason is that existing production guidelines, evaluation metrics, and benchmark datasets are oriented toward entertainment content and do not transfer to lectures, where ADs must complement the lecturer's speech, resolve deictic references, and fit within narrow pauses. In this work, we propose AD4Edu, consisting of an AD benchmark, AD4Edu-Bench, and a rule-diagnosable evaluation metric, AD4Edu-Eval, for generating and evaluating ADs in higher education. Specifically, we consolidated established accessibility guidance, including WCAG 2.2, the DCMP Description Key, and NCAM's STEM description practices. We further collected silver reference ADs, judge-ordered preference pairs, and pairwise preferences from four BLV raters, and post-trained models with direct preference optimization (DPO) for detailed analysis. Our findings indicate that while the metric reliably detects violations of the standard, ADs that comply with it do not necessarily fulfill the subjective preferences of BLV persons, most notably on description length, underscoring the necessity of active participation of BLV persons in co-designing inclusive AI solutions.
Read the full abstract

Computation and Language

Beyond Model Bias: Cross-Entropy Based Rankings in Literary and Academic Texts Remain Stable across Small Language Models

Ankit Bhattacharjee, Indrasabarna Hari, Krityapriya Bhaumik et al. (+2 more)

While language models provide powerful mechanisms for evaluating textual predictability, it remains unclear whether the relative extents of LM surprisal remain stable across models. This study presents empirical evidence of the invariance of ranking based on cross-entropy related measures (e.g. mean, variance of each data-point), across four distinct small scale transformer-based models, for a given dataset of text sources. We consider two stylistic environments: a corpus of contemporary poetry and a corpus of academic research papers. We demonstrate that mean and variance of cross-entropy maintain robust cross-model rank correlation ($\rho > 0.90$) across both regimes. To test whether this agreement merely reflects generic differences in textual complexity, we compare it against six parameter-free compression baselines. The extent of agreement among language models is found to be substantially higher than that among model-baselines, for most of the measures. Geometric analysis of the structurally constrained academic regime reveals that higher-order moments are sensitive to specific models, although strong agreement persists. The findings suggest that relative extents of textual unpredictability may be independent of the language model they are computed against.
Read the full abstract

All authors Ankit Bhattacharjee, Indrasabarna Hari, Krityapriya Bhaumik, Arka Das, Piyush Sarangi

Computation and Language

MENTOR: Procedural Shielding of Tool-Using LLM Agents

Ashwani Anand, Ivi Chatzi, Ritam Raha et al. (+1 more)

Tool-using LLM agents are increasingly used to automate workflows governed by procedural manuals, such as company rules for booking travel or procuring hardware. These manuals are written in natural language for humans and describe general requirements over such complex workflows. Ensuring compliance of LLM agents with these manuals is difficult, and state-of-the-art techniques use either LLM judges or manually written rules to forbid problematic tool-calls locally at runtime. This is insufficient because (i) only blocking tool calls keeps the agent clueless about how to make correct progress towards fulfilling its task, and (ii) can lead to overly conservative behavior of the agent, as tool calls might only be problematic in certain contexts. To overcome these challenges, we introduce MENTOR, which (i) additionally provides guidance for the agent to make procedural-compliant progress on the task, and (ii) ensures that this compliance is context-aware. We achieve these goals by building a novel history-dependent runtime shield that is obtained from the procedural manual via a multi-step autoformalization pipeline. At its core, MENTOR utilizes a game-based technique to ensure robustness against runtime uncertainty without over-resticting agents' behavior. This yields a formally assisted framework for procedural shielding of LLM agents that improves compliance well beyond existing methods. In particular, MENTOR lifts the compliance of shielded light-weight open-source LLMs above that of non-shielded proprietary models. Since a shield is synthesized once per manual and then serves every task, a MENTOR-shielded agent becomes a viable choice for repeatedly executed workflows, potentially reducing deployment costs when the offline construction cost is amortized over repeated tasks.
Read the full abstract

All authors Ashwani Anand, Ivi Chatzi, Ritam Raha, Anne-Kathrin Schmuck

Computation and Language · Cross-listed from cs.AI

Understanding Personalization Gains in Whisper

Sergei Katkov and Antonio Liotta

Whisper-style ASR models provide strong zero-shot transcription, but their residual errors are not evenly distributed across datasets, recording conditions, and metadata-based groups. We study when lightweight low-rank adaptation (LoRA) improves Whisper and whether observed gains require group-specific adapters. Across Common Voice, LibriSpeech, and VoxPopuli, we compare zero-shot Whisper, pooled LoRA, and group-specific LoRA under matched within-group splits. We find that adaptation is not uniformly beneficial: gains are concentrated in difficult Common Voice groups, are consistent on VoxPopuli, and largely disappear on low-error LibriSpeech. Crucially, pooled LoRA often matches, and sometimes clearly outperforms, group-specific LoRA. Gains over zero-shot should therefore be tested against matched pooled adaptation before being interpreted as evidence of personalization.
Read the full abstract

Computation and Language

Predicting Mental Health Help-Seeking Behaviour on Reddit: A Survival Analysis of Linguistic and Behavioural Signals

Nisar Ahmed, Ammara Kazmi, Irtaza Bashir Raja et al. (+1 more)

Most prior work on predicting mental health help-seeking treats it as post-hoc classification of users already active in clinical communities, which cannot tell whether someone approaching a transition can be identified in advance. We instead frame help-seeking as a time-to-event problem and ask which Reddit users will eventually seek mental health support, and how imminent that transition is. We build a two-arm cohort: a control frame of 6,337 ordinary users with genuine right-censoring (12.1% transition rate) to answer who will transition, and a clinical frame of 5,023 eventual transitioners to answer how soon. Features come only from posts written before any clinical activity, and user-grouped five-fold cross-validation plus a four-year temporal holdout guard against leakage. We compare TF-IDF, MentalBERT embeddings and behavioural activity features using Cox proportional hazards models, Random Survival Forests and an LSTM sequential model. Behavioural features alone outperform text for predicting who will seek help (C-index 0.797 vs. 0.763), and combining both is best (0.823 under cross-validation, 0.856 on the holdout). The pattern reverses for predicting how soon a known transition will happen: text remains a stable predictor (C-index 0.698) while behavioural features become unstable. Structural engagement patterns therefore carry information about future help-seeking beyond linguistic content, and separating who from when yields a cleaner analysis.
Read the full abstract

All authors Nisar Ahmed, Ammara Kazmi, Irtaza Bashir Raja, Didier Devaurs

Computation and Language

Event Memory and Narrative Dynamics: A Large Language Model Framework for Multidimensional State Spaces

Joshua Sariñana

Studying event memory requires defining events and their relationships within remembered experience. Autobiographical narratives provide evidence of this organization, but their analytical units must first be established. This study introduces a large language model framework for constructing event representations from narratives and modeling their structure and dynamics with excerpt-level traceability. Explicit boundary rules and a feature schema informed by hippocampal models of episodic memory and navigation guide event construction. In a proof-of-concept application to interviews from five participants, the framework constructed 256 event-states. Multiple correspondence analysis embedded these states in feature-based spaces, with analyses focused on a selected 4-D geometry and 10-D and 13-D reference geometries. Consecutive event-states were significantly closer than expected under randomized narrative order. In the selected 4-D ten-basin solution, similarity between basin-state feature profiles corresponded with similarity between one-step routing profiles; context, mechanism, and theme differentiated full-system pathway engagement. A successor representation modeled expected basin-state occupancy across successive steps. Correlations between basin-centroid distances and complete successor-profile distances were positive across all five basin solutions, with exploratory tests supporting correspondence in at least two solutions across all three geometries. These findings connect feature-based event representations with observed narrative transitions and modeled multistep organization. The framework translates hippocampal principles into source-linked event boundaries and state representations, providing an analytical basis for testing neural accounts of event segmentation and predictive organization during naturalistic recall.
Read the full abstract

Computation and Language

SarcasTraps: Effect of Sarcasm on Turkish Hate Speech Detection

Niyazi Ahmet Metin and Dilara Keküllüoğlu

Standard hate speech detection datasets often emphasize explicit harmful language, but directed harm can also be expressed indirectly, without clearly hateful wording. Sarcasm is one such form of indirect expression, as it can convey harmful intent through mock praise, exaggeration, humor, or apparent politeness. In this paper, we introduce SarcasTraps, a targeted external test set for analyzing the effect of sarcasm on Turkish hate speech detection. SarcasTraps contains 479 entries annotated along two dimensions: hate speech and sarcasm. We train classifiers using BERTurk and XLM-RoBERTa on two existing Turkish hate speech datasets and evaluate them on SarcasTraps, using it strictly as an external test set. We also evaluate GPT-4.1 and Llama-3.3-70B in a zero-shot setting on SarcasTraps. Results show that sarcasm substantially affects hate speech detection. Fine-tuned encoder models lose recall on sarcastic hate, while zero-shot LLMs maintain higher recall but suffer substantial precision drops on sarcastic examples. These findings show that both fine-tuned hate speech detectors and zero-shot LLMs struggle when sarcasm is involved, and SarcasTraps provides quantitative evidence of this challenge through a targeted Turkish test set. Due to the presence of sensitive and hateful language, SarcasTraps will be shared upon request for academic research purposes. Warning: This paper has examples from the dataset which contains offensive and hateful content towards groups of people.
Read the full abstract

Computation and Language

Lexical Redefinition Reveals Context-Sensitive Instruction Following

Roksana Goworek and Haim Dubossarsky

How do language models reconcile conflicting contextual evidence within a prompt? We study this question using lexical redefinition, where an explicit instruction specifies how a word should be interpreted and a subsequent sentence provides additional evidence through its usage. Across 870 word pairs, we find that lexical redefinitions are not applied as fixed rules. Instead, the behaviour depends systematically on the relationship between the explicit instruction and the intervening sentence: the contextualised usage can preserve, weaken, or completely reverse the instruction. These results suggest that the interpretation of an explicit instruction remains sensitive to subsequent contextual evidence. Lexical redefinition therefore provides a controlled behavioural paradigm for studying how language models integrate multiple sources of information within a prompt.
Read the full abstract

Computation and Language

Estimating Uncertainty of Omnimodal Large Language Models via External Omnimodal Embeddings

Iacopo Ghinassi, Hy Nguyen Gia La and Luke Ong

Recent work has investigated Omni-modal Large Language Models (OLLMs), models which can take multiple modalities and their combinations as input, as a way forward to have systems really capable of interacting with their environment. Uncertainty quantification (UQ) for OLLMs, however, remains an open problem: existing methods largely target at most two modalities at a time, require access to model internals, or re-invocation of the evaluated model. We propose a model-agnostic UQ framework that scores a Chain-of-Thought response entirely through external omni-modal embeddings, measuring the grounding of the reasoning in the multimodal input and its internal coherence. Our contribution is the family of signals and an empirical characterisation of when it is reliable, rather than any single estimator: we find the geometric scores to be largely redundant with one another (mean within-split rank correlation 0.91-0.94) but strongly complementary to answer-dispersion baselines, and their reliability to depend on the generator and the modality. The resulting confidence estimates require no access to logits, hidden states or gradients of the evaluated model and no additional inference from it; they do require segmenting the reasoning trace with an auxiliary model and, for the aggregated score, a labelled calibration set that we show can be as small as 25 examples (although we also show that such a set buys little over fixing one signal a priori). On UNO-Bench, with both a closed and an open model, our scores yield small gains over self-consistency (systematic with the closed generator, mixed with the open one) and are competitive with self-verbalisation and internal-state baselines, especially in open-ended, omni-modal questions where previous state-of-the-art baselines struggle.
Read the full abstract

Computation and Language

SpecGuardBench: Can Coding Agents Construct Reliable Validation Harnesses from Natural Language Specifications?

Zhiqing Zhong, Mingyu Chen, Aoyang Fang et al. (+5 more)

As coding agents become more capable, tests remain the primary evidence for judging whether their code is correct. Yet passing available tests does not establish that an implementation satisfies its natural language specification. Reliable validation requires checks that faithfully capture the specification and scenarios that thoroughly exercise the code, but current benchmarks provide limited insight into whether agents can construct such evidence themselves. We introduce SpecGuardBench, a benchmark of 291 tasks from 13 real-world repositories, covering 8 types of verification logic such as functional postconditions and interaction protocols. Given a natural language specification, repository context, and a reference implementation, an agent constructs a validation harness comprising executable checks and test scenarios. The checks should reject implementations that violate the specification while accepting those that satisfy it, and the scenarios should thoroughly exercise the code. Across four proprietary and open-weight coding agents, none constructs a harness scored as faithful on the benchmark's implementations for more than 63% of the tasks. The bottleneck lies less in exploring test scenarios than in constructing checks that faithfully reflect the specification. Specifically, the agents show a clear tendency to write checks that mirror the reference implementation rather than the specification, and therefore reject implementations that the specification permits.
Read the full abstract

All authors Zhiqing Zhong, Mingyu Chen, Aoyang Fang, Yuzhong Zhang, Haoye Tian, Lionel Briand, Boxi Yu, Pinjia He

Computation and Language · Cross-listed from cs.LG

Davinci-Tool: Scaling Agentic Tool-Use Supervision with Verifiable Executable Environments

Xiaoyang Yuan, Ji Zhao, Xingyu Hu et al. (+7 more)

Reliable agentic tool use requires training data that pair diverse multi-step workflows with executable evidence that the requested outcomes were achieved. Scaling either trajectory coverage or verification alone is insufficient as new environments introduce distinct states, rules, and dependencies that must also be reflected in the success criteria. To address this challenge, we introduce Davinci-Tool, a 30B tool-use model, and Davinci-ToolData, an execution-grounded dataset designed to scale environment coverage and outcome verification together. Davinci-ToolData preserves restorable states, execution traces, outcomes, and verifiers as a unified substrate for supervision. Two complementary pipelines populate this substrate: Execution-Grounded Task Synthesis (EGTS) derives verified demonstrations from independently authored systems, and Executable Environment Synthesis (EES) synthesizes auditable environments, state-grounded tasks, and executable verifiers across single and coordinated domains. These pipelines yield 30,395 verified trajectories spanning 2,260 invoked tool names. We first train on these demonstrations with supervised fine-tuning, then reuse EES states and verifiers to compare alternative actions and construct outcome-grounded preferences for Direct Preference Optimization. Davinci-Tool improves Qwen3-Coder-30B-A3B-Instruct from 52.4 to 88.2 on $\tau^2$-bench, from 30.4 to 62.8 on BFCL v4 multi-turn, and from 16.9 to 56.2 on MCP-Atlas, outperforming Qwen3-Coder-480B-A35B-Instruct on all three benchmarks with only 1/16 of its total parameters. Code is available at https://github.com/MassLab-SII/Davinci-Tool.
Read the full abstract

All authors Xiaoyang Yuan, Ji Zhao, Xingyu Hu, Wenhan Wang, Yutao Hu, Qiaosheng Zhang, Pengfei Liu, Yu Qiao, Yi Bin, Wenqi Shao

Computation and Language

ANCHOR: Assessing Naturalistic Constraint Hierarchies and Ordering in Repository-Grounded Agentic Coding

Wenqiang Zhu, Xinping Lei, Zhiyi Lai et al. (+17 more)

For repository-grounded coding agents, a functionally correct patch is not necessarily an acceptable one: developers also constrain how the task must be carried out-modification scope, workflow order, validation, tool use, and reporting. This capability, which we call engineering constraint following, remains largely invisible to current evaluations. We introduce ANCHOR (Assessing Naturalistic Constraint Hierarchies and Ordering in Repository-Grounded Agentic Coding), a benchmark built on naturalistic, multilingual repository tasks spanning eight language groups. ANCHOR combines (i) a hierarchical taxonomy of 10 categories and 78 leaf types distilled from real developer interactions, (ii) an automated pipeline that instantiates task-grounded atomic constraints and composes them into typed AND/THEN/PRIORITY relations, and (iii) a trajectory-grounded evaluation framework that checks both the code diff and the execution trajectory with rule-based and LLM judges. Across 19 agent configurations spanning 2 scaffolds and 10 foundation models on 358 tasks, the strongest measured system, Claude Opus 4.8 under Cline, reaches 70.1% atomic compliance and 63.7% relation compliance, while 19.0% of tasks pass every scored checklist item under the strict all-item FCCR criterion. A five-engineer audit of constraints from 50 instances reports 94% feasibility, 94% repository consistency, and 96% realism. In the primary paired set (14 configurations), the mean signed change in task completion after adding constraints is +0.7 percentage points; individual changes vary in both directions. These results distinguish functional success from constraint compliance, with failures concentrated in workflow ordering, tool-use discipline, and reporting, and substantial interactions between models and scaffolds.
Read the full abstract

All authors Wenqiang Zhu, Xinping Lei, Zhiyi Lai, Zhengshuo Gong, Tianzhuang He, Songwei Yu, Weihao Li, Jingxuan Xu, Zihang Xu, Gang Wu, Zhongqiu Wang, Yuhui Bai, Jie Li, Zhao Liu, Junhao Fan, Yan Wu, Yanan Wu, Jiaheng Liu, Quan Gu, Ariel Deng