Computation and Language · Cross-listed from cs.LG
LiSA: Lifelong Safety Adaptation via Conservative Policy Induction
Minbeom Kim, Lesly Miculicich, Bhavana Dalvi Mishra et al. (+6 more)
As AI
agents move from chat interfaces to systems that read private data, call tools, and execute multi-step workflows, guardrails become a last line of defense against concrete deployment harms. In these settings, guardrail failures are no longer merely answer-quality errors: they can leak secrets, authorize unsafe actions, or block legitimate work. The hardest failures are often contextual: whether an action is acceptable depends on local
privacy norms, organizational policies, and user expectations that resist pre-deployment specification. This creates a practical gap: guardrails must adapt to their own operating environments, yet deployment feedback is typically limited to sparse, noisy user-reported failures, and repeated fine-tuning is often impractical. To address this gap, we propose LiSA (Lifelong Safety Adaptation), a conservative policy induction framework that improves a fixed base guardrail through structured
memory. LiSA converts occasional failures into reusable policy abstractions so that sparse reports can generalize beyond individual cases, adds conflict-aware local rules to prevent overgeneralization in mixed-label contexts, and applies evidence-aware confidence gating via a posterior lower bound, so that
memory reuse scales with accumulated evidence rather than empirical accuracy alone. Across PrivacyLens+, ConFaide+, and AgentHarm, LiSA consistently outperforms strong
memory-based baselines under sparse feedback, remains robust under noisy user feedback even at 20% label-flip rates, and pushes the latency–performance frontier beyond backbone model scaling. Ultimately, LiSA offers a practical path to secure AI
agents against the unpredictable long tail of real-world edge risks.
Read the full abstract ▾
All authors Minbeom Kim, Lesly Miculicich, Bhavana Dalvi Mishra, Mihir Parmar, Phillip Wallis, Bharath Chandrasekhar, Kyomin Jung, Tomas Pfister, Long T. Le
Computation and Language · Cross-listed from cs.LG
When Answers Stray from Questions: Hallucination Detection via Question-Answer Orthogonal Decomposition
Siyang Yao, Erhu Feng and Yubin Xia
Hallucination detection in large
language models (LLMs) requires balancing accu racy, efficiency, and robustness to distribution shift. Black-box consistency methods are effective but demand repeated inference; single-pass white-box probes are effi cient yet treat answer representations in isolation, often degrading sharply under domain shift. We propose QAOD (Question-Answer Orthogonal Decomposition), a single-pass framework that projects away the question-aligned direction from the answer representation to obtain a question-orthogonal component that suppresses domain-conditioned variation. To identify informative signals, QAOD further selects layers via diversity-penalized Fisher scoring and discriminative
neurons via Fisher importance. To address both in-domain detection and cross-domain generalization, we design two complementary probing strategies: pairing the or thogonal component with question context yields a joint probe that maximizes in-domain discriminability, while using the orthogonal component alone preserves domain-agnostic factuality signals for robust transfer. QAOD's joint probe achieves the best in-domain AUROC across all evaluated model-
dataset pairs, while the orthogonal-only probe delivers the strongest OOD transfer, surpassing the best white-box baseline by up to 21% on BioASQ at under 25% of generation cost.
Read the full abstract ▾
Computation and Language · Cross-listed from cs.CV
Think When Needed: Adaptive Reasoning-Driven Multimodal Embeddings with a Dual-LoRA Architecture
Longxiang Zhang, Weilong Dai, Guanghao Zhang et al. (+2 more)
Multimodal large
language models (MLLMs) have emerged as a powerful backbone for
multimodal embeddings. Recent methods introduce chain-of-thought (CoT)
reasoning into the
embedding pipeline to improve
retrieval quality, but remain costly in both model size and inference cost. They typically employ separate reasoner and embedder with substantial parameter overhead, and generate CoT indiscriminately for every input. However, we observe that for simple inputs, discriminative
embeddings already perform well, and redundant
reasoning can even mislead the model, degrading performance. To address these limitations, we propose Think When Needed (TWN), a unified
multimodal embedding framework with adaptive
reasoning. TWN introduces a dual-LoRA architecture that attaches
reasoning and
embedding adapters to a shared frozen backbone, detaching gradients at their interface to mitigate gradient conflicts introduced by joint optimization while keeping parameters close to a single model. Building on this, an adaptive think mechanism uses a self-supervised routing gate to decide per input whether to generate CoT, skipping unnecessary
reasoning to reduce inference overhead and even improve
retrieval quality. We further explore
embedding-guided RL to optimize CoT quality beyond supervised training. On the 78 tasks of MMEB-V2, TWN achieves state-of-the-art
embedding quality while being substantially more efficient than existing generative methods, requiring only 3-5% additional parameters relative to the backbone and up to 50% fewer
reasoning tokens compared to the full generative mode.
Read the full abstract ▾
All authors Longxiang Zhang, Weilong Dai, Guanghao Zhang, Hao Jiang, Pipei Huang
Computation and Language
A Calculus-Based Framework for Determining Vocabulary Size in End-to-End ASR
Sunil Kumar Kopparapu
In hybrid automatic
speech recognition (ASR) systems, the vocabulary size is unambiguous, typically determined by the number of phones, bi-phones, or tri-phones present in the
language. In contrast, end-to-end ASR systems derive their vocabulary, often referred to as
tokens from the text corpus used for training. The choice and, more importantly, the size of this vocabulary is a critical hyper-parameter in training end-to-end ASR systems. Tokenization
algorithms such as Byte Pair Encoding (BPE), WordPiece, and Unigram
Language Model (ULM) use the vocabulary size as an input hyper-parameter to generate the sub-words employed during ASR training. Popular toolkits like ESPNet provide a fixed vocabulary size in their training recipes, but there is little documentation or discussion in the literature regarding how these values are determined. Recent work [1] has formalized an approach to identify the vocabulary size best suited for end-to-end ASR, introducing a cost function framework that treats the tokenization process as a black box. In this paper, we build upon that foundation by curve fitting the training data and using the principle of first and second derivative tests in calculus to formally estimate the vocabulary size hyper-parameter. We demonstrate the utility and usefulness of our approach by applying it on a standard Librispeech corpus and show that the optimal choice of vocabulary size hyper-parameter improves the performance of the ASR. The main contribution of this paper in formalizing an approach to identify the vocabulary size best suited for training an end-to-end ASR system.
Read the full abstract ▾
Computation and Language · Cross-listed from cs.SE
SWE-Chain: Benchmarking Coding Agents on Chained Release-Level Package Upgrades
Man Ho Lam, Chaozheng Wang, Hange Liu et al. (+5 more)
Coding
agents powered by large
language models are increasingly expected to perform realistic software maintenance tasks beyond isolated issue resolution. Existing
benchmarks have shifted toward realistic software
evolution, but they rarely capture continuous maintenance at the granularity of package releases, where changes are bundled, shipped, and inherited by subsequent versions. We present SWE-Chain, a
benchmark for evaluating
agents on chained release-level package upgrades, where each transition builds on the
agent's prior codebase. To produce upgrade specifications, we design a divide-and-conquer synthesis pipeline that aligns release notes with code diffs for each version transition, ensuring the requirements are grounded in actual code changes, informative to
agents, and feasible to implement. SWE-Chain contains 12 upgrade chains across 9 real Python packages, with 155 version transitions and 1,660 grounded upgrade requirements. Across nine frontier
agent-model configurations,
agents achieve an average of 44.8% resolving, 65.4% precision, and 50.2% F1 under the Build+Fix regime, with Claude-Opus-4.7 (Claude Code) leading at 60.8% resolving, 80.6% precision, and 68.5% F1. These results show that SWE-Chain is both feasible and discriminative, and reveal that current
agents still struggle to make correct upgrades across chained package releases without breaking existing functionality.
Read the full abstract ▾
All authors Man Ho Lam, Chaozheng Wang, Hange Liu, Jingyu Xiao, Haau-sing Li, Jen-tse Huang, Terry Yue Zhuo, Michael R. Lyu
Computation and Language
Knowledge Beyond Language: Bridging the Gap in Multilingual Machine Unlearning Evaluation
Kyomin Hwang, Hyeonjin Kim, Sangyeon Cho et al. (+1 more)
While LLMs are increasingly used in commercial services, they pose
privacy risks such as leakage of sensitive personally identifiable information (PII). For LLMs trained on
multilingual corpora,
Multilingual Machine Unlearning (MMU) aims to remove information across multiple
languages. However, prior MMU evaluations fail to capture such cross-linguistic distribution of information, being largely limited to direct extensions of per-
language evaluation protocols. To this end, we propose two metrics to evaluate the information spread across
languages: the Knowledge Separability Score (KSS) and the Knowledge Persistence Score (KPS). KSS measures the overall unlearning quality across multiple
languages, while KPS more specifically aims to assess consistent removal of information among different
language pairs. We evaluated various unlearning methods in the
multilingual setting with these metrics and conducted comprehensive analyses. Through our investigation, we provide insights into unique phenomena exclusive to MMU and offer a new perspective on MMU evaluation.
Read the full abstract ▾
All authors Kyomin Hwang, Hyeonjin Kim, Sangyeon Cho, Nojun Kwak
Computation and Language
Agentic Recommender System with Hierarchical Belief-State Memory
Xiang Shen, Yuhang Zhou, Yifan Wu et al. (+8 more)
Memory-augmented LLM
agents have advanced personalized recommendation, yet existing approaches universally adopt flat
memory representations that conflate ephemeral signals with stable preferences, and none provides a complete lifecycle governing how
memory should evolve. We propose MARS (
Memory-Augmented Agentic Recommender System), a framework that treats recommendation as a partially observable problem and maintains a structured belief state that progressively abstracts noisy behavioral observations into a compact estimate of user preferences. MARS organizes this belief state into three tiers: event
memory buffers raw signals, preference
memory maintains fine-grained mutable chunks with explicit strength and evidence tracking, and profile
memory distills all preferences into a coherent natural
language narrative. A complete lifecycle of six operations – extraction, reinforcement, weakening, consolidation, forgetting, and resynthesis – is adaptively scheduled by an LLM-based planner rather than fixed-interval heuristics. Experiments on four InstructRec
benchmark domains show that MARS achieves state-of-the-art performance with average improvements of 26.4% in HR@1 and 10.3% in NDCG@10 over the strongest baselines with further gains from agentic scheduling in evolving settings.
Read the full abstract ▾
All authors Xiang Shen, Yuhang Zhou, Yifan Wu, Zhuokai Zhao, Siyu Lin, Lei Huang, Qianqian Zhong, Lizhu Zhang, Benyu Zhang, Xiangjun Fan, Hong Yan
Computation and Language · Cross-listed from cs.AI
Nexus : An Agentic Framework for Context-Aware Time Series Forecasting
Sarkar Snigdha Sarathi Das, Palash Goyal, Mihir Parmar et al. (+6 more)
Language models have emerged as highly capable reasoners, yet their performance in time series
forecasting remains uneven across domains and levels of contextual grounding. This is primarily due to their suboptimal time series
reasoning and numerical modeling capabilities. To tackle this, we introduce Nexus, a
multimodal forecasting framework that bridges qualitative event
reasoning with structural numerical modeling through principled statistical synthesis. Nexus follows three structured stages: (1) Temporal profiling and anchor generation, which distills the
multimodal history into a timestamped causal-temporal profile and generates a forecast anchor (2) Dual contextual outlook generation, which produces decoupled macro and micro level forecast trajectories, and (3)
Bayesian synthesis, which reconciles the
forecasting anchor with both contextual forecast trajectories via closed-form Gaussian updating weighted by empirical error covariances. Evaluated across three diverse
forecasting benchmarks constituting eight
datasets spanning rich event contexts, abrupt shocks, and purely numerical series, Nexus consistently outperforms leading time-series foundation models and
language model baselines, while providing interpretable causal
reasoning.
Read the full abstract ▾
All authors Sarkar Snigdha Sarathi Das, Palash Goyal, Mihir Parmar, Nanyun Peng, Vishy Tirumalashetty, Chun-Liang Li, Rui Zhang, Jinsung Yoon, Tomas Pfister
Computation and Language · Cross-listed from cs.LG
NodeSyn: Socially Aligned Synthetic Data for AI Evaluation
Qazi Mamunur Rashid, Xuan Yang, Zhengzhe Yang et al. (+4 more)
Recent advancements in generative AI facilitate large-scale synthetic data genera- tion for model evaluation. However, without targeted approaches, these
datasets often lack the sociotechnical nuance required for sensitive domains. We introduce NodeSyn, an evidence-grounded methodology that generates socially relevant syn- thetic queries by leveraging a fine-tuned taxonomy generator (TaG) anchored in real-world evidence. Evaluated against four mainstream LLMs, NodeSyn elicited failure rates up to five times higher than human-authored
benchmarks. Ablation studies confirm that our granular taxonomic expansion significantly drives these failure rates, while independent validation reveals critical deficiencies in prominent guard models.
Read the full abstract ▾
All authors Qazi Mamunur Rashid, Xuan Yang, Zhengzhe Yang, Yanzhou Pan, Erin van Liemt, Darlene Neal, Kshitij Pancholi
Computation and Language
VISHC at PsyDefDetect: Mitigating Data Scarcity in Psychological Defense Classification with Context-Aware Synthetic Augmentation
Hoang-Thuy-Duong Vu, Quoc-Cuong Pham and Huy-Hieu Pham
Psychological defense mechanisms (PDMs) are unconscious cognitive processes that modulate how individuals perceive and respond to emotional distress. Automatically classifying PDMs from text is clinically valuable but severely hindered by data scarcity and class imbalance, challenges which generative augmentation alone cannot resolve without psychological grounding. In this work, we address these challenges in the PsyDefDetect shared task (BioNLP@ACL 2026) by proposing a context-aware synthetic augmentation framework combined with a hybrid classification model. Our hybrid model integrates contextual
language representations with basic clinical features, along with 150 annotated defense items. Experiments demonstrate that definition quality in prompting directly governs generation fidelity and downstream performance. Our method surpasses DMRS Co-Pilot, reaching an accuracy of 58.26% (+40.25%) and a macro-F1 of 24.62% (+15.99%), thereby establishing a strong baseline for psychologically grounded defense mechanism classification in low-resource settings. Source code is available at: https://github.com/htdgv/CASA-PDC.
Read the full abstract ▾
Computation and Language
Where Should Diffusion Enter a Language Model? Geometry-Guided Hidden-State Replacement
Injin Kong, Hyoungjoon Lee and Yohan Jo
Continuous
diffusion language models require choosing a representation space in which denoising operates, yet it remains unclear which representations are most compatible with
diffusion. We study a basic design question: where inside a pretrained
language model should continuous
diffusion operate? We formulate this as a hidden-state interface-selection problem and hypothesize that
diffusion-friendly interfaces can be identified from representation geometry. We operationalize this hypothesis using three training-free
geometric proxies: local compactness, global stiffness, and effective rank. Across two 8B-scale backbones, the resulting geometry score predicts fixed-budget
diffusion bridgeability beyond the dominant effect of layer depth. We then instantiate DiHAL, a Locate-and-Replace framework that replaces the
transformer prefix below a selected interface with conditional
diffusion while retaining the pretrained suffix and LM head. Under full training, geometry-selected interfaces remain close to validation-loss oracles and outperform worst-layer controls, while the
diffusion bridge outperforms a parameter-matched deterministic replacement under matched diagnostic conditions. These results suggest that the representation space in which
diffusion operates should itself be treated as a first-class design variable for continuous
diffusion in
language models.
Read the full abstract ▾
Computation and Language
Reinforcement Learning with Semantic Rewards Enables Low-Resource Language Expansion without Alignment Tax
Zeli Su, Ziyin Zhang, Zhou Liu et al. (+7 more)
Extending large
language models (LLMs) to low-resource
languages often incurs an "
alignment tax": improvements in the target
language come at the cost of catastrophic forgetting in general capabilities. We argue that this trade-off arises from the
rigidity of supervised fine-tuning (SFT), which enforces
token-level surface imitation on narrow and biased data distributions. To address this limitation, we propose a semantic-space
alignment paradigm powered by Group Relative Policy Optimization (GRPO), where the model is optimized using
embedding-level semantic rewards rather than likelihood maximization. This objective encourages meaning preservation through flexible realizations, enabling controlled updates that reduce destructive interference with pretrained knowledge. We evaluate our approach on Tibetan-Chinese machine translation and Tibetan headline generation. Experiments show that our method acquires low-resource capabilities while markedly mitigating
alignment tax, preserving general competence more effectively than SFT. Despite producing less rigid surface overlap, semantic RL yields higher semantic quality and preference in open-ended generation, and few-shot transfer results indicate that it learns more transferable and robust representations under limited supervision. Overall, our study demonstrates that reinforcement learning with semantic rewards provides a safer and more reliable pathway for inclusive low-resource
language expansion.
Read the full abstract ▾
All authors Zeli Su, Ziyin Zhang, Zhou Liu, Xuexian Song, Zhankai Xu, Longfei Zheng, Xiaolu Zhang, Rong Fu, Guixian Xu, Wentao Zhang
Computation and Language · Cross-listed from cs.HC
A Formative Study of Brief Affective Text as a Complement to Wearable Sensing for Longitudinal Student Health Monitoring
Tamunotonye Harry, Johanna Hidalgo, Matthew Price et al. (+7 more)
Wearable devices capture physiological and behavioral data with increasing fidelity, but the psychological context shaping these outcomes is difficult to recover from sensor data alone, limiting the utility of passive sensing for digital health goals such as early detection of distress, personalized intervention, and timely clinical outreach. We examined whether ultra-brief naturalistic concern text could serve as a scalable complement to passive sensing. In a year-long study of 458 university students (3,610 person-waves) tracked with Oura rings, participants responded bimonthly to an open-ended
prompt about what concerned them most; responses had a median length of three words. We compared dictionary-based, general pretrained, and domain-adapted NLP approaches using within-person mixed-effects models across nine sleep and physical activity outcomes to determine which method best recovers physiologically relevant signal from brief naturalistic text. Weeks dominated by academic concern framing were associated with lower physical activity; weeks characterized by emotional exhaustion
language were associated with poorer sleep quality and lower heart rate variability. General pretrained
embeddings performed as well as or better than domain-adapted models across most outcomes, with differences between the two generally small and within the range of estimation noise. Zero-shot classification of concern topics showed no consistent evidence of association with outcomes; affective dimensions across all three methods showed more associations, though these did not survive correction for multiple comparisons, offering preliminary evidence that emotional register may carry more signal than topical content. These findings offer design guidance: ultra-brief affective
prompts enrich the psychological interpretability of passive physiological data at minimal burden.
Read the full abstract ▾
All authors Tamunotonye Harry, Johanna Hidalgo, Matthew Price, Yuanyuan Feng, Kathryn Stanton, Connie Tompkins, Peter Sheridan Dodds, Mikaela Irene Fudolig, Laura Bloomfield, Christopher Danforth
Computation and Language · Cross-listed from cs.AI
Herculean: An Agentic Benchmark for Financial Intelligence
Xueqing Peng, Zhuohan Xie, Yupeng Cao et al. (+60 more)
As AI
agents advance, the central question is no longer whether they can solve isolated financial tasks, but whether their capabilities generalize across complete professional workflows. Existing
benchmarks cannot answer this question because they evaluate either static competencies or
agents within a single interaction structure. We introduce Herculean, a
benchmark of end-to-end agentic financial intelligence that crosses
agent architecture, backbone model, and four structurally distinct workflows: Trading, Hedging,
Market Insights, and
Auditing. Each workflow is implemented as an MCP skill that preserves its native state, actions, constraints, and outcome criteria, allowing intact
agent systems to be evaluated under matched external conditions. Across 30 architecture–backbone configurations, we find that
agent capability is not globally ordered. Workflow rankings are weakly aligned, changing the architecture frequently reverses performance under the same backbone, and each architecture concentrates failure in different forms of workflow control. These results establish that
agent performance is an interaction between architecture and workflow demand, rather than an intrinsic property of the underlying model. Consequently, success on a single financial workflow provides little evidence of dependable, general financial intelligence.
Read the full abstract ▾
All authors Xueqing Peng, Zhuohan Xie, Yupeng Cao, Haohang Li, Lingfei Qian, Yan Wang, Vincent Jim Zhang, Huan He, Xuguang Ai, Linhai Ma, Ruoyu Xiang, Yueru He, Yi Han, Shuyao Wang, Yuqing Guo, Mingyang Jiang, Yilun Zhao, Youzhong Dong, Xiaoyu Wang, Yankai Chen, Ye Yuan, Qiyuan Zhang, Fuyuan Lyu, Haolun Wu, Yonghan Yang, Zichen Zhao, Yuyang Dai, Fan Zhang, Rania Elbadry, Ayesha Gull, Muhammad Usman Safder, Nuo Chen, Fengbin Zhu, Tianshi Cai, Zimu Wang, Polydoros Giannouris, Yuechen Jiang, Zhiwei Liu, Mohsinul Kabir, Yuyan Wang, Yixiang Zheng, Yangyang Yu, Weijin Liu, Wenbo Cao, Anke Xu, Peng Lu, Jerry Huang, Mingquan Lin, Prayag Tiwari, Yijia Zhao, Víctor Gutiérrez-Basulto, Xiao-Yang Liu, Kaleb E Smith, Jiahuan Pei, Arman Cohan, Jimin Huang, Yuehua Tang, Alejandro Lopez-Lira, Xi Chen, Xue Liu, Junichi Tsujii, Jian-Yun Nie, Sophia Ananiadou
Computation and Language
LLM-based Detection of Manipulative Political Narratives
Sinclair Schneider, Florian Steuber and Gabi Dreo Rodosek
We present a new computational framework for detecting and structuring manipulative political narratives. A task that became more important due to the shift of political discussions to social media. One of the primary challenges thereby is differentiating between manipulative political narratives and legitimate critiques. Some posts may also reframe actual events within a manipulative context. To achieve good clustering results, we filter manipulative posts beforehand using a detailed few-shot
prompt that combines documented campaign narratives with legitimate criticisms to differentiate them. This
prompt enables a
reasoning model to assign labels, retaining only manipulative narrative posts for further processing. The remaining posts are subsequently embedded and dimensionality-reduced using UMAP, before HDBSCAN is applied to uncover narrative groups. A key advantage of this unsupervised approach is its independence from a predefined list of target categories, enabling it to uncover new narrative clusters. Finally, a
reasoning model is employed to uncover the narrative behind each cluster. This approach, applied to over 1.2 million social media posts, effectively identified 41 distinct manipulative narrative clusters by integrating
prompt-based filtering with unsupervised clustering.
Read the full abstract ▾
Computation and Language
Ideology Prediction of German Political Texts
Sinclair Schneider, Florian Steuber, Joao A. G. Schneider et al. (+1 more)
Elections represent a crucial milestone in a nation's ongoing development. To better understand the political rhetoric from various movements, ranging from left to right, we propose a
transformer-based model capable of projecting the political orientation of a text on a continuous left-to-right spectrum, represented by a normalized scalar d between -1 and 1. This approach enables analysts to focus on specific segments of the political landscape, such as conservatives, while excluding liberal and far-right movements. Such a task can only be achieved with multiclass classifiers, provided that the desired orientation is incorporated within one of their predefined classes. To determine the most suitable foundation model among 13 candidate
transformers for this task, we constructed four distinct corpora. One corpus comprised annotated plenary notes from the German Bundestag, while another was based on an official online decision-making tool, Wahl-O-Mat. The third corpus consisted of articles from 33 newspapers, each identified by its political orientation, and the fourth included 535,200 tweets from 597 members of the 20th and 21st German Bundestag. To mitigate overfitting, we used two distinct corpora for training and two for testing, respectively. For in-domain performance, DeBERTa-large achieved the highest F1 score F1=0.844 as well as for the X (Twitter) out-of-domain test ACC=0.864. Regarding the newspaper out-of-domain test, Gemma2-2B excelled (MAE = 0.172). This study demonstrates that
transformer models can recognize political framing in German news at the level of public opinion polls. Our findings suggest that both the model architecture and the availability of domain-specific training data can be as influential as model size for estimating political bias. We discuss methodological limitations and outline directions for improving the robustness of bias measurement.
Read the full abstract ▾
All authors Sinclair Schneider, Florian Steuber, Joao A. G. Schneider, Gabi Dreo Rodosek
Computation and Language · Cross-listed from cs.LG
Dynamic Latent Routing
Fangyuan Yu, Xin Su and Amir Abdullah
We investigate the temporal concatenation of sub-policies in Markov Decision Processes (MDP) with time-varying reward functions. We introduce General Dijkstra Search (GDS), and prove that globally optimal goal-reaching policies can be recovered through temporal composition of intermediate optimal sub-policies. Motivated by the "search, select, update" principle underlying GDS, we propose Dynamic Latent Routing (DLR), a
language-model post-training method that jointly learns discrete latent codes, routing policies, and model parameters through dynamic search in a single training stage. In low-data fine-tuning settings, DLR matches or outperforms supervised fine-tuning across four
datasets and six models, achieving a mean gain of +6.6 percentage points, while prior discrete-latent baselines consistently underperform SFT. Mechanistic analyses and targeted code ablations show that DLR learns structured routing behaviors with distinct causal roles.
Read the full abstract ▾
Computation and Language
Factorization-Error-Free Discrete Diffusion Language Model via Speculative Decoding
Xun Fang, Yunchen Li, Hang Yuan et al. (+1 more)
Discrete
diffusion language models improve generation efficiency through parallel
token prediction, but standard $X_0$ prediction methods introduce factorization errors by approximating the clean
token posterior with independent
token-wise distributions. This paper proposes Factorization-Error-Free Discrete
Diffusion Language Modeling (FeF-DLLM), which replaces independent clean-
token prediction with an exact prefix-conditioned factorization of the clean posterior to better preserve
token dependencies. To reduce the sequential cost introduced by prefix conditioning, FeF-DLLM further incorporates speculative decoding within
diffusion denoising, accelerating inference while maintaining the parallel prediction and re-masking properties of DLLMs. Theoretically, we prove that FeF-DLLM generates from the true joint distribution and derive its expected acceleration ratio. Experiments on GSM8K, MATH, HumanEval, and MBPP demonstrate that our method improves accuracy by an average of 5.04 percentage points while achieving an average inference speedup of $3.86\times$.
Read the full abstract ▾
All authors Xun Fang, Yunchen Li, Hang Yuan, Zhou Yu
Computation and Language · Cross-listed from cs.LG
Minimal-Intervention KV Retention via Set-Conditioned Diversity
Libo Sun, Po-wei Harn, Peixiong He et al. (+1 more)
KV-cache compression at small budgets is a crowded design space spanning cache representation, head-wise routing, compression cadence, decoding behavior, and within-budget scoring. We study seven mechanisms across these five families under matched mean cache on long-form mathematical
reasoning (MATH-500~\cite{hendrycks2021math}) with two distilled-
reasoning models (Qwen-7B and Llama-8B variants of DeepSeek-R1-Distill~\cite{deepseek2025r1}) at budgets $b \in \{64, 128\}$. All seven were rejected. We then propose $\alpha$, a one-function modification to the TriAttention~\cite{mao2026triattention} retention scorer that replaces argmax-top-$k$ with greedy facility-location-inspired selection under a V-space redundancy penalty controlled by a single weight $\lambda$. A pre-registered protocol tunes $\lambda$ on a frozen development split and confirms on a disjoint held-out split; with $\lambda = 0.5$, $\alpha$ clears Bonferroni on two of the four (model, budget)
cells (Qwen $b{=}128$ and Llama $b{=}64$), no
cell is significantly negative, and the pre-registered Branch~A triggers. The finding is asymmetric: a minimal scoring modification beat heavier structural redesigns in this regime, and the combined matched-
memory, sympy-graded, held-out confirmation protocol is the evidence standard that made the asymmetry visible.
Read the full abstract ▾
All authors Libo Sun, Po-wei Harn, Peixiong He, Xiao Qin
Computation and Language · Cross-listed from cs.CR
To See is Not to Learn: Protecting Multimodal Data from Unauthorized Fine-Tuning of Large Vision-Language Model
Chengshuai Zhao, Zhen Tan, Dawei Li et al. (+2 more)
The rapid advancement of Large
Vision-
Language Models (LVLMs) is increasingly accompanied by unauthorized scraping and training on
multimodal web data, posing severe copyright and
privacy risks to data owners. Existing countermeasures, such as machine unlearning and watermarks, are inherent post-hoc approaches that act only after intellectual property infringement has already occurred. In this work, we propose MMGuard to empower data owners to proactively protect their
multimodal data against unauthorized LVLM fine-tuning. MMGuard generates unlearnable examples by injecting human-imperceptible perturbations that actively exploit the learning dynamics of LVLMs. By minimizing the training loss, the perturbation creates an optimization shortcut, causing the model to overfit to the noise and thereby degrading downstream performance when the perturbation is absent during inference. To further strengthen this defense, MMGuard introduces a cross-modal binding disruption, strategically shifting LVLM
attention to enforce a spurious correlation between the noise and the training target with theoretical guarantees. Enhanced by an ensemble learning strategy for cross-model transferability, MMGuard is evaluated against nine open-source LVLMs across six
datasets. Our comprehensive results demonstrate effective, stealthy, and robust protection under white-box, gray-box, and black-box threat models, establishing a mechanistic advantage in proactively defending against aggressive fine-tuning exploitation.
Read the full abstract ▾
All authors Chengshuai Zhao, Zhen Tan, Dawei Li, Zhiyuan Yu, Huan Liu
Computation and Language · Cross-listed from cs.CR
Web Agents Should Adopt the Compile-and-Execute Paradigm
Julien Piet, Annabella Chow, Yiwei Hou et al. (+5 more)
ReAct has become the default architecture across LLM
agents, and many existing web
agents follow this paradigm. We argue that it is the wrong default for web
agents. Instead, web
agents should default to compile-and-execute: commit to a task-specific program before observing runtime web content, then execute it. The reason is that web content mixes inputs from many parties. An e-commerce product page may combine a seller's listing, customer reviews, sponsored advertisements, and platform-generated recommendations. Under ReAct, all of this content flows into the model when deciding on the next action, creating a direct path for
prompt injections to steer the
agent's control flow. Compile-and-execute changes this boundary: untrusted data may influence values or branches inside a predefined execution graph, but it cannot redefine the user task or cause the model to synthesize new actions at runtime. We analyze WebArena, a popular web
agent benchmark, and find that all tasks are compatible with compile-and-execute, while 81% can be completed with a purely static program, without any runtime LLM subroutine. We identify the main barrier to adopting compile-and-execute on the web: For it to work well, tools must map cleanly to semantic actions, with effects known before execution, so a compiler has enough information to emit a program. The web does not naturally expose that interface. Browser tools such as click, type, and scroll have page-dependent meanings. Planning at this layer is near-sighted: the
agent can only see actions on the current page, and later actions appear only after it acts. Closing this gap requires typed interfaces that turn website interactions from clicks and keystrokes to task-level operations. This is an infrastructure problem, not a modeling problem. Web tasks do not need reactivity by default; they need typed, complete, auditable website APIs.
Read the full abstract ▾
All authors Julien Piet, Annabella Chow, Yiwei Hou, Muxi Lyu, Sylvie Venuto, Jinhao Zhu, Raluca Ada Popa, David Wagner
Computation and Language · Cross-listed from cs.LG
MetaMoE: Diversity-Aware Proxy Selection for Privacy-Preserving Mixture-of-Experts Unification
Weisen Jiang, Shuhao Chen and Sinno Jialin Pan
Mixture-of-Experts (MoE) models scale capacity by combining specialized experts, but most existing approaches assume centralized access to training data. In practice, data are distributed across clients and cannot be shared due to
privacy constraints, making unified MoE training challenging. We propose MetaMoE, a
privacy-preserving framework that unifies independently trained, domain-specialized experts into a single MoE using public proxy data as surrogates for inaccessible private data. Central to MetaMoE is diversity-aware proxy selection, which selects client-domain-relevant and diverse samples from public data to effectively approximate private data distributions and supervise router learning. These proxies are further used to align expert training, improving expert coordination at unification time, while a context-aware router enhances expert selection across heterogeneous inputs. Experiments on computer
vision and natural
language processing
benchmarks demonstrate that MetaMoE consistently outperforms recent
privacy-preserving MoE unification methods. Code is available at https://github.com/ws-jiang/MetaMoE.
Read the full abstract ▾
Computation and Language
Chengzhi Liu, Yichen Guo, Yepeng Liu et al. (+8 more)
LLM
agents increasingly run inside execution harnesses that dispatch tools, allocate resources, and route messages between specialized components. However, a harness can return a correct, benign answer over a trajectory that accesses unauthorized resources or leaks context to the wrong
agent. Output-level evaluation cannot see these failures, yet most safety
benchmarks score only final outputs or terminal states, even though many violations occur mid-trajectory rather than at termination. The central question is whether the harness respects user intent, permission boundaries, and information-flow constraints throughout execution. To address this gap, we propose HarnessAudit, a framework that audits full execution trajectories across boundary compliance, execution fidelity, and system stability, with a focus on multi-
agent harnesses where these risks are most pronounced. We further introduce HarnessAudit-Bench, a
benchmark of 210 tasks across eight real-world domains, instantiated in both single-
agent and multi-
agent configurations with embedded safety constraints. Evaluating ten harness configurations across frontier models and three multi-
agent frameworks, we find that: (i) task completion is misaligned with safe execution, and violations accumulate with trajectory length; (ii) safety risks vary across domains, task types, and
agent roles; (iii) most violations concentrate in resource access and inter-
agent information transfer; and (iv) multi-
agent collaboration expands the safety risk surface, while harness design sets the upper bound of safe deployment.
Read the full abstract ▾
All authors Chengzhi Liu, Yichen Guo, Yepeng Liu, Yuzhe Yang, Qianqi Yan, Xuandong Zhao, Wenyue Hua, Sheng Liu, Sharon Li, Yuheng Bu, Xin Eric Wang
Computation and Language · Cross-listed from cs.AI
Hypergraph Enterprise Agentic Reasoner over Heterogeneous Business Systems
Ling Wang, Xin Liu, Songnan Liu et al. (+8 more)
Applying Large
Language Models (LLMs) to heterogeneous enterprise systems is hindered by
hallucinations and failures in multi-hop, n-ary
reasoning. Existing paradigms (e.g., GraphRAG, NL2SQL) lack the semantic grounding and auditable execution required for these complex environments. We introduce HEAR, an enterprise agentic reasoner built on a Stratified Hypergraph Ontology. Its base Graph Layer virtualizes provenance-aware data interfaces, while the Hyperedge Layer encodes n-ary business rules and procedural protocols. Operating an evidence-driven
reasoning loop, HEAR dynamically orchestrates ontology tools for structured multi-hop analysis without requiring LLM retraining. Evaluations on supply-chain tasks, including order fulfillment blockage root cause analysis (RCA), show HEAR achieves up to 94.7% accuracy. Crucially, HEAR demonstrates adaptive efficiency: utilizing procedural hyperedges to minimize
token costs, while leveraging topological exploration for rigorous correctness on complex queries. By matching proprietary model performance with open-weight backbones and automating manual diagnostics, HEAR establishes a scalable, auditable foundation for enterprise intelligence.
Read the full abstract ▾
All authors Ling Wang, Xin Liu, Songnan Liu, Jianan Wang, Cheng Cheng, Yihan Zhu, Enyu Li, Yu Xiao, Jiangyong Xie, Duogong Yan, Jiangyi Chen
Computation and Language
Sakura at BEA 2026 Shared Task 1: What Makes Vocabulary Difficult?
Adam Nohejl, Xuanxin Wu, Yusuke Ide et al. (+3 more)
We describe two types of models for vocabulary difficulty prediction: a high-accuracy black-box model, which achieved the top shared task result in the open track, and an explainable model, which outperforms a fine-tuned encoder baseline. As the black-box model, we fine-tuned an LLM using a soft-target loss function for effective application to the rating task, achieving r > 0.91. The explainable model provides insights into what impacts the difficulty of each item while maintaining a strong correlation (r > 0.77). We further analyze the results, demonstrating that the difficulty of items in the British Council's Knowledge-based Vocabulary Lists (KVL) is often affected by spelling difficulty or the construction of the test items, in addition to the genuine production difficulty of the words. We make our code available online at https://github.com/ynklab/vocabulary-difficulty .
Read the full abstract ▾
All authors Adam Nohejl, Xuanxin Wu, Yusuke Ide, Maria Angelica Riera Machin, Yi-Ning Chang, Hitomi Yanaka
Computation and Language · Cross-listed from cs.LG
Active Learners as Efficient PRP Rerankers
Jeremías Figueiredo Paschmann, Juan Kaplan, Francisco Nattero et al. (+3 more)
Pairwise Ranking Prompting (PRP) elicits pairwise preference judgments from an LLM, which are then aggregated into a ranking, usually via classical sorting
algorithms. However, judgments are noisy, order-sensitive, and sometimes intransitive, so sorting assumptions do not match the setting. Because sorting aims to recover a full permutation, truncating it to meet a call budget does not produce a dependable top-K. We thus reframe PRP reranking as active learning from noisy pairwise comparisons and show that active rankers are drop-in replacements that improve NDCG@10 per call in the call-constrained regime. Our noise-robust framework also introduces a randomized-direction oracle that uses a single LLM call per pair. This approach converts systematic position bias into zero-mean noise, enabling unbiased aggregate ranking without the cost of bidirectional calls.
Read the full abstract ▾
All authors Jeremías Figueiredo Paschmann, Juan Kaplan, Francisco Nattero, Santiago Barron, Juan Wisznia, Luciano del Corro
Computation and Language · Cross-listed from cs.LG
DT-Transformer: A Foundation Model for Disease Trajectory Prediction on a Real-world Health System
Yunying Zhu, Andrew R Weckstein, Kueiyu Joshua Lin et al. (+1 more)
Accurate disease trajectory prediction is critical for early intervention, resource allocation, and improving long-term outcomes. While electronic health records (EHRs) provide a rich longitudinal view of patient health in clinical environments, models trained on curated research cohorts may not reflect routine deployment settings, and those trained on single-hospital
datasets capture only fragments of each patient's trajectory. This highlights the importance of leveraging large, multi-hospital health systems for training and validation to better reflect real-world clinical complexity. In this work, we develop DT-
Transformer, a foundation model trained on 57.1M structured EHR entries over 1.7M patients from Mass General Brigham (MGB), spanning 11 hospitals and a broad network of outpatient clinics. DT-
Transformer achieves strong discrimination in both held-out and prospective validation settings. Next-event prediction achieves a median age- and sex-stratified AUC of 0.871 across 896 disease categories, with all categories exceeding AUC 0.5. These results support health system-scale training as a path toward foundation models suited to real-world clinical
forecasting.
Read the full abstract ▾
All authors Yunying Zhu, Andrew R Weckstein, Kueiyu Joshua Lin, Jie Yang
Computation and Language · Cross-listed from cs.LG
Diagnosing Training Inference Mismatch in LLM Reinforcement Learning via a Zero-Mismatch Reference
Tianle Zhong, Neiwen Ling, Yifan Pi et al. (+5 more)
Modern LLM RL systems separate rollout generation from policy optimization. These two stages are expected to produce
token probabilities that match exactly. However, implementation differences can make them assign different values to the same sequence under the same model weights, inducing Training-Inference Mismatch (TIM). TIM is difficult to inspect because it is entangled with off-policy drift and common stabilization mechanisms. In this work, we isolate TIM in a zero-mismatch diagnostic setting (VeXact), and show that small
token-level numerical disagreements can independently cause training collapse. We further show that TIM changes the effective optimization problem, and identify a set of remedies that could mitigate TIM. Our results suggest that TIM is not benign numerical noise, but a systems-level perturbation that should be treated as a first-order factor in analyzing LLM RL stability.
Read the full abstract ▾
All authors Tianle Zhong, Neiwen Ling, Yifan Pi, Zijun Wei, Tianshu Yu, Geoffrey Fox, Peng Wu, Xiao Yu
Computation and Language · Cross-listed from cs.LG
PreFT: Prefill-only finetuning for efficient inference
Andrew Lanpouthakoun, Aryaman Arora, Zhengxuan Wu et al. (+4 more)
Large
language models can now be personalised efficiently at scale using parameter efficient finetuning methods (PEFTs), but serving user-specific PEFTs harms throughput, even with specialised kernels and
memory management techniques. This is because, theoretically and empirically, a mismatch exists between prefill (processing a large number of
tokens at once) and decode (generating a single
token autoregressively): the latter has far lower throughput when serving multiple adapters. Rather than optimising performance relative to parameter count, for efficient multi-adapter serving, we instead ought to optimise performance relative to serving throughput. We therefore propose PreFT (Prefill-only Finetuning), wherein we only apply the adapter to prefill
tokens and discard it afterwards. PreFT significantly increases throughput with minimal effect on performance. We develop and release an efficient implementation of two prefill-only PEFTs, LoRA and ReFT, on the vLLM inference engine. We first show that serving multi-user PreFTs is more efficient than traditional PEFTs ($1.9\times$ the throughput when serving $512$ adapters on Llama 3.1 70B). Then, we compare the performance of prefill-only vs. all-
token adapters on a variety of supervised finetuning and reinforcement learning tasks with LMs at varying scales. On SFT, we observe that the evaluation loss of PreFTs is higher than PEFTs, but can be compensated by increasing rank with nearly no reduction in throughput. On RL, we consistently find that PreFTs approach parity with standard PEFTs. Together, this work validates prefill-only adaptation of LLMs as a more favourable accuracy-throughput tradeoff than existing PEFTs for personalised serving.
Read the full abstract ▾
All authors Andrew Lanpouthakoun, Aryaman Arora, Zhengxuan Wu, Dhruv Pai, Ben Keigwin, Dan Jurafsky, Christopher Potts