Peer Revenue An exchange of ideas

Papers

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

Computer Science Computation and Language

cs.CL Watch Computation and Language

Tuesday 2 June

Computation and Language

PhotoCraft: Agentic Reasoning with Hierarchical Self-Evolving Memory for Deep Image Search

Kailin Lyu, Zhiqiang Yuan, Jianwei He et al. (+9 more)

Deep Image Search requires multi-step reasoning over rich contextual cues, such as time, location, and event relations. However, most existing LLM-based agents are stateless and reactive, lacking persistent memory to maintain long-horizon context or transfer experience across tasks, which often leads to execution drift and experience isolation. To address these limitations, we propose PhotoCraft, a training-free, hierarchical memory system for photo-search agents. Inspired by human cognition, PhotoCraft equips MLLMs with working, episodic, and semantic memory, which are dynamically invoked during reasoning to preserve logical consistency and knowledge transferability throughout multi-step reasoning and answer generation. Extensive experiments on DISBench demonstrate that PhotoCraft consistently improves context-aware retrieval across diverse MLLM backbones, achieving gains of up to 18.5% and effectively mitigating key bottlenecks in memoryless deep image search, offering a practical path toward reliable and generalizable multimodal search agents.
Read the full abstract

All authors Kailin Lyu, Zhiqiang Yuan, Jianwei He, Qiwei Yan, Xuanbo Su, Nanxing Hu, Yang Liu, Ce Hao, Shengqian Qin, Lianyu Hu, Jinchao Zhang, Jie Zhou

Computation and Language

Can Factual Opinions Be Edited (Manipulated) in Large Language Models?

Yuanpu Cao, Ziyi Yin, Fenglong Ma et al. (+1 more)

Large Language Models (LLMs) are increasingly integrated into various domains, making knowledge editing techniques crucial yet potentially hazardous. Current editing methods primarily target atomic facts, overlooking the significant risks associated with manipulating factual opinions, e.g., documented stances of public figures on societal issues. Such manipulation could reshape public images, influence elections, and alter societal views. To systematically assess this threat, we introduce the Factual Opinion Editing with Evidence (FOE) benchmark, which encompasses 261 public figures, 19 issue categories, and 2,178 complete opinion records. Our evaluations demonstrate that current editing techniques struggle significantly with factual opinions, often achieving only superficial changes while failing to preserve consistency between the edited opinion and the supporting evidence generated by the model. To address this limitation, we further propose a simple yet effective Self-Generated Evidence-Aligned method that achieves opinion-evidence alignment without relying on explicit instructions. Together, our benchmark and method provide a foundation for understanding the emerging security implications of factual opinion editing in LLMs.
Read the full abstract

All authors Yuanpu Cao, Ziyi Yin, Fenglong Ma, Jinghui Chen

Computation and Language · Cross-listed from cs.LG

Multi-component Causal Tracing in Large Language Models

Zirui Yan, Dennis Wei, Dmitriy A. Katz et al. (+2 more)

Causal tracing systematically intervenes on a large language model's (LLM's) internal representations to uncover and quantify the causal pathways linking specific inputs or computations to specific metrics of interest, quantifying the LLM's behavior. Building on previous single-component or single-layer studies, this paper presents a unified framework for causally tracing multiple components simultaneously. This framework systematically identifies the subsets of components (e.g., attention heads and multi-layer perceptron neurons) most critical to a desired target performance metric (e.g., accuracy and fairness). This is achieved by incorporating flexible interventions applied to a wide range of desired metrics. To address the combinatorial complexity of the multi-component problem, an efficient algorithm is designed that leverages soft interventions and a carefully designed metric transformation, converting the combinatorial search problem into a continuous one that can be solved efficiently under proper constraints, thereby generating proper binary decisions for selecting components. Experimental results demonstrate that the proposed method efficiently identifies subsets of the model's components that have a high impact on the target metric, outperforming existing baseline approaches. Our code is available at https://github.com/ZiruiYan/multi-component-causal-tracing.
Read the full abstract

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

Computation and Language

Regret Pre-training: Bridging Prior and Posterior Views for Enhanced Knowledge Grounding

Mingkuan Zhao, Xiayu Sun, Wentao Hu et al. (+5 more)

Causal language models factorize sequence probabilities using only preceding context, leaving future information unexploited during training despite its availability in the training data. This paper introduces Regret Pre-training, a self-supervised framework grounded in the Learning Using Privileged Information (LUPI) paradigm. The framework employs a dual-view architecture in which a single model generates both a causal Student distribution and a future-conditioned Teacher distribution. The training objective augments standard language modeling with a regret loss that minimizes the KL divergence from teacher to student, transferring future-aware signals to the causal representations. We investigate two teacher configurations on the OLMoE-1B-7B architecture:LocalRegret, which extends attention by one future token, andGlobalRegret, which conditions on bidirectional context with the target position masked. Experiments on nine downstream tasks following 4 billion tokens of training demonstrate that both configurations consistently outperform the baseline. On average,GlobalRegret andLocalRegret achieve 33.9% and 32.2% accuracy respectively, surpassing the baseline's 30.2%. Most notably,GlobalRegret improves BoolQ performance by 18.1 percentage points (61.0% vs 42.9%). The framework introduces no additional parameters and requires only one extra inference-mode forward pass per training step.
Read the full abstract

All authors Mingkuan Zhao, Xiayu Sun, Wentao Hu, Suquan Chen, Jiaxuan Li, Xiaoyan Zhu, Xin Lai, Jiayin Wang

Computation and Language

G^2C-MT: Graph-Guided Context Selection for Document-Level Machine Translation

Baijun Ji, Zixuan Zhou, Xiangyu Duan et al. (+4 more)

Effective document-level machine translation (DocMT) requires capturing long-range discourse dependencies. Recent work has explored retrieval-based and discourse-aware context selection. However, these approaches often lack an explicit mechanism for modeling structured discourse dependencies between distant paragraphs in a document. In this paper, we propose G^2C-MT (Graph-Guided Context for Machine Translation), which views DocMT context selection as a structured path discovery problem on a lightweight discourse graph, rather than retrieving unstructured context sets or relying on expensive LLM-based discourse modeling. In detail, we represent each paragraph as a node and model the relationship between each pair of nodes, considering their semantic similarity, adjacency, and keyword overlap. Furthermore, we propose a depth-biased random walk over the graph to sample a backward context path for each target paragraph. The context path will be used to prompt a large language model (LLM) for translation. This framework naturally supports multi-path context sampling, which can improve robustness by aggregating diverse translation candidates for discourse-ambiguous inputs. Experiments conducted across various domains show that G^2C-MT outperforms strong baselines on multiple LLMs, including DeepSeek-V3, Gemini-2.5-Flash-lite, and the Qwen-2.5/3 series.
Read the full abstract

All authors Baijun Ji, Zixuan Zhou, Xiangyu Duan, Yu Liu, Longbo Sun, Rupu Wei, Bohong Zhao

Computation and Language · Cross-listed from cs.LO

ZX-Calculus:Trace-Indexed Dependent Types and Epistemic Semantics

Peng Chen

We propose ZX-Calculus (Knowledge Evolution Calculus), a conservative extension of Martin-Lof Dependent Type Theory (MLTT) integrating trace-indexed types, presheaf non-monotone semantics, and constructive AGM belief revision. A Coq mechanisation accompanies the paper (34 complete proofs; zero admits for the two central results). (I) Trace types. FinTrace(s0,sn) is an inductive family of typed execution traces. FinTrace and Star(Step) are isomorphic as path types but not judgementally equal; TraceElim exposes the event label e:Event explicitly, giving a more ergonomic interface for event-driven induction. We prove the Trace-Reachability Correspondence, Deterministic Replay, and a canonicity framework via reducibility candidates with a Transport Lemma (RC-elim deferred; all other Core results are Coq-verified). (II) Sheaf semantics. Trace-indexed propositions are contravariant sheaves over the free trace partial-order category Tf. A Separation Theorem (explicit countermodel) distinguishes proof-theoretic monotonicity from semantic non-monotonicity. The term model is an initial CwF (syntactic universal property, not classical completeness). (III) AGM belief revision. We give an explicit constructive partial meet contraction algorithm verified against (C1)-(C4). All eight AGM postulates (R1)-(R8) are theorems. Proofs of R7 and R8 use the Disjunctive Entrenchment Lemma, given a self-contained constructive derivation. (IV) Integration. B^AGM fails the sheaf composition law BP-comp for sequential revision (explicit countermodel, Coq-verified). We introduce Single-Step Revision Systems (SSRS), prove B^AGM is a valid SSRS (Coq-verified), and show this suffices for trace morphisms, retraction characterisation, and revision witnesses. The BP-comp failure reveals a fundamental tension between path-dependent belief revision and functor consistency, not previously identified.
Read the full abstract

Computation and Language

The Geometry of LLM-as-Judge: Why Inter-LLM Consensus Is Not Human Alignment

Sourabrata Mukherjee, Hamna Hamna, Kalika Bali et al. (+1 more)

LLM judges now score most open-ended NLP output, and their mutual agreement is routinely read as evidence that the scores can be trusted. That reading is unsafe: judges may agree because they capture quality, or because they share the same blind spots, and agreement statistics alone cannot tell these apart. We develop a geometric test that can. Treating each judge's scores as a vector, we measure spread, effective rank, the angle to human scores, and the judge-judge, judge-human, and human-human agreement triple for 42 judges on two community-built Indic benchmarks covering four domains and eight languages. Every comparison is reference-matched: a judge and a held-out rater are scored against the same two-rater mean, since an averaged reference otherwise flatters judges by several degrees. On subjective rubrics, judges agree with one another as much as humans do yet reach only 58-66% of human agreement and often concentrate on an axis humans do not weight. On the one rubric with a verifiable answer, most of that gap closes. Ensembles converge on the judges' shared axis rather than the human one, and training widens scores without rotating them. Inter-judge agreement is evidence of human alignment only after this check passes.
Read the full abstract

All authors Sourabrata Mukherjee, Hamna Hamna, Kalika Bali, Sunayana Sitaram

Computation and Language

The Deliberative Illusion: Diagnosing Factual Attrition and Stance Homogenization in Multi-Agent LLM Deliberation

Herun Wan, Jiaying Wu, Minnan Luo et al. (+4 more)

Multi-agent LLM systems often treat consensus as evidence of successful interaction. For deliberative problems, however, reliability depends on whether agents preserve the facts and viewpoints needed to interpret an issue. We identify the deliberative illusion: discussion produces (1) factual attrition, the progressive loss of issue-critical facts, alongside (2) stance homogenization, the collapse of diverse positions toward consensus. To measure this process, we introduce DelibTrace, a framework that decomposes each issue into atomic facts, labels issue-critical ones, distributes them across agents, and tracks their survival across discussion rounds. Across ethical and news-based deliberation with three representative LLM families, multi-agent discussion erases up to 72% of issue-critical facts. This loss is consequential: retained evidence can reconstruct the issue misleadingly, final stances remain anchored in base-model priors, and a single malicious agent can inject misinformation into the shrinking shared context. These results reveal a sharper risk: agents can agree more while knowing less. We call for evaluations that measure which facts, uncertainties, and legitimate disagreements survive interaction.
Read the full abstract

All authors Herun Wan, Jiaying Wu, Minnan Luo, Fanxiao Li, Ningnan Wang, Nancy F. Chen, Min-Yen Kan

Computation and Language

Conditional Hypothesis Generation for LLM-Based Text Analysis with Researcher-Specified Covariates

Paiheng Xu, Jing Liu and Wei Ai

A core goal of computational social science is to discover interpretable differences in how language varies across outcomes of interest, such as political affiliation or instructional quality. Recent LLM-based hypothesis generation methods describe such differences in natural language, but select for globally discriminative patterns without accounting for covariates that shape the data based on researchers' domain knowledge. When covariates are ignored, selected patterns can reflect confounds rather than differences of substantive interest. We introduce conditional hypothesis generation, a framework that incorporates researcher-specified covariates to steer hypothesis discovery toward differences that hold within relevant subgroups. Two challenges arise: the target subgroup may be underrepresented (stratum imbalance), and the direction of a difference may reverse across subgroups (sign reversal). We propose two econometrics-inspired methods: one introduces feature–covariate interactions to detect sign reversals, and the other applies within-stratum demeaning and inverse-frequency reweighting to equalize underrepresented strata. Synthetic experiments show each method outperforms global baselines in its targeted setting, and expert evaluation on two real-world datasets confirms that covariate-aware generation surfaces more useful hypotheses within relevant subgroups.
Read the full abstract

Computation and Language

SEA-LION-Embedding: Open and Reproducible Text Embeddings for Southeast Asia

Peerat Limkonchotiwat, Raymond Ng, Sarana Nutanong et al. (+1 more)

Text embeddings are fundamental to many downstream applications, making robustness important for real-world NLP. However, most recent state-of-the-art embedding models are not reproducible because they rely on closed or undisclosed training data, and they remain insufficiently robust for Southeast Asian languages. We present SEA-LION-Embedding, a fully open and reproducible text-embedding pipeline for Southeast Asian languages trained only on publicly available data, and use it to study three core factors of robust embedding design: data composition, training objective, and base encoder initialization. SEA-LION-Embedding achieves state-of-the-art results on SEA-BED while enabling systematic and reproducible analysis of robust text embeddings for the region.
Read the full abstract

All authors Peerat Limkonchotiwat, Raymond Ng, Sarana Nutanong, Jian Gang Ngui

Computation and Language

Hallucinations as Orthogonal Noise: Inference-Time Manifold Alignment via Dynamic Contextual Orthogonalization

Mingkuan Zhao, Wentao Hu, Tianchen Huang et al. (+6 more)

Hallucination in Large Language Models (LLMs), characterized by the generation of content inconsistent with contextual facts or logical constraints – remains a persistent challenge for reliable deployment. In this work, we address this issue through a geometric framework rooted in the linear representation hypothesis. We propose that hallucinations manifest as orthogonal noise relative to the semantic manifold of the residual stream. Specifically, we hypothesize that while attention heads ideally propagate information congruent with the context subspace, hallucinations arise when specific heads introduce components orthogonal to this subspace, disrupting the coherence of the latent representation. Based on this formulation, we introduce Dynamic Contextual Orthogonalization (DCO), an inference-time intervention method. DCO utilizes the input residual stream as a dynamic context anchor to perform orthogonal decomposition on attention head outputs. To distinguish between context-aligned semantic updates and divergent noise, DCO employs a layer-wise Z-score suppression mechanism that selectively attenuates outlier orthogonal components based on statistical distributions. Evaluations on Llama-3-8B and 70B across benchmarks such as XSum, NQ-Swap, and IFEval demonstrate that DCO achieves superior contextual faithfulness compared to state-of-the-art intervention baselines. Furthermore, DCO maintains high performance on knowledge-intensive tasks like TriviaQA and TruthfulQA, effectively mitigating the trade-off between hallucination suppression and parametric knowledge retention often observed in existing methods. Our findings validate the geometric interpretation of hallucinations and establish DCO as a computationally efficient approach for enforcing manifold alignment.Our code is available at https://github.com/Harry-Miral/DCO
Read the full abstract

All authors Mingkuan Zhao, Wentao Hu, Tianchen Huang, Yuheng Min, Suquan Chen, Yide Gao, Yanbo Zhai, Shuangyong Song, Xuelong Li

Computation and Language

Hint-Guided Diversified Policy Optimization for LLM Reasoning

Zhiyu Cao, Kaixin Wu, Mingjie Zhong et al. (+3 more)

Recent developments in Large Language Models (LLMs) have showcased impressive reasoning capabilities, with Reinforcement Learning with Verifiable Rewards (RLVR) being a promising enhancement strategy. However, existing reward mechanisms are constrained to the outcome-level correctness and lack explicit signals to guide the model to consider diverse solutions. In contrast, human problem solving typically involves evaluating multiple potential approaches and selecting the most reliable solution, a cognitive process that current RLVR frameworks do not explicitly incentivize. Inspired by this, we propose Hint-Guided Diversified Policy Optimization (HDPO), allowing the model to first list all potential candidate solution outlines as hints and then select the most reliable one for further reasoning. HDPO comprises two stages of Cold Start for Structured Reasoning and Hint-Guided Diversified Reinforcement Learning to incentivize the model to generate diverse and reliable solutions following the “propose-select-think” trajectory. Experimental results show that HDPO effectively boosts LLM reasoning and enhances the diversity of candidate solutions as well as the LLM's ability to identify reliable solutions.
Read the full abstract

All authors Zhiyu Cao, Kaixin Wu, Mingjie Zhong, Peifeng Li, Can Ye, Qiaoming Zhu

Computation and Language · Cross-listed from cs.AI

Learning Reusable Agent Subroutines from Experience

Zhihan Lei, Jiarui Yan, Joshua Momo et al. (+1 more)

Existing methods induce various forms of knowledge from agent traces, from sample workflows to programmatic skills. We propose ARISE (Agent Routine Induction from Successful Experience), which identifies recurring procedures in agent traces and synthesizes them into a reusable library of agent subroutines. Across five tasks in narrative deduction, rule application, and constraint planning, it is shown to be a more effective learning strategy than prompting methods and prose and code artifact induction methods. ARISE outperforms chain-of-thought in all ten model-task settings of the main evaluation. Compared to state-of-the-art methods in agent experience induction, our method scores 6.1 percentage points over the strongest evaluated baselines, and 11.5 percentage points over code-skill induction. In our controlled experiment, ARISE also improves average accuracy by up to 15.6 percentage points compared to formatting the induced knowledge as skills and by up to 19.1 points over rewriting it as code tools.
Read the full abstract

All authors Zhihan Lei, Jiarui Yan, Joshua Momo, William W. Cohen

Computation and Language

A Language Model from 1913: Pretraining on Historical Text

Xiaoxi Luo, Zachary Shinnick, Niclas Griesshaber et al. (+5 more)

While modern language models increasingly rely on ever-larger web corpora, we show that pretraining on historical text (e.g., pre-1913 text) in a data-constrained setting can produce a temporally grounded language model that still shows reasonable performance on language understanding. However, developing History LMs requires addressing challenges in data quality, preventing temporal leakage in post-training, and constructing temporally aligned evaluations. We address these challenges and pretrain TypewriterLM, a 7.24B-parameter model with a 1913 knowledge cutoff. We construct TypewriterCorpus, a 54B-token historical corpus with extensive temporal filtering, propose lexically grounded instruction tuning that constrains all responses to vocabulary from historical source documents, and introduce History-Event, a benchmark of 2,344 events for evaluating both competence and cutoff adherence. We release TypewriterLM and all associated resources to support future research on History LMs.
Read the full abstract

All authors Xiaoxi Luo, Zachary Shinnick, Niclas Griesshaber, Yixuan Wang, Junchi Yu, Freda Shi, Philip Torr, Yao Lu

Computation and Language

A Locally Deployed RAG-Based Academic Advising System for Course Selection

Feng Li and Yoritaka Iwata

The correct sequence of courses in the curriculum based on prerequisites between courses is of great importance for students to develop their knowledge and skills holistically. However, students crafting this sequence in isolation frequently struggle with recognition limitations and information overload that leads to confusion. Simultaneously, education institutions encounter difficulties in providing adequate academic advice for the correct sequence due to limited education resources. To address these challenges, we propose a locally deployed RAG-based academic advising system grounded in syllabus information. By combining large language models with retrieval from structured syllabus data, the system is designed to support course selection, prerequisite understanding, and personalized study planning in a privacy-preserving manner.
Read the full abstract

Computation and Language

Predicting Inference-Time Scaling Gains from Labeled Validation-Set Output Statistics

Luyang Zhang and Jingyan Li

Best-of-$N$ inference scaling (drawing $N$ candidate answers from a language model and returning the one a reward model ranks highest) improves accuracy by an amount that varies across models, but predicting that amount in advance currently requires running the procedure end-to-end. Prior work links cheap statistics of a model's sampled outputs and validation-set correctness (how often samples agree, how diverse they are, how confident the model is, and where correct samples appear) to model behavior, but does not isolate which of these form a stable, compact predictor of best-of-$N$ gain. We fit ridge predictors on features computed from a single labeled validation-set sampling pass, use bootstrap-Lasso as a stability analysis of the candidate feature set, and give a concentration analysis with an explicit linear-approximation residual. Across three base-model families, six post-training methods, and math and reasoning task domains, the stability analysis identifies a strict three-feature core spanning prompt-level agreement spread, label-assisted first-correct-sample position, and completion-length variance; a compact ridge predictor built from this core plus an entropy add-on reaches Spearman $\rho = 0.90$ with actual best-of-$N$ gain under a reward-model verifier. The intended use is labeled validation-set screening of candidate configurations before paying the full reward-model scoring cost.
Read the full abstract

Computation and Language

Memory Retrieval for Changing Preferences

Yuehan Qin, Li Li, Linxin Song et al. (+4 more)

Long-context dialogue systems must decide both when to access memory and which parts of the interaction history are relevant. Existing approaches typically rely on heuristic retrieval signals or always-on memory usage, failing to account for the changing and potentially inconsistent nature of user preferences. In this work, we propose a unified framework for memory access and selection based on changing preferences. We formulate personalized memory retrieval as identifying which historical turns provide evidence about a user's latent preference state, rather than relying on surface-level semantic similarity. To this end, we quantify the utility of each memory turn using a Bayes factor, defined as the improvement in the model's likelihood of the reference response when the turn is included in context. This provides a principled measure of evidence strength and a unified signal for both memory access and selection. By framing memory retrieval as utility estimation, the model learns to identify salient turns and regulate memory usage based on expected utility. Experiments on four heterogeneous memory benchmarks show that our approach outperforms existing embedding-based retrieval on long-context, preference-intensive tasks where modeling changing preferences is essential, while remaining competitive in low-density regimes where semantic similarity suffices.
Read the full abstract

All authors Yuehan Qin, Li Li, Linxin Song, Wei Yang, Jiate Li, Yuqing Yang, Yue Zhao

Computation and Language

Chatbots Output Meaningful (but Problematic) Language

Matthew Stone and Una Stojnić

Are utterances by AI chatbots meaningful? Concretely, if a user asks, say, Anthropic's agent Claude, "What is the capital of Spain?" and Claude answers, "Madrid is the capital of Spain," does that sentence have its ordinary meaning – and does it express a true proposition? Most ordinary users, as well as AI engineers, take the answer to be trivially "yes." However, many cognitive scientists, linguists, and philosophers of language argue that dominant intentionalist accounts of language and meaning deliver the opposite conclusion. Theorists more sympathetic to ordinary users' intuitions have therefore advocated a radical "de-anthropomorphization" of language, revising our understanding of mental states, intentions, and semantic content to capture the intuition that the outputs of LLMs are meaningful. We take a different approach. While we, too, argue that LLM outputs are meaningful, we contend that a proper theory of human language already applies, as is, to current chatbots. Meaning is a low bar: claiming that LLM outputs are meaningful does not require positing mental states, intentions, rationality, or the cognitive capacities requisite for communication in LLMs – or, indeed, making any other anthropomorphic assumptions. People do have communicative intentions (typically successful ones), but nevertheless, even in humans, language production can depart from what the speaker has in mind. Our view has important consequences for how we should theorize about – and critically engage with – both human linguistic output and synthetically generated text. In particular, to say that chatbots produce meaningful text is not by any means to endorse what they output, or to assume that the technology is (or is not) good, powerful, appropriate, or useful.
Read the full abstract

Computation and Language

EURO-5K: When Does Domain Pretraining Matter? Benchmarking Transformers for EU Reporting Obligation Extraction

Marios Koniaris, Vasileios Kotronis, Eugenia Giannini et al. (+1 more)

Extracting reporting obligations from EU legislation is critical for assessing and reducing regulatory reporting burden. However, distinguishing reporting requirements from structurally similar provisions requires specialised legal understanding. Current legal NLP methods lack specialised datasets with clear guidelines and comparative evaluation of extraction paradigms and domain adaptation strategies. We curate EURO-5K, a corpus of sentence-level reporting obligations and challenging negative examples from 136 EU legislative acts. On this dataset, we train and compare discriminative token-classification models (BERT-style) and generative span-extraction models (LLMs), evaluating both full fine-tuning and parameter-efficient QLoRA against baselines (pattern and dependency-based extraction, few-shot prompting). Results show that fully fine-tuned generic and legal BERT models achieve similar performance (0.89 F1), while fine-tuned LLMs match encoder accuracy for sentence-level extraction. Legal pretraining offers only small gains for generative models. In contrast, it is clearly beneficial when adaptation capacity is constrained, as parameter-efficient tuning of Legal-BERT outperforms its generic counterpart. Learning curve analysis demonstrates that legal pretraining accelerates early learning with minimal data. All approaches converge around 3K samples with diminishing returns thereafter, validating dataset sufficiency. Cross-dataset evaluation on two external regulatory corpora shows that our models behave as specialised reporting obligation extractors rather than generic regulatory classifiers. We release EURO-5K, trained models, and an interactive demo with explainability visualizations and structured RDF export. These demonstrate that both paradigms and parameter-efficient training provide practical tools for regulatory compliance automation.
Read the full abstract

All authors Marios Koniaris, Vasileios Kotronis, Eugenia Giannini, Panayiotis Tsanakas

Monday 1 June

Computation and Language

Thought-Aware KV Cache Compaction for Reasoning via Adaptive Attention Matching

Yang Liu, Bin Chong, Chongyang Zhang et al. (+3 more)

Reasoning language models generate lengthy chain-of-thought (CoT) sequences whose key-value (KV) cache grows linearly and becomes a memory bottleneck during decoding. Existing compaction methods treat reasoning trajectories as flat token sequences and apply uniform compression, ignoring the hierarchical structure of CoT reasoning where different steps vary drastically in importance. We propose Thought-Aware Attention Matching (TAM), which exploits this structure through three mechanisms: (i)~thought segmentation that decomposes the trajectory into reasoning blocks, (ii)~adaptive budget allocation that assigns compression budget based on each segment's importance and size, and (iii)~pivotal token protection that preserves high-attention reasoning anchors. We prove that the allocation rule is optimal under a convex error model and that cumulative error under sequential compaction remains bounded. Experiments on AIME 2024 and MATH-500 with Qwen3-4B show that TAM improves accuracy over uniform compaction at the same memory footprint, with periodic compaction bounding peak memory to 3.1–3.2\,GB (a 65% reduction) while maintaining competitive accuracy.
Read the full abstract

All authors Yang Liu, Bin Chong, Chongyang Zhang, Hao Zheng, Jiayu Liang, Xu Kefu

Computation and Language

Reliability-Aware Sexism Detection: Combining DPO with Annotator Agreement and Token-Level Confidence Scoring

Hadi Mohammadi, Shihan Wang, Masoume M. Raeissi et al. (+1 more)

The detection of online sexism remains an open problem. Sexism detection is inherently subjective, yet most existing systems reduce multi-annotator labels to a single majority decision and treat all instances uniformly. This ignores two informative signals: annotator agreement and model uncertainty. We propose RA-DPO (Reliability-Aware Direct Preference Optimization), which integrates annotator agreement, model confidence, and a token-level uncertainty signal into a single reliability score. RA-DPO uses this score to select high-value preference pairs during training and to support inference-time abstention, which allows the model to trade coverage for accuracy. We evaluate RA-DPO on 6,920 multilingual posts from EXIST 2023, fine-tune OpenAI gpt-4o base via DPO, and validate on two open-weight 3B models (Llama, Qwen). Results show that training on the top 30% most reliable pairs matches full-data DPO, which indicates that reliability-aware selection can reduce training cost without sacrificing performance. At inference, selective prediction reaches 96.2% accuracy at 50% coverage in the true-agreement setting and 88.7% in the deployable predicted-agreement setting, both exceeding the 85.3% no-agreement baseline. These results suggest that accounting for annotation uncertainty is beneficial for both efficient training and reliable deployment in subjective classification.
Read the full abstract

All authors Hadi Mohammadi, Shihan Wang, Masoume M. Raeissi, Anastasia Giachanou

Computation and Language

AnchorSIPS: A Synthetic Dataset and Evaluation Resource for Evidence-Supported Psychosis-Risk Symptom Measurement

Guilherme C. Oliveira, Stephanie Fong, Zimu Wang et al. (+10 more)

Progress on AI for psychosis-risk assessment is limited by a data-access bottleneck. Real clinical interviews are difficult to share because of privacy, governance, and consent constraints. We present AnchorSIPS, a synthetic dataset of 10K structured psychosis-risk interviews with transcript-grounded measurement targets. Each interview is modeled on Mini-SIPS, a clinician-administered psychosis-risk interview. It captures history, 24 symptom questions, follow-up evidence for items the patient affirms, decisions about delusion-like symptoms (unusual beliefs), hallucination-like symptoms (unusual perceptions), and disorganized communication, exclusion of clear psychotic-level symptoms ("frank psychosis"), and a final attenuated psychosis syndrome (APS) diagnosis, a high-risk state of milder or early psychotic symptoms. The APS diagnosis is not a standalone label. It depends on earlier endorsements, supporting follow-up details, symptom-class decisions, and the frank-psychosis check. Every intermediate decision is anchored to its supporting transcript turns. AnchorSIPS is generated by a plan-then-realize pipeline. A hidden case sheet specifies the patient's clinical state, a deterministic planner fixes the interview structure, and an LLM realizes only the patient utterances under validation and bounded repair. Fixing labels and structure before generation avoids the inter-turn inconsistencies typical of multi-turn LLM dialogue. Across seven LLM baselines, models recover coarse decisions but fail to extract follow-up details or cite supporting transcript turns, so final-label performance overstates interview competence. AnchorSIPS is intended for research on evidence extraction, transcript-grounded measurement, and uncertainty under partial disclosure.
Read the full abstract

All authors Guilherme C. Oliveira, Stephanie Fong, Zimu Wang, Clarice Lee, Xiangyu Zhao, Duy Khoa Pham, Duong Nhu, Yiwen Jiang, Jiahe Liu, Zhongxing Xu, Dwarikanath Mahapatra, Dominic Dwyer, Zongyuan Ge

Computation and Language · Cross-listed from cs.CV

PragyaDoc: A Universal Document Intelligence Framework for Multilingual Medical Document Understanding in Low-Resource Settings

Jagpal Singh Jhala

India's 22 official languages create a critical accessibility barrier: the majority of medical documentation exists exclusively in English, yet the patients who most urgently require this information - rural populations, ASHA workers, and patient families - are functionally excluded from understanding it. This paper presents PragyaDoc, a Universal Document Intelligence Framework that addresses this gap through a four-layer pipeline: a parallel ensemble OCR extraction layer, a geometric-lexical fusion layer, a deterministic domain structuring layer, and a dual-LLM medical reasoning and localization layer
Read the full abstract

Computation and Language

Adversarial Style Optimization: Enhancing VLM Jailbreaks by GRPO-based Stylistic Triggers Optimization

Bingjun Luo, Jialin Guo, Yue Yao et al. (+1 more)

Multimodal Large Language Models (MLLMs) have achieved impressive performance, but their safety alignment remains vulnerable to jailbreak attacks. Existing content-based jailbreaks are often inconsistent and show unsatisfying performance against the rapidly evolving MLLMs, failing to exploit non-content-based vulnerabilities. Unlike previous research, we empirically find that MLLMs exhibit a Stylistic Inconsistency between their comprehension ability and safety ability: MLLMs can robustly understand content regardless of visual style, yet their defense mechanisms can be easily bypassed by specific stylistic triggers. Based on this finding, we propose Adversarial Style Optimization (ASO), a plug-and-play enhancement module to amplify existing visual jailbreaks. ASO fine-tunes an image-editing model to superimpose an optimized stylistic modification onto a given adversarial image, using a Group Relative Policy Optimization (GRPO) agent guided by a Structurally-Tiered Reward Function that combines a logit-based signal for detecting explicit refusals with a high-fidelity semantic evaluation from a powerful judge model. Extensive experiments show that ASO significantly enhances the ASR of SOTA attacks, demonstrating that stylistic biases are a scalable vector for red-teaming MLLMs. Our code is available at https://github.com/bingjunluo/ASO.
Read the full abstract

All authors Bingjun Luo, Jialin Guo, Yue Yao, Xinpeng Ding

Computation and Language · Cross-listed from cs.IR

PIXELRAG: Web Screenshots Beat Text for Retrieval-Augmented Generation

Yichuan Wang, Zhifei Li, Zirui Wang et al. (+5 more)

Augmenting large language models (LLMs) with retrieved web text has become a dominant paradigm, yet the web is not natively textual: existing systems depend on complex parsing pipelines that linearize HTML and discard layout, visual structure, and formatting. We introduce PixelRAG, a new retrieval-augmented method that represents websites in their native visual form and performs retrieval and reading entirely in pixel space, enabling an end-to-end architecture that eliminates text abstraction. PixelRAG is, to our knowledge, the first pipeline to operate over a full Wikipedia corpus in this form, scaling to a datastore of 30 million screenshot images with an efficient visual retrieval index. Built on an existing visual embedding model (i.e., Qwen3-VL-Embedding), PixelRAG further fine-tunes this model on screenshot data with carefully curated contrastive training data. Retrieved screenshots are then fed directly as pixel inputs to a VLM, without intermediate text conversion. PixelRAG consistently outperforms both no-retrieval and text-based RAG baselines, most surprisingly on widely studied text-centric tasks such as NQ and SimpleQA. It also achieves strong gains on multimodal open-domain QA (e.g., MMSearch), benchmarks over noisy news corpora (e.g., LiveVQA), and agentic benchmarks (e.g., MoNaCo), improving accuracy by up to 18.1% over text-based baselines. Finally, pixel representations enable a new efficiency lever for RAG through image compression, achieving up to 3x token cost reduction at lower resolutions while maintaining accuracy. Our results challenge the necessity of text representations in web retrieval, suggesting that web RAG can operate directly in the web's native visual form while improving both performance and efficiency.
Read the full abstract

All authors Yichuan Wang, Zhifei Li, Zirui Wang, Paul Teiletche, Lesheng Jin, Matei Zaharia, Joseph E. Gonzalez, Sewon Min

Computation and Language · Cross-listed from cs.LG

Do Transformers Need Three Projections? Systematic Study of QKV Variants

Ali Kayyam, Anusha Madan Gopal and M Anthony Lewis

Transformers have become the standard solution for various AI tasks, with the query, key, and value (QKV) attention formulation playing a central role. However, the individual contribution of these three projections and the impact of omitting some remain poorly understood. We systematically evaluate three projection sharing constraints: a) Q-K=V (shared key-value), b) Q=K-V (shared query-key), and c) Q=K=V (single projection). The last two variants produce symmetric attention maps; to address this, we also explore asymmetric attention via 2D positional encodings. Through experiments spanning synthetic tasks, vision (MNIST, CIFAR, TinyImageNet, anomaly), and language modeling (300M and 1.2B parameter models on 10B tokens), we discovered that our transformers perform on par or occasionally better than the QKV transformer. In language modeling, Q-K=V projection sharing achieves 50% KV cache reduction with only 3.1% perplexity degradation. Crucially, projection sharing is complementary to head sharing (GQA/MQA): combining Q-K=V with GQA-4 yields 87.5% cache reduction, while Q-K=V + MQA achieves 96.9%, enabling practical on-device inference. We show that Q-K=V preserves quality because keys and values can occupy similar representational spaces and attention operates in a low-rank regime, whereas Q=K-V breaks attention directionality. Our results systematically characterize projection sharing as an underexplored instance of weight tying in attention, with direct, quantifiable inference memory benefits, particularly valuable for edge deployment. The code is publicly available at https://github.com/Brainchip-Inc/Do-Transformers-Need-3-Projections
Read the full abstract

Computation and Language · Cross-listed from cs.AR

Multi-Segment Attention: Enabling Efficient KV-Cache Management for Faster Large Language Model Serving

Chunan Shi, Yilei Chen, Yilin Chen et al. (+2 more)

Large Language Model (LLM) inference relies on key-value (KV) caches to avoid redundant attention computation. While approximate KV cache retention techniques reduce memory usage by sacrificing model accuracy, lossless approaches instead evict KV cache blocks from GPU memory and reconstruct them on demand to preserve exact outputs. Existing lossless KV cache management systems primarily base eviction decisions on access frequency or positional heuristics, without considering how different KV cache blocks affect the execution efficiency of GPU attention kernels. In this paper, we propose AsymCache, a computation-latency-aware KV cache management system for LLM inference that explicitly aligns cache residency decisions with GPU attention kernel performance, including three key components: Multi-Segment Attention (MSA) for efficient non-contiguous KV context processing, a cache eviction policy that jointly optimizes hit rate and position-aware recomputation cost, and an adaptive chunking scheduler for high hardware utilization. Experiments show that AsymCache reduces TTFT by up to 1.90-2.03x and time-per-output-token (TPOT) by 1.62-1.71x over latest baselines, confirming the effectiveness of the method in common workloads and validating its design goal of balancing computational efficiency with cache hit rate. Moreover, the low-level design of AsymCache allows seamless integration into agent serving systems such as Continuum, where it further reduces average job latency by up to 18.1%.
Read the full abstract

All authors Chunan Shi, Yilei Chen, Yilin Chen, Xupeng Miao, Bin Cui

Computation and Language

Fast-dLLM++: Fréchet Profile Decoding for Faster Diffusion LLM Inference

Siva Rajesh Kasa, Yasong Dai, Sumit Negi et al. (+1 more)

Diffusion large language models promise parallel token generation, yet inference remains bottlenecked by deciding which masked tokens can be safely committed together. Fast-dLLM addressed this with KV caching and confidence-guided parallel decoding, but its decoding theory uses a homogeneous high-confidence assumption that effectively reduces each candidate set to its weakest selected token. We argue that this leaves speed on the table because real decoding steps exhibit heterogeneous confidence profiles. We propose Fast-dLLM++, a training-free extension that introduces Fréchet profile decoding: selecting parallel commit sets from the full sorted confidence profile rather than a single worst-case confidence. The resulting rule is a heterogeneous-confidence generalization of Fast-dLLM's factor selector and it recovers the previous rule exactly in the equal-confidence case and adds a provable heterogeneity bonus when the selected tokens have uneven confidences. Fast-dLLM++ leaves the model, diffusion process, and cache implementation entirely unchanged, making it a drop-in replacement for existing Fast-dLLM decoding. Experiments on GSM8K, MATH, HumanEval, and MBPP with the LLaDA-8B model show that the theoretical improvement translates directly into empirical gains: profile-aware selection improves the accuracy–throughput frontier by exploiting safe parallelism that weakest-token rules miss, achieving up to 37% higher throughput at comparable accuracy. Our code release is at https://github.com/Ringo-Star/FastdLLM_plusplus.
Read the full abstract

All authors Siva Rajesh Kasa, Yasong Dai, Sumit Negi, Hongdong Li

Computation and Language

Linguistic Productivity in Large Language Models: Models Coerce, but do not Preempt

Claire Bonial, Claire Benet Post, Laura Michaelis et al. (+1 more)

Usage-based theories of grammar explain language users' ability to interpret and produce novel combinations of words through two complementary inferential strategies. In cases of coercion, a language user, when encountering a word that is mismatched to its grammatical context, reinterprets that word in a way that resolves the conflict. In cases of preemption, the user infers that an otherwise plausible structure is unavailable or dispreferred because it fails to occur in contexts where it would be expected. Large Language Models are also usage-based, in the sense that the structures of language are learned through exposure to vast amounts of text. Here, we test whether or not the opposing forces of coercion and preemption also encourage and constrain linguistic productivity in LLMs. We demonstrate across models of various sizes and architectures that larger models recognize constructional productivity in cases of coercion, even when these involve novel words. However, we also show that even the largest models do not replicate the conservative language strategy of preemption for novel words: in particular, they do not appear to draw on negative evidence when making choices among competing grammatical options. We conclude that coercion is detectable because it is both surprising and semantically motivated: when a word is mismatched to its syntactic context, the word's meaning can be adjusted to fit the context. Preemption has no such signature. There is no record of the lexeme-construction pairings that were avoided–only of those that were produced. This asymmetry helps to explain why models coerce (create mismatched sign combinations) but do not preempt (avoid unconventional yet semantically defensible ones).
Read the full abstract

All authors Claire Bonial, Claire Benet Post, Laura Michaelis, Harish Tayyar Madabushi

Computation and Language · Cross-listed from cs.RO

SCOPE: Real-Time Natural Language Camera Agent at the Edge

Nikolaj Hindsbo, Sina Ehsani and Pragyana Mishra

Deploying language-driven agents in robotics requires evaluations that reflect real-world task demands: natural-language instructions with reproducible outcomes. Such agents must connect language models to callable perception and control tools, and be assessed using deployment-critical metrics including latency, accuracy, and error modes. We present SCOPE (Simulation and Camera Operations for Perception and Evaluation), a modular agent for natural-language, open-vocabulary pan-tilt-zoom (PTZ) camera control and visual scene understanding, designed explicitly for edge deployment. SCOPE operates both in a Blender-based simulation environment and on a physical PTZ camera, executing all perception, planning, and control locally at the deployment site using edge-accessible compute. We release a 536-task benchmark spanning QA, single- and multi-step commands, counting, spatial reasoning, descriptions, and optical character recognition in a Blender-based simulation environment that exposes realistic PTZ control affordances. Execution traces are combined with an LM-as-Judge to evaluate latency, accuracy, and error modes. We evaluate 19 planner-perception model combinations pairing Qwen3 small language models (SLMs) with Moondream and Qwen vision-language models (VLMs). Stronger SLMs substantially reduce hallucinations and improve tool routing, leading to more reliable closed-loop behavior. Once a sufficiently capable SLM is used, perception becomes the dominant performance bottleneck. Mixture-of-Experts models on both the planning and perception side consistently match or exceed dense alternatives at latencies and memory footprints comparable to much smaller networks. Quantization provides additional efficiency gains with minimal accuracy degradation, identifying a practical, sim-to-real validated design point for real-time, edge-feasible language-driven PTZ control.
Read the full abstract