cs.CL · 2026-06-29 · No. 38

Computation and Language, 2026-06-29.

25 new papers in cs.CL. Titles, authors, abstracts. Links to arXiv. Want this in your inbox every morning? Subscribe →

01 — The papers

25 entries
  1. 01

    Cognitive Episodes in LLM Reasoning Traces Enable Interpretable Human Item Difficulty Prediction

    Chenguang Wang, Ming Li, Xinyue Zeng, Zhuochun Li, Hong Jiao, Tianyi Zhou, Dawei Zhou

    cs.CL · cs.AI · cs.CY · cs.LG

    Predicting human item difficulty is central to educational assessment, where reliable estimates support fairness and effective test construction. Existing methods often depend on costly human calibration or item-level textual representations, providing limited evidence about the cognitive processes that make items difficult. We argue that difficulty should be viewed not only as a property of item text, but also as an observable consequence of...

    arxiv.org/abs/2606.28186 · PDF

  2. 02

    From Tokens to States: LLMs as a Special Case of World Models and the Continuous Path Beyond

    Paul Dubois

    cs.CL · cs.AI · cs.LG

    The AI community has framed the relationship between large language models (LLMs) and world models as a dichotomy: LLMs predict tokens; world models simulate reality. Yann LeCun argues in 2022 that reaching general intelligence requires abandoning autoregressive token prediction in favour of latent-space architectures. This framing is unnecessarily binary. Two claims will be defended. First, LLMs are a degenerate special case of world models:...

    arxiv.org/abs/2606.28127 · PDF

  3. 03

    MultiHashFormer: Hash-based Generative Language Models

    Huiyin Xue, Atsuki Yamaguchi, Nikolaos Aletras

    cs.CL · cs.AI · cs.LG

    Language models (LMs) represent tokens using embedding matrices that scale linearly with the vocabulary size. To constrain the parameter footprint, prior work proposes hashing many tokens into a single vector within encoder-only models. While this offers parameter efficiency, many-to-one collisions prevent its use in causal LMs. In this paper, we propose MultiHashFormer, a new framework that allows hash-based autoregression. Each token is...

    arxiv.org/abs/2606.28057 · PDF

  4. 04

    Can LLMs Judge Better Than They Generate? Evaluating Task Asymmetry, Mechanistic Interpretability and Transferability for In-Context QA

    Sambaran Bandyopadhyay

    cs.CL · cs.AI

    LLM-as-a-Judge and self-evaluation pipelines implicitly assume that evaluation is easier than generation. We test this in a controlled in-context QA setting where a context passage is the sole information source and each model judges the answer it generated, removing the parametric-knowledge confound of open-domain comparisons. Across four benchmarks (SQuAD 2.0, DROP, HotpotQA, MuSiQue) and two models, evaluation is not uniformly easier:...

    arxiv.org/abs/2606.28050 · PDF

  5. 05

    Dialogue to Detection: A Multimodal Hybrid NLP Pipeline for Insurance Fraud Detection

    Muhammad Shakeel Akram, Amal Htait, Abdul Hamid Sadka, Emma Meisingseth, Karishma Jaitly

    cs.CL · cs.AI · eess.AS

    Insurance fraud imposes substantial financial losses and operational inefficiencies, raising premiums and impacting trust among legitimate policyholders. Early detection at FNOL remains a persistent challenge. Existing approaches rely largely on private, text-only datasets, limiting progress on multimodal methods that integrate linguistic, behavioural, and speaker-based indicators. We introduce a synthetic multimodal framework that replicates...

    arxiv.org/abs/2606.28002 · PDF

  6. 06

    From Black-Box to Clinical Insight: A Multi-Stage Explainable Framework for Speech-Based Cognitive Impairment Detection

    Yasaman Haghbin, Sina Rashidi, Ali Zolnour, Fatemeh Taherinezhad, Ali Fartoot, Hossein Azadmaleki, James M Noble,...

    cs.CL · cs.AI

    Speech-based cognitive impairment detection offers a noninvasive, accessible alternative to costly biomarker assays, yet transformer-based models remain clinically uninterpretable. We propose a multi-stage explainability framework that translates black-box transformer predictions into clinically grounded narratives by integrating SHapley Additive exPlanations (SHAP)-based token attribution, theory-informed linguistic features, and a...

    arxiv.org/abs/2606.27973 · PDF

  7. 07

    VASAE: Naming SAE Dictionary Directions with Vocabulary-Aligned Anchoring

    Kairui Zhang, Ziwen Yu, Zahraa S. Abdallah, Martha Lewis

    cs.CL · cs.AI · cs.LG

    Sparse autoencoders (SAEs) provide useful decompositions of Transformer residual streams, but their learned features are usually named post hoc rather than directly connected to the Transformer's token vocabulary. We introduce Vocabulary-Aligned Sparse Autoencoder (VASAE), a method that trains SAE features under vocabulary-aligned anchoring and assigns each feature an intrinsic token name: the token string whose embedding is nearest to that...

    arxiv.org/abs/2606.27941 · PDF

  8. 08

    Triadic Werewolf: A Jester Role for Multi-Hop Theory of Mind in LLMs

    Avni Mittal

    cs.CL · cs.AI · cs.GT · cs.MA

    Theory-of-mind evaluations of large language models typically use dyadic social-deduction games, where every observable cue points to a single hidden side, so a model with strong language priors can score well without ever simulating opponents' incentives. We extend the Werewolf game with a Jester, a third faction whose utility on peer suspicion is inverted because it wins by being voted out, so optimal play requires reasoning across three...

    arxiv.org/abs/2606.27909 · PDF

  9. 09

    A Study of Temporal Fusion Strategies for Named Entity Recognition in Historical Texts

    Emanuela Boros

    cs.CL · cs.AI

    Temporal variation poses a unique challenge for named entity recognition (NER) in historical texts, where entities drift in surface form and salience across time. While language models (LMs) have made progress in various NLP tasks, their ability to reason about temporality, especially in diachronic contexts, remains limited or at least, questionable. In this paper, we systematically study how temporal metadata can be structurally embedded...

    arxiv.org/abs/2606.27881 · PDF

  10. 10

    Position Bias Correction is Insufficient for One-Pass Attention Sorting

    Qiong Tang, Xiangkun Hu, Xiangyang Liu, Yiran Chen, Yunfan Shao

    cs.CL · cs.AI

    Long-context language models suffer from position bias, where information in middle positions is underutilized. Attention Sorting addresses this by iteratively reordering documents based on attention patterns, but its multiple sort-and-generate cycles increase deployment cost. We hypothesize that position bias is the primary bottleneck and propose Debiased One-Pass Attention Sorting, which estimates a per-prompt position-bias curve from the...

    arxiv.org/abs/2606.27793 · PDF

  11. 11

    NLL-Guided Full-Attention Layer Selection for Training-Free Sliding-Window Adaptation

    Qiong Tang, Xiangkun Hu, Xiangyang Liu, Yiran Chen, Yunfan Shao

    cs.CL · cs.AI

    Hybrid attention models that mix full and sliding-window attention across layers offer a promising approach to efficient long-context inference, but the critical question of \emph{which layers} should retain full attention remains unsolved. Existing methods use either fixed periodic patterns or attention-based heuristics that may not capture what matters for downstream accuracy. We propose NLL-guided layer selection, a training-free method...

    arxiv.org/abs/2606.27791 · PDF

  12. 12

    SHIFT: Gate-Modulated Activation Steering for Knowledge Conflict Mitigation in Retrieval-Augmented Generation

    Ruochang Li, Pengcheng Huang, Zhenghao Liu, Yukun Yan, Huiyuan Xie, Yu Gu, Ge Yu, Maosong Sun

    cs.CL · cs.AI

    Retrieval-augmented generation (RAG) enhances LLMs by incorporating external knowledge to support response generation. However, conflicts between retrieved context and parametric knowledge have emerged as a critical challenge in RAG systems. To mitigate such conflicts, numerous studies have attempted to identify and edit knowledge-related internal neurons, aiming to improve the ability of LLMs to rely on contextual evidence during generation....

    arxiv.org/abs/2606.27786 · PDF

  13. 13

    Output-Space Allocation Costs for Calibration-Guided LLM Compression: An Empirical Study

    Qiong Tang, Xiangkun Hu, Xiangyang Liu, Yiran Chen, Yunfan Shao

    cs.CL · cs.AI

    Training-free compression methods for large language models (LLMs) often use calibration data to guide compression decisions. ROCKET, a recent method combining sparse-dictionary factorization with multi-choice knapsack problem (MCKP) allocation, derives its per-layer factorization from an output reconstruction objective but uses weight-space Frobenius error as the MCKP allocation cost. We investigate whether aligning the allocation cost with...

    arxiv.org/abs/2606.27785 · PDF

  14. 14

    KG2Cypher: Data-Centric Pipeline for Building Enterprise Text-to-Cypher Systems

    Minjun Choi, Yerin Kim, Junghyuk Seo, Sujin Mo, Hyemin Lee, Youngjoong Ko

    cs.CL · cs.AI

    Enterprise Knowledge Graphs (KGs) are increasingly used for internal search, analytics, and question answering, but building natural-language interfaces for private enterprise graphs remains costly. We present KG2Cypher, a data-centric pipeline for building enterprise text-to-Cypher systems from existing KGs. KG2Cypher first constructs an executable Cypher query from observed graph facts and then uses LLMs to generate its associated...

    arxiv.org/abs/2606.27742 · PDF

  15. 15

    Enhancing Numerical Prediction in LLMs via Smooth MMD Alignment

    Zhuo Zuo, Li Yue, Wenhao Zheng, Chenpeng Wang, Xianggen Liu

    cs.CL · cs.AI · cs.CE · cs.LG

    Despite their strong general capabilities, large language models (LLMs) often remain unreliable when outputs must be numerically precise. A key reason is the training objective: standard cross-entropy treats numeric tokens as unstructured categories and ignores the metric structure of their values. We address this mismatch with Smooth Maximum Mean Discrepancy (SMMD), which builds on the classic MMD by incorporating value-distance kernels over...

    arxiv.org/abs/2606.27731 · PDF

  16. 16

    Do Speech Emphasis Models Generalize across Languages and Emotions?

    Megan Wei, Deepali Aneja, Jiaqi Su, Yunyun Wang, Haonan Chen, Zeyu Jin

    cs.CL · cs.AI · cs.LG · cs.SD · eess.AS

    Prosodic emphasis varies across languages, emotions, and speaking styles, yet existing emphasis detection models are largely trained and evaluated on monolingual neutral read speech. We introduce MMEE (Multilingual Multi-Emotion Emphasis), a corpus of 10,000 professionally recorded expressive utterances (14.13 hours) across 7 languages and 34 emotion/style categories, with three-level perceptual labels (10 annotations per sample). We...

    arxiv.org/abs/2606.27717 · PDF

  17. 17

    Low-Agreeableness Persona Conditioning for Safe LLM Fine-Tuning

    Austin MY Cheung, Yi Yang

    cs.CL · cs.AI

    Recent work has shown that fine-tuning large language models (LLMs) for social warmth degrades factual reliability and increases sycophancy. We investigate a related but distinct failure mode: warmth fine-tuning also weakens adversarial safety, making models more susceptible to jailbreaks and harmful output generation. We examine whether this reflects an inherent consequence of empathetic adaptation or an artifact of data construction. To...

    arxiv.org/abs/2606.27709 · PDF

  18. 18

    Mitigating LLM-based p-Hacking by Preregistering for the Next LLM

    Maria Thomas, Kristina Gligoric, Nihar B. Shah

    cs.CL · cs.AI · cs.DL

    Large language models (LLMs) are increasingly used to generate, classify, and annotate data whose outputs feed downstream hypothesis tests. However, LLM-based research is easy to p-hack: a researcher can tune the prompts, decoding parameters, or output format until a desired result is reached. We propose a protocol to mitigate p-hacking in LLM-based research: preregistering the experiment and eligible models, and then running it on the first...

    arxiv.org/abs/2606.27687 · PDF

  19. 19

    From Signals to Transfer: A Factorised Study of Probe-Based Uncertainty Estimation in Large Language Models

    Ponhvoan Srey, Xiaobao Wu, Cong-Duy Nguyen, Quang Minh Nguyen, Duc Anh Vu, Anh Tuan Luu

    cs.CL · cs.AI

    Probe-based uncertainty estimation (UE) has emerged as a prominent approach to detect hallucinations in Large Language Models (LLMs) by learning uncertainty from internal model signals. Yet, recent methods vary simultaneously across feature design, training data construction, and evaluation setting, obscuring what actually drives performance. To address this issue, we propose a factorised study of probe-based UE under matched conditions. Our...

    arxiv.org/abs/2606.27679 · PDF

  20. 20

    Cross-Platform Chinese Offensive Comment Detection via Dual-Threshold Hard Example Mining

    Ruixing Ren, Junhui Zhao, Fangfang Wang

    cs.CL · cs.AI · eess.SY

    Cross-platform deployment of offensive comment detection for Chinese social media suffers performance degradation. The paper proposes a dual-threshold hard mining method to address this. First, the clean-Chinese-base RoBERTa is finetuned on COLD to establish a binary baseline for fair comparison. Second, a three-class fine-labeled test set covering Weibo, Xiaohongshu, Tieba, and Zhihu is constructed, domain distances from the source are...

    arxiv.org/abs/2606.27629 · PDF

  21. 21

    Masked Language Flow Models

    Iskander Azangulov, Kianoosh Ashouritaklimi, Leo Zhang, Simon Vary, Patrick Rebeschini

    cs.CL · cs.LG

    Masked Diffusion Models (MDMs) promise fast, parallel language generation, but their reverse transition factorises across token positions -- an approximation that breaks down in the few-step sampling regime where parallel generation ought to provide the greatest efficiency gains. Flow Language Models (FLMs) sidestep this limitation by learning a continuous flow that transports noise toward clean sequences represented in Euclidean space,...

    arxiv.org/abs/2606.27617 · PDF

  22. 22

    Narrative-UFET: Narrative Generation for Ultra-Fine Entity Typing

    Mreedul Gupta, Advait Deshmukh, Ashwin Umadi, Matt Pauk, Maria Leonor Pacheco

    cs.CL · cs.AI

    Ultra-fine entity typing (UFET) assigns highly specific types to entity mentions, but current approaches struggle with types in the long tail. We hypothesize that a key limitation is the reliance on sentence-level context, since disambiguating evidence is often spread across multiple sentences. Testing this has been difficult because all existing UFET resources are sentence-level. We present Narrative-UFET, a controlled extension of UFET in...

    arxiv.org/abs/2606.27598 · PDF

  23. 23

    EntMTP: Accelerating LLM Inference with Entropy Guided Multi Token Prediction

    Carrie Chen

    cs.CL · cs.LG

    Multi-token prediction has been shown to increase data density during training, improve downstream text-generation quality, and serves as the defacto approach for self-speculative decoding. Existing foundation and open source models that use MTP heads commit to a static tree-based attention topology throughout the entire generation sequence, meaning the speculation depth, and thus the compute required during verification, stays constant...

    arxiv.org/abs/2606.27550 · PDF

  24. 24

    The Context-Ready Transformer

    Mahesh Godavarti

    cs.CL · cs.AI

    We introduce the context-ready transformer, a new recurrent neural network architecture built from a D-layer transformer block that pre-contextualizes each token before it enters the block. During left-to-right generation, a correction network combines the previous position's block output -- a cached summary of past context -- with the current token embedding, so the tokenenters the block already contextualized rather than as a raw embedding....

    arxiv.org/abs/2606.27538 · PDF

  25. 25

    Supersede: Diagnosing and Training the Memory-Update Gap in LLM Agents

    Vedant Patel

    cs.CL · cs.AI · cs.LG

    Large language model (LLM) agents operate over long, multi-session interactions in which facts change: a user moves, a price updates, a plan is revised. Acting correctly requires using the current value of a fact and discarding values that have been superseded. We isolate this ability on real conversational data and show that it is a distinct, unsolved failure. On the knowledge-update subset of LongMemEval, replacing an agent's full context...

    arxiv.org/abs/2606.27472 · PDF

This edition is part of The Daily Abstract — cs.CL archive. Subscribe to receive these in your inbox each morning, automatically translated to Spanish, with reply-to-PDF: arxivdaily.ignorelist.com.

Colophon Set in Georgia, with system sans for interface chrome and a monospaced stack for code and paper identifiers. Sole accent: amber #D99C5E. Built and served on an always-free VM. The masthead is set 14% letterspaced because newspapers do that and it works.