探索全球前沿学术脉络

01.

arXiv (CS.CL) 2026-06-11 DOI: arXiv:2602.03141

Short Chains, Deep Thoughts: Balancing Reasoning Efficiency and Intra-Segment Capability via Split-Merge Optimization

作者:

Runquan Gui↗Jie Wang↗Zhihai Wang↗Chi Ma↗Jianye Hao↗Feng Wu↗

While Large Reasoning Models (LRMs) have demonstrated impressive capabilities in solving complex tasks through the generation of long reasoning chains, this reliance on verbose generation results in significant latency and computational overhead. To address these challenges, we propose CoSMo (Consistency-Guided Split-Merge Optimization), a framework designed to eliminate structural redundancy rather than indiscriminately restricting token volume. Specifically, CoSMo utilizes a split-merge algorithm that dynamically refines reasoning chains by merging redundant segments and splitting logical gaps to ensure coherence. We then employ structure-aligned reinforcement learning with a novel segment-level budget to supervise the model in maintaining efficient reasoning structures throughout training. Extensive experiments across multiple benchmarks and backbones demonstrate that CoSMo achieves superior performance, improving accuracy by 3.3 points while reducing segment usage by 28.7\% on average compared to reasoning efficiency baselines.

阅读与讨论 → 访问原文 →

02.

arXiv (CS.AI) 2026-06-15 DOI: arXiv:2511.07368

Distributional Biases in Post-Training: A Markovian Analysis of Reasoning Trajectories

作者:

Dake Bu↗Wei Huang↗Andi Han↗Atsushi Nitanda↗Bo Xue↗Qingfu Zhang↗Hau-San Wong↗Taiji Suzuki↗

arXiv:2511.07368v3 Announce Type: replace-cross Abstract: Foundation models exhibit broad knowledge but limited task-specific reasoning, motivating post-training strategies such as RL with verifiable rewards (RLVR) and test-time scaling (TTS). While recent work highlights the role of exploration in improving pass@K, empirical evidence points to a paradox: RLVR and ORM/PRM typically reinforce existing paths rather than expanding the reasoning scope, raising the question of why exploration helps if no new patterns emerge. To reconcile this paradox, we adopt the perspective of Kim et al. (2025), viewing easy (e.g., simplifying a fraction) versus hard (e.g., discovering the some symmetry) reasoning steps as low versus high probability Markov transitions. In this tractable model, pretraining corresponds to tree-graph discovering, while post-training corresponds to CoT reweighting. We provably show that, both RLVR and ORM/PRM would favor heavily to several high-probability paths, and thereby forget rare-but-crucial CoTs. Building on this, we further prove that exploration strategies such as rejecting easy instances and KL regularization help preserve rare CoTs. Empirical simulations corroborate our theoretical results.

阅读与讨论 → 访问原文 →

03.

arXiv (CS.AI) 2026-06-16 DOI: arXiv:2606.15899

SkillVetBench: LLM-as-Judge for Multi-Dimensional Security Risk Evaluation in Open-Source LLM Agent Skills

作者:

Ismail Hossain↗Sai Puppala↗Md Jahangir Alam↗Tanzim Ahad↗Sajedul Talukder↗

arXiv:2606.15899v1 Announce Type: cross Abstract: Open-source LLM agent ecosystems are growing rapidly, yet the security of community-contributed skills - modular tool definitions that extend agent capabilities - remains largely unvetted. The gap we fill: existing scanners operate at the code layer and are structurally blind to instruction-layer and multi-agent risk - natural-language directives that hijack an agent, exfiltrate data through encoded side channels, or chain harm across pipelines - so what is needed is a semantic, multi-dimensional vetting system rather than another signature matcher. We present SKILLVETBENCH, a live public leaderboard on Hugging Face that uses an LLM-as-Judge to vet agent skills. What is new: SARS (Skill Agentic Risk Score), a five-dimensional agentic-risk metric with a principled weighted formula for instruction-following systems. What is integrated: full CVSS v4.0 vector decomposition and a ClawHub dual-view that places our LLM-generated review beside the official marketplace verdict. What is demonstrated: drawing on our companion benchmark paper [ 1], the LLM-as-Judge stage achieves zero false negatives across 78 confirmed-malicious skills and zero false positives across 22 benign controls, while the best static baseline (SKILLSIEVE) still misses 15%; for instruction-layer categories such as Prompt Injection and Memory Poisoning, conventional tools miss between 89% and 100% of threats (e.g., CODEBERT detects none of nine memory-poisoning skills). Detection rates vary from 35% to 95% across four LLM evaluators, motivating ensemble scoring in production deployments.

阅读与讨论 → 访问原文 →

04.

arXiv (CS.CL) 2026-06-11 DOI: arXiv:2603.06910

Language Shapes Mental Health Evaluations in Large Language Models

作者:

Jiayi Xu↗Xiyang Hu↗

Multilingual large language models (LLMs) are increasingly used in socially sensitive mental health contexts, including support chatbots, screening, and content moderation. This raises a reliability question: do semantically equivalent mental health inputs elicit comparable evaluations across languages, or systematic shifts consistent with language-associated social and cultural contexts? We examine this question in an English-Chinese setting with GPT-4o and Qwen3-32B using a two-level framework: construct-level evaluative orientation, measured by psychometric stigma instruments, and decision-level behavior, measured by binary stigma detection and four-class depression severity classification. Across instruments and models, Chinese prompts elicit higher stigma-related scores than English prompts. At the decision level, Chinese prompts reduce sensitivity to stigmatizing content and produce more conservative depression severity judgments, leading to more under-estimation errors. These findings show that prompt language can shift both evaluative orientation and downstream behavior in LLM-based mental health evaluation. They highlight the need to evaluate multilingual LLMs not only for aggregate performance, but also for whether they apply comparable evaluative standards across languages in socially sensitive domains.

阅读与讨论 → 访问原文 →

05.

bioRxiv (Bioinfo) 2026-06-18 DOI: HASH:3fd1513f17e693a820d87068c3efe76b

Predicting optimal growth temperatures of bacteria using learned structural information from a single protein

作者:

Hoffert↗Myerscough↗Dragone↗N. B↗Gebert↗M. J↗Silberg↗J. J↗Fierer↗

Temperature is a fundamental determinant of bacterial physiology and ecology. Optimal growth temperature (OGT) is highly variable across species, contributing to differences in where and when species are most likely to thrive. Although the OGTs for most bacteria remain unknown, the increasing availability of genomes from uncultivated and cultivated taxa has made it advantageous to build genomic, cultivation-independent models to infer OGT. However, pre-existing genomic models often lack the generalizability and mechanistic grounding required for robust inferences of OGT. We propose a novel framework for predicting bacterial OGT which uses learned protein structural signatures of thermal adaptation. We hypothesize that biophysical tradeoffs which dictate enzymatic functions across variable temperatures provide a more robust empirical basis for OGT prediction than broad genomic features. Our OGT-predicting model, ROSEATE, is based on a single gene, adenylate kinase (ADK), that encodes for a ubiquitous enzyme essential for energy homeostasis. ROSEATE uses high-dimensional latent space encoding via MSA Transformer, a protein language model which embeds ADKs in a manner which preserves biophysical information about embedded proteins. We show that the accuracy of the ROSEATE model is on par with other genome-based models, has a high degree of phylogenetic generalizability, and the ESM embeddings effectively capture key temperature-adaptive enzyme characteristics derived from AlphaFold structures. Because ROSEATE is based on analyses of a single ubiquitous protein, it can be used with metagenomic data to infer the community-level variation in bacterial OGTs. We demonstrate this feature of ROSEATE by reconstructing ADK sequences from over 500 environmental and host-associated metagenomes, successfully distinguishing community-wide thermal preferences across diverse habitats, from polar oceans to mammalian guts. By transitioning from genomic proxies to informationally dense protein structural features, this work provides an efficient, interpretable tool for predicting bacterial OGTs across taxa and whole communities.

阅读与讨论 → 访问原文 →

06.

arXiv (CS.AI) 2026-06-18 DOI: arXiv:2606.18594

Benchmarking Action Spaces in Reinforcement Learning for Vision-based Robotic Manipulation

作者:

Seyed Alireza Azimi↗Homayoon Farrahi↗Abhishek Naik↗Colin Bellinger↗A. Rupam Mahmood↗

arXiv:2606.18594v1 Announce Type: cross Abstract: In real-world reinforcement learning (RL), the choice of action space can play a key role in shaping motion smoothness, safety, and overall task performance. In this study, we evaluate pose increment, pose velocity, joint position increment, and joint velocity across two vision-based manipulation tasks: object picking and pushing. We train policies in simulation and deploy them to the real world using sim-to-real transfer. We find that action-space representation indeed significantly affects sim-to-real performance. In particular, we find that the joint velocity action space is best for the vision-based picking and pushing tasks in terms of smoothness and final task performance. We also provide practical guidance for RL practitioners in choosing action spaces for both simulation and real-world experiments.

阅读与讨论 → 访问原文 →

07.

arXiv (CS.CL) 2026-06-17 DOI: arXiv:2606.17092

Securing Multi-Agent GIS Systems: Risk Evaluation and Prompt Hardening Optimization

作者:

Kyle Gao↗Pranavi Kotta↗Linlin Xu↗Jonathan Li↗David A. Clausi↗

Agentic systems are increasingly integrated with geographic information systems (GIS), where multi-agent coordination enables complex conversational and spatial analysis but introduces security risks. This work presents a security-oriented framework for risk identification, evaluation, and mitigation in a multi-agent GIS system while maintaining adaptability to broader agentic architectures. We test the agentic system of a commercial geospatial partner while developing a modular state-machine-based orchestration framework that abstracts agent behavior into reusable components. We evaluate robustness using a red-teaming framework with an adaptive attacker LLM and a deterministic judge that produces binary outcomes with supporting rationales across multi-turn attacks. We further improve resilience with a prompt optimization framework that treats prompts as structured signatures and injects adversarial demonstrations, enabling systematic security improvements without degrading task performance.

阅读与讨论 → 访问原文 →

08.

arXiv (CS.CV) 2026-06-19 DOI: arXiv:2606.19662

Learning When to Denoise: Optimizing Asynchronous Schedules for Latent Diffusion

作者:

Bingshuo Qian↗Xiang Cheng↗

Multi-representation diffusion models can improve visual synthesis by denoising complementary views of an image, but their performance depends critically on the asynchronous schedule that determines when each representation is denoised. We propose to learn this schedule. Our method formulates asynchronous flow matching over multiple representation spaces and uses a schedule-corrected objective that keeps each representation's local noising-time weights fixed as the schedule changes. We instantiate the schedule with a flexible parametric class that is convex and monotone by construction, and learn it using a fast joint probe with less than 1% additional training compute. On ImageNet 256x256, the learned schedule substantially improves both convergence speed and final quality under a matched 675M-parameter XL backbone. With AutoGuidance, our 200-epoch model reaches FID 1.05, matching the 800-epoch SFD-XL baseline with 4x less training. Training to 600 epochs further improves to FID 1.02, outperforming the 1B-parameter SFD-XXL result of FID 1.04 while using a smaller model. In the unguided setting, our 200-epoch model reaches FID 2.37, already below the best 800-epoch SFD-XL result (2.54) at 4x less training, and improves to FID 2.14 at 600 epochs. Code is available at https://github.com/bsq532087/LWD

阅读与讨论 → 访问原文 →

09.

arXiv (quant-ph) 2026-06-15 DOI: arXiv:2504.17731

Synchronization of Quasi-Particle Excitations in a Quantum Gas with Cavity-Mediated Interactions

作者:

Gabriele Natale↗Alexander Baumg\"artner↗Justyna Stefaniak↗David Baur↗Simon Hertlein↗Dalila Rivero↗Tilman Esslinger↗Tobias Donner↗

arXiv:2504.17731v2 Announce Type: replace-cross Abstract: Driven-dissipative quantum systems can undergo transitions from stationary to dynamical phases, reflecting the emergence of collective non-equilibrium behavior. We study such a transition in a Bose-Einstein condensate coupled to an optical cavity and develop a cavity-assisted Bragg spectroscopy technique to resolve its collective modes. We observe dissipation-induced synchronization at the quasiparticle level, where two roton-like modes coalesce at an exceptional point. This reveals how dissipation microscopically drives collective dynamics and signals a precursor to a dynamical phase transition.

阅读与讨论 → 访问原文 →

10.

arXiv (CS.LG) 2026-06-12 DOI: arXiv:2606.13657

Dense Supervision, Sparse Updates: On the Sparsity and Geometry of On-Policy Distillation

作者:

Guo Yu↗Wenlin Liu↗Yulan Hu↗Hao-Xuan Ma↗Jun-Peng Jiang↗Han-Jia Ye↗

arXiv:2606.13657v1 Announce Type: new Abstract: On-policy distillation (\textsc{OPD}) has recently become a prominent post-training recipe as it combines two desirable ingredients: on-policy student trajectories and dense teacher supervision, yet how this hybrid changes a model's parameters remains unclear. Across several language and vision-language model pairs and use cases, our analysis yields two main findings. On sparsity, \textsc{OPD}-style updates are small and coordinate-sparse. They are distributed across layers and are usually FFN-heavy. This sparse structure is operationally useful: training only the discovered subnetwork recovers nearly the same performance as full \textsc{OPD}. However, the sparsity-inducing SGD optimizer underperforms AdamW in our optimizer ablation, likely because dense teacher supervision preserves heterogeneous coordinate-wise gradient scales where AdamW's adaptive scaling remains useful. On geometry, the updates are numerically full-rank but spectrally concentrated; they lie mostly away from the principal singular subspaces of the source weights and fall disproportionately on coordinates where the source weights are close to zero. These findings suggest that dense teacher supervision does not turn \textsc{OPD} into ordinary dense parameter rewriting; instead, \textsc{OPD} retains important geometric signatures of on-policy post-training.

阅读与讨论 → 访问原文 →

11.

arXiv (quant-ph) 2026-06-12 DOI: arXiv:2606.12685

Multiple Topological Haldane Phases for Symmetry-Protected Quantum Information Processing

作者:

Jo\~ao Pedro Gama D'Elia↗Irene D'Amico↗Thereza Paiva↗

arXiv:2606.12685v1 Announce Type: new Abstract: Symmetry-protected topological phases have attracted significant interest at the fundamental level and as a potential platform for quantum information processing, owing to their protected edge states and resilience to perturbations. Applying these features for practical and efficient quantum computation is highly desirable, but remains an open challenge. Here, we demonstrate the partitioning into multiple independent Haldane phase subsystems of a single spin-1/2 ladder system and propose this as a scalable architecture for gate-based quantum computation, which takes advantage of the symmetry-protected topological order. We encode qubits in the two topological states of the $S^{z}=0$ sector of each subsystem. Finite-size effects, typically viewed as detrimental, instead provide a controllable energy splitting that enables single-qubit rotations using only local magnetic fields. An Ising-type interaction between neighboring subsystem edges generates entangling gates, enabling universal quantum computation driven by two control parameters that are easily accessible experimentally. Our results demonstrate how symmetry-protected topological phases can be directly harnessed for circuit-model quantum computation in realistic systems.

阅读与讨论 → 访问原文 →

12.

arXiv (quant-ph) 2026-06-17 DOI: arXiv:2606.17751

Cavity method for permutation models on Cayley trees

作者:

Masayuki Ohzeki↗

arXiv:2606.17751v1 Announce Type: new Abstract: Motivated by permutation statistical models arising in random tensor networks, we study permutation models on a Cayley tree whose variables take values in the symmetric group $\Sn$. The pair interaction is assumed to depend only on the cycle type of the relative permutation. Then the Boltzmann weight is written as a class function on $\Sn$. This property diagonalizes the edge convolution operator in irreducible representation sectors. As a result, the linear stability of the uniform paramagnetic cavity solution is controlled by the character eigenvalue ratios. For cycle-factorized weights, these eigenvalues can be expressed as specializations of Schur functions. We derive the instability criteria and also verify their validity by comparison with direct numerical iterations of the cavity equation.

阅读与讨论 → 访问原文 →

13.

medRxiv (Medicine) 2026-06-22 DOI: HASH:763a9e8b1ac419f857e7be3be64047ca

Leishmaniasis on YouTube: a critical appraisal of the quality, reliability, and transparency of educational content

作者:

NANTALAGA↗K. C↗Nantege↗

Background: Leishmaniasis is a neglected tropical disease of significant global public health importance, for which accurate information is essential to support prevention and early care-seeking, particularly in endemic, resource-limited settings. YouTube is a widely used source of health information, but the quality and reliability of leishmaniasis-related content have not been evaluated. We aimed to assess the quality, reliability, and transparency of English-language YouTube videos on leishmaniasis. Methods: We conducted a cross-sectional analysis of YouTube videos retrieved via the YouTube Data API on 15 June 2026 using the terms "leishmaniasis," "cutaneous leishmaniasis," and "visceral leishmaniasis." After applying eligibility criteria and screening the 150 most-viewed eligible videos, 48 videos were included. Two reviewers independently assessed each video using the modified DISCERN (mDISCERN) tool, the Global Quality Score (GQS), and the JAMA benchmark criteria, with disagreements resolved by consensus. Inter-rater agreement was assessed using the intraclass correlation coefficient (ICC), and associations were examined using Spearman's rank correlation. Results: Of 402 videos retrieved, 48 met the inclusion criteria. The median GQS was 3.00 (IQR 2.00-4.00) and median mDISCERN was 3.00 (IQR 2.38-4.50), indicating moderate quality and reliability, while the median JAMA score was 2.00 (IQR 1.00-2.00), reflecting limited transparency; no video met all four JAMA criteria. The overwhelming majority of videos (47/48, 97.9%) were of professional or institutional origin. Inter-rater agreement was good to excellent (ICC 0.883 for GQS, 0.896 for mDISCERN, 1.000 for JAMA). The instruments were strongly inter-correlated (mDISCERN-GQS rho = 0.841, p < 0.001). Quality scores did not correlate positively with views, likes, or video duration; comments correlated weakly and negatively with mDISCERN (rho = -0.337, p = 0.031) and JAMA (rho = -0.381, p = 0.014). Conclusions: YouTube videos on leishmaniasis are of moderate quality and reliability but limited transparency, and are produced almost exclusively by professional sources. Video popularity, length, and age were not indicators of quality. There is a need for experts and institutions to produce clearly authored, well-sourced, and transparent educational content on this neglected tropical disease.

阅读与讨论 → 访问原文 →

14.

arXiv (CS.AI) 2026-06-19 DOI: arXiv:2606.19588

Analyzing the Narration Gap in LLM-Solver Loops

作者:

Zunchen Huang↗Songgaojun Deng↗

arXiv:2606.19588v1 Announce Type: new Abstract: Formal tools such as SAT and SMT solvers are increasingly embedded in language model reasoning pipelines when a safety or security critical question can be formulated in logic. Unlike chain of thought whose steps are sampled from the model distribution without formal guarantee, a solver produces a sound and independently verifiable answer. However, the soundness guarantee can be lost in the interaction between the solver and the model. The hybrid pipeline has three components: formalizing the question, deciding it, and narrating the result. Prior work has studied the formalization and decision, but not narration, which is the step that turns a formal tool's output into the user answer. To fill the narration gap, we first model the LLM-solver loop as a verified decision procedure. We further evaluate five open-sourced models under prompt injection, and we find certificate gating makes the solver verdict sound, while an adversary can invert a verified conclusion across phrasings and channels. We study the mitigation through hardened prompt that reduces injection significantly but cannot eliminate it and still suffers under adaptive attack. Combining the formal analysis and empirical studies, we show in the LLM-solver loop, robustness does not reach to the answer that the user finally reads.

阅读与讨论 → 访问原文 →

15.

arXiv (quant-ph) 2026-06-16 DOI: arXiv:2606.16375

Scalable Graph State Generation with O(1) Local Feedforward in Quantum Networks

作者:

Xiaoyi Zheng↗Chan-Tong Lam↗Lin Chen↗Zheng Xing↗

arXiv:2606.16375v1 Announce Type: new Abstract: The development of quantum networks faces a key challenge: the contradiction between probabilistic long-range entanglement generation and finite coherence time. Existing routing protocols typically focus on global state computation or path optimization. As the network scales up, classical delays accumulate and exacerbate decoherence, leading to a decrease in entanglement fidelity. To reduce routing decision delays to levels far below the coherence time of qubits, we propose a protocol based on local measurement and classical feedforward. This protocol reduces the local decision complexity to amortized O(1) level, ensuring that the decision delay is always much smaller than the coherence time of qubits. We map this protocol onto a dual-species trapped-ion platform and perform hybrid simulations. The results show that the proposed protocol performs well in terms of both resource efficiency and time feasibility. Noise analysis indicates that readout fidelity is the main bottleneck of this protocol, but noise suppression can be achieved by employing an erasure transformation in the dual-species architecture, combined with spatial multiplexing and branch independence, thereby ensuring the generation of high-fidelity star subgraphs. This protocol provides a clear path to achieving high-fidelity star subgraphs. These subgraphs can serve as general modules, merging to construct arbitrary subgraphs, providing a feasible solution for future fault-tolerant distributed quantum computing.

阅读与讨论 → 访问原文 →

16.

arXiv (CS.LG) 2026-06-17 DOI: arXiv:2503.04507

The Morse Transform for Discrete Shape Analysis

作者:

Alexander M. Tanaka↗Aras T. Asaad↗Richard Cooper↗Vidit Nanda↗

arXiv:2503.04507v2 Announce Type: replace-cross Abstract: The geometry of an object plays a vital role in modulating its interactions with the physical world. It nevertheless remains difficult to describe geometric information numerically for the purposes of statistical inference or classification tasks. Here, we introduce a new topological transform which leverages directional piecewise-linear Morse theory to quantify the geometry of an embedded object by cataloguing critical points across multiple height-functions. The output of this Morse transform records both the heights and the local topological type (peak, trough or saddle) of the critical points that characterise the underlying shape, retaining finer information than the Euler characteristic transform whilst naturally prioritising a shape's outermost regions. Crucially, this output can be further compressed into a rich but compact feature vector. We benchmark the Morse feature vector as a descriptor for ligand-based virtual screening (LBVS), which intrinsically depends on the shape of molecules. Under a common gradient-boosted tree classification pipeline, Morse descriptors achieve the highest mean AUROC when compared to other topological transform descriptors and to standard shape-based LBVS descriptors.

阅读与讨论 → 访问原文 →

17.

arXiv (CS.CV) 2026-06-16 DOI: arXiv:2606.10550

LentiAvatar: Pseudo-Multiview Reconstruction and Subpixel Prism Rendering for Real-Time Stereoscopic Communication

作者:

Chufeng Fang↗Dongdong Teng↗Lilin Liu↗

Real-time stereoscopic video communication has long been a goal of immersive telepresence, yet practical systems still require specialized capture rigs or reduce remote users to a single portrait view. We present LentiAvatar, a Gaussian head-avatar system that connects monocular avatar capture with subpixel-encoded glasses-free lenticular display for real-time autostereoscopic communication. From a monocular portrait video, LentiAvatar reconstructs a controllable head avatar and optimizes it for the lateral viewing zones induced by the display. The method uses natural head turns as pseudo-multiview (PMV) supervision to constrain regions that are otherwise weakly observed in monocular training, including hair, ears, jaw contours, and neck boundaries. Reliable side frames are yaw-binned, aligned to virtual cameras, and supervised within a strict head-and-hair domain; contour-aware losses and staged regularization further suppress ghosting, alpha leakage, and depth instability while preserving lateral detail. At runtime, LentiAvatar renders 32 virtual views and encodes them into a 4K lenticular raster with calibrated subpixel-routing masks. The live-tracker prototype sustains 10.65 FPS, and a subject-specific distilled driver raises the same display pipeline to 38.49 FPS.

阅读与讨论 → 访问原文 →

18.

arXiv (CS.CL) 2026-06-18 DOI: arXiv:2602.06470

Improve Large Language Model Systems with User Logs

作者:

Changyue Wang↗Weihang Su↗Qingyao Ai↗Xingzhao Yue↗Rui Zhang↗Xiaojia Chang↗Yiqun Liu↗

Scaling training data and model parameters has long driven progress in large language models (LLMs), but this paradigm is increasingly constrained by the scarcity of high-quality data and diminishing returns from rising computational costs. As a result, recent work is increasing the focus on continual learning from real-world deployment, where user interaction logs provide a rich source of authentic human feedback and procedural knowledge. However, learning from user logs is challenging due to their unstructured and noisy nature. Vanilla LLM systems often struggle to distinguish useful feedback signals from noisy user behavior, and the disparity between user log collection and model optimization (e.g., the off-policy optimization problem) further strengthens the problem. To this end, we propose UNO (User log-driveN Optimization), a unified framework for improving LLM systems (LLMsys) with user logs. UNO first distills logs into semi-structured rules and preference pairs, then employs query-and-feedback-driven clustering to manage data heterogeneity, and finally quantifies the cognitive gap between the model's prior knowledge and the log data. This assessment guides the LLMsys to adaptively filter out noisy feedback and construct different modules for primary and reflective experiences extracted from user logs, thereby improving future responses. Extensive experiments show that UNO achieves state-of-the-art effectiveness and efficiency, significantly outperforming Retrieval Augmented Generation (RAG) and memory-based baselines. We have open-sourced our code at https://github.com/bebr2/UNO .

阅读与讨论 → 访问原文 →

19.

arXiv (CS.CL) 2026-06-18 DOI: arXiv:2502.02904

ScholaWrite: A Dataset of End-to-End Scholarly Writing Process

作者:

Khanh Chi Le↗Linghe Wang↗Minhwa Lee↗Ross Volkov↗Luan Tuyen Chau↗Dongyeop Kang↗

Writing is a cognitively demanding activity that requires constant decision-making, heavy reliance on working memory, and frequent shifts between tasks of different goals. To build writing assistants that truly align with writers' cognition, we must capture and decode the complete thought process behind how writers transform ideas into final texts. We present ScholaWrite, the first dataset of end-to-end scholarly writing, tracing the multi-month journey from initial drafts to final manuscripts. We contribute three key advances: (1) a Chrome extension that unobtrusively records keystrokes on Overleaf, enabling the collection of realistic, in-situ writing data; (2) a novel corpus of full scholarly manuscripts, enriched with fine-grained annotations of cognitive writing intentions. The dataset includes \LaTeX-based edits from five computer science preprints, capturing nearly 62K text changes over four months; and (3) analyses and insights into the micro-dynamics of scholarly writing, highlighting gaps between human writing processes and the current capabilities of large language models (LLMs) in providing meaningful assistance. ScholaWrite underscores the value of capturing end-to-end writing data to develop future writing assistants that support, not replace, the cognitive work of scientists.

阅读与讨论 → 访问原文 →

20.

arXiv (CS.CL) 2026-06-16 DOI: arXiv:2606.15770

ttda704 at SemEval-2026 Task 6: Structured Chain-of-Thought Prompting for Political Evasion Detection

作者:

Tai Tran Tan↗An Dinh Thien↗

This paper describes our system for SemEval-2026 Task 6, which addresses the classification of political evasion strategies in English question-answer pairs extracted from U.S. presidential interviews. We systematically compare two distinct paradigms: (1) Parameter-Efficient Fine-Tuning of Qwen3 models (4B-32B) using QLoRA, enhanced with tiered upsampling and weighted cross-entropy loss to address severe class imbalance, and (2) structured Chain-of-Thought (CoT) prompting of reasoning-capable API models, namely DeepSeek-V3.2 and Grok-4-Fast. Our evaluation demonstrates that structured CoT prompting of reasoning-enabled models substantially outperforms our baseline parameter-efficient fine-tuning implementation in absolute Macro F1. Our best system, Grok-4-Fast with extended reasoning and few-shot hierarchical CoT prompting, achieves a Macro F1 of 0.5147 on Subtask 2 (9-class evasion) and 0.7979 on Subtask 1 (3-class clarity), ranking 8th out of 33 teams on Subtask 2 and 13th out of 41 teams on Subtask 1 on the official leaderboard. Furthermore, our ablation studies reveal key insights into effective prompt design for evasion detection: presenting labels within a hierarchical taxonomy helps structure model reasoning, while few-shot exemplars provide task calibration. However, the strongest prompt variants are not statistically distinguishable in Macro F1, and explicitly enabling extended reasoning modes yields substantial performance gains by facilitating the multi-step pragmatic analysis required to detect evasive intent.

阅读与讨论 → 访问原文 →

21.

arXiv (CS.AI) 2026-06-11 DOI: arXiv:2504.09762

Position: Stop Anthropomorphizing Intermediate Tokens as Reasoning/Thinking Traces!

作者:

Subbarao Kambhampati↗Karthik Valmeekam↗Siddhant Bhambri↗Vardhan Palod↗Lucas Saldyt↗Kaya Stechly↗Soumya Rani Samineni↗Durgesh Kalwar↗Upasana Biswas↗

arXiv:2504.09762v4 Announce Type: replace Abstract: Intermediate token generation (ITG), where a model produces output before the solution, has become a standard method to improve the performance of language models on reasoning tasks. These intermediate tokens have been called \say{reasoning traces} or even \say{thinking traces} – implicitly anthropomorphizing the traces, and implying that these traces resemble steps a human might take when solving a challenging problem, and as such can provide an interpretable window into the operation of the model's thinking process to the end user. In this position paper, we present evidence that this anthropomorphization isn't a harmless metaphor, and instead is quite dangerous – it confuses the nature of these models and how to use them effectively, and leads to questionable research. We call on the community to avoid such anthropomorphization of intermediate tokens.

阅读与讨论 → 访问原文 →

22.

arXiv (CS.LG) 2026-06-19 DOI: arXiv:2606.20431

Sparsity, Superposition, and Forgetting: A Mechanistic Study of Representation Retention in Continual Learning

作者:

Jan Wasilewski↗J\k{e}drzej Kozal↗Micha{\l} Wo\'zniak↗Bartosz Krawczyk↗

arXiv:2606.20431v1 Announce Type: new Abstract: Continual learning (CL) systems often forget previously acquired knowledge, yet the mechanisms driving forgetting remain hard to isolate in practice because real datasets entangle many factors. We present a controlled, toy-world framework that makes these mechanisms observable and testable. Using a synthetic generator-separator pipeline, we define ground-truth latent features, build tasks with tunable sparsity and overlap, and introduce measurable quantities for representation strength and superposition (directional overlap among features). We then study retention dynamics-the temporal change of representation strength by fitting sparse dynamical relations (via SINDy) between retention, superposition, and exposure history. A complementary task-level analysis based on effective rank characterizes how representational capacity is allocated across tasks. Our controlled experiments yield three takeaways. (1) Superposition tends to increase over time with transient dips at task boundaries, suggesting boundary-specific interference rather than steady drift. (2) Higher feature sparsity induces more superposition yet does not inevitably cause forgetting; when representations remain strong, forgetting can be reduced despite overlap. (3) Task-level effective rank grows with sparsity, indicating broader capacity usage under sparse regimes. Together, these results nuance the common intuition that more superposition leads to more forgetting by showing that overlap interacts with representation strength and capacity allocation. Our toy analysis provides falsifiable hypotheses and diagnostic tools for CL.

阅读与讨论 → 访问原文 →

23.

arXiv (CS.CL) 2026-06-16 DOI: arXiv:2508.01401

MedSynth: Realistic, Synthetic Medical Dialogue-Note Pairs

作者:

Ahmad Rezaie Mianroodi↗Amirali Rezaie↗Niko Grisel Todorov↗Nadine A. Friedrich↗Maria P Mogollon↗Alexander Hernandez-Tirado↗Guillermo Lopez Garcia↗Cyril Rakovski↗Frank Rudzicz↗

Physicians spend significant time documenting clinical encounters, a burden that contributes to professional burnout. To address this, robust automation tools for medical documentation are crucial. We introduce MedSynth – a novel dataset of synthetic medical dialogues and notes designed to advance the Dialogue-to-Note (Dial-2-Note) and Note-to-Dialogue (Note-2-Dial) tasks. Informed by an extensive analysis of disease distributions, this dataset includes over 10,000 dialogue-note pairs covering over 2000 ICD-10 codes. We demonstrate that our dataset markedly enhances the performance of models in generating medical notes from dialogues, and dialogues from medical notes. The dataset provides a valuable resource in a field where open-access, privacy-compliant, and diverse training data are scarce. Code is available at https://github.com/ahmadrezarm/MedSynth/tree/main and the dataset is available at https://huggingface.co/datasets/Ahmad0067/MedSynth.

阅读与讨论 → 访问原文 →

24.

arXiv (CS.CL) 2026-06-16 DOI: arXiv:2503.23688

Mapping Geopolitical Bias in 11 Large Language Models: A Bilingual, Dual-Framing Analysis of U.S.-China Tensions

作者:

William Guey↗Wei Zhang↗Pierrick Bougault↗Vitor D. de Moura↗Jos\'e O. Gomes↗

Large language models are how hundreds of millions of people now encounter contested political questions, raising a subtle measurement problem: a model that simply agrees with whatever it is told can masquerade as biased, contaminating any claim that models hold political opinions. We address this by importing balanced keying from survey psychometrics, posing each proposition and its swapped reverse and signing the response so acquiescence cancels and genuine conviction accumulates. The result is a reproducible, quantitative instrument that maps geopolitical stance across 11 models and 2 languages (19,712 responses). Developer origin, query language and issue domain emerge as three near-equal, additive factors; every model, including those built in the United States, leans more Pro-China in Mandarin; and two models with identical agreement bias are told apart, one neutral, one biased. We release it as an open, interactive tool that extends to any contested-opinion domain.

阅读与讨论 → 访问原文 →

25.

arXiv (CS.AI) 2026-06-19 DOI: arXiv:2606.19868

A Systematic Evaluation of Black-Box Uncertainty Estimation Methods for Large Language Models

作者:

Jiayi Wang↗Xu-Yao Zhang↗

arXiv:2606.19868v1 Announce Type: new Abstract: Although large language models (LLMs) have shown strong capabilities across a wide range of tasks, their outputs often remain unreliable and may contain hallucinations, making uncertainty estimation (UE) essential for building trustworthy LLMs. In practice, many mainstream LLMs are only accessible through restricted APIs, where internal signals such as logits and hidden states are unavailable, making black-box UE especially important. However, existing work on black-box UE for LLMs remains fragmented in methodology and lacks a unified empirical comparison. To address this gap, we present a systematic review of black-box UE methods and organize them into five categories: verbalization-based, sampling-based, explanation-based, multi-agent, and hybrid methods. We further build a unified evaluation framework and benchmark 24 representative methods across 4 models and 4 dataset settings. Our results show that no single method consistently dominates across all settings. Nevertheless, methods that reason over and compare candidates in the answer space are generally effective, and hybrid methods that combine multiple uncertainty signals perform well under most conditions. By releasing the benchmark data and a unified evaluation framework, we aim to facilitate reproducible comparisons and support future research, while our empirical findings provide practical guidance for developing future black-box UE methods for LLMs.

阅读与讨论 → 访问原文 →