Academic Intelligence · Curated Daily

探索全球前沿学术脉络

AcademicHub 汇聚顶级期刊与预印本平台的实时文献。定制您的专属科研雷达,利用大语言模型自动生成交叉领域文献分析简报。

01.
arXiv (CS.CV) 2026-06-12

MPMWorlds: Material-Point-Method Simulations for Inferring and Extrapolating Physical Dynamics

To study the ability to infer physical dynamics from videos and extrapolate them forward in time, we assemble a dataset of 2D Material Point Method (MPM) physical simulations covering rich physical phenomena such as deformable objects, fluids, kinetic objects, and emitters. We study code generation and video diffusion approaches on this dataset, identifying their strengths and weaknesses by varying the amount of physically relevant side information. The code generation model, beyond giving a working demonstration of automatic synthesis of MPM simulations, reveals that such an approach struggles with inferring physical parameters from visual input, but relative to video diffusion, produces physically and temporally stable extrapolations forward in time, while the video diffusion model more strongly identifies geometric properties from visual input but produces physically implausible extrapolations.

02.
arXiv (CS.AI) 2026-06-15

GAGPO: Generalized Advantage Grouped Policy Optimization

arXiv:2605.13217v1 Announce Type: cross Abstract: Reinforcement learning has become a powerful paradigm for post-training large language model agents, yet credit assignment in multi-turn environments remains a challenge. Agents often receive sparse, trajectory-level rewards only at the end of an episode, making it difficult to determine which intermediate actions contributed to success or failure. As a result, propagating delayed outcomes back to individual decision steps without relying on costly auxiliary value models remains an open problem. We propose Generalized Advantage Grouped Policy Optimization (GAGPO), a critic-free reinforcement learning method for precise, step-aligned temporal credit assignment. GAGPO constructs a non-parametric grouped value proxy from sampled rollouts and uses it to compute TD/GAE-style temporal advantages, recursively propagating outcome supervision backward through time. Combined with group-wise advantage normalization and an action-level importance ratio, GAGPO extracts stable, localized optimization signals directly from multi-turn trajectories. Experiments on ALFWorld and WebShop show that GAGPO outperforms strong reinforcement learning baselines. Further analyses demonstrate faster early-stage learning, improved interaction efficiency, and smoother optimization dynamics, suggesting that GAGPO offers a simple yet effective framework for multi-turn agentic reinforcement learning.

03.
arXiv (CS.AI) 2026-06-16

Multi-agent Framework for Time-Sensitive Complementary Collaboration in Minecraft

arXiv:2606.15684v1 Announce Type: new Abstract: We present TickingCollabBench, a Minecraft-based multi-agent benchmark for a novel class of time-sensitive complementary collaboration tasks. Our benchmark reflects four core characteristics of real-world collaboration: agent heterogeneity, mandatory collaboration, dynamic environments, and strict real-time constraints with failure risks. To enable this, we develop the TickingCollab framework, which supports the generation of diverse dynamic environments and abstracts Minecraft's primitive APIs to enable declarative YAML task specifications for composing these events. Building on this, we design a feasibility-aware automated benchmark generation pipeline, where an LLM drafts structurally diverse task configurations and feasibility verifier filters out invalid ones using approximate constraints. Evaluations demonstrate that lang latency and inherent difficulty of coordinating under partial observability and agent heterogeneity cause LLMs to frequently fail under dynamic environments and fall significantly short of a global-knowledge oracle.

04.
arXiv (quant-ph) 2026-06-24

On the Limits of Stretching Quantum Pseudorandomness

arXiv:2606.24736v1 Announce Type: new Abstract: Pseudorandom states, introduced by Ji, Liu, and Song (CRYPTO '18), are quantum analogues of classical pseudorandom generators. A fundamental property of classical pseudorandom generators is that their output can be stretched to arbitrary polynomial length. Whether an analogous stretching property holds for quantum pseudorandom states remains unclear. In this work, we prove the first black-box separation between single-copy secure pseudorandom states ($\mathsf{1PRS}$) with different output lengths. Specifically, we construct a quantum oracle relative to which $\mathsf{1PRS}$ with output length $m(n)=1.1n$ exist, but $\mathsf{1PRS}$ with output length $m(n)=\Omega(n^{2+\epsilon})$ do not, for any $\epsilon>0$. Our proof leverages the Common Haar Random State (CHRS) model introduced by Chen, Coladangelo, and Sattath (EUROCRYPT '25), and introduces a technique to bound the effective number of resource CHRS states utilized by any $\mathsf{1PRS}$ generator in this model.

06.
arXiv (CS.CL) 2026-06-19

Toward Human-Centered AI-Assisted Terminology Work

Generative AI is likely to transform terminology work by creating new opportunities for automation. At the same time, it raises concerns about the future of terminologists and terminological resources, as efficiency pressures may encourage excessive automation based on the perception that human expertise can be replaced by AI. However, large language models remain unreliable for terminological purposes due to errors, hallucinations, and various forms of bias, making terminologists indispensable for ensuring the accuracy and reliability of terminological data. This paper argues that human-centered AI, an approach that emphasizes that AI's primary goal should be to contribute to human well-being, provides a framework for maximizing the benefits of generative AI while mitigating its risks. It contends that high levels of automation and meaningful human control are compatible and desirable, and that AI should enhance terminologists' capabilities while preserving their agency and decision-making authority. The implications of AI-assisted terminology work are examined through three interrelated dimensions: the augmented terminologist, ethical AI, and human-centered design. In particular, the paper examines how AI integration reshapes the role of the terminologist, affects professional values and working conditions, requires the management of AI-generated bias, and calls for the design of AI tools around the terminologist's needs. The paper concludes that a human-centered orientation is necessary to ensure that AI strengthens, rather than undermines, the essential role of terminology work in supporting specialized communication and the accurate transmission of knowledge across languages and cultures.

07.
bioRxiv (Bioinfo) 2026-06-23

CellOS: Learning a World Model of Cellular State through Joint Embedding Prediction

Foundation models learned from single-cell transcriptomes are central to the prospect of AI virtual cell that can represent, query and predict cellular state. However, most current single-cell foundation models learn from a single view of gene expression and are optimized primarily through reconstruction or next-token prediction. As a result, they capture expression abundance but can-not explicitly reconcile complementary views of cellular state. Here we present CellOS, a multi-view foundation model that learns cellular representations from paired expression and perception views. CellOS integrates complementary views through a scalable three-stage training strategy that combines causal cell-sentence language modelling, function-preserving dense-to-mixture-of-experts expansion and latent-space alignment via an LLM-JEPA objective. Using this framework, we trained a 12-billion-parameter model on 390.5 million single-cell transcriptomes. Across diverse benchmarks spanning cell-state annotation, batch integration and perturbation-response prediction, CellOS consistently outperformed state-of-the-art single-cell foundation models in cell-state annotation and perturbation-response prediction while preserving robust batch integration. Together, these results suggest that predictive alignment between complementary cellular views provides a scalable path toward representation-centric cellular world models and transferable AI virtual cells.

09.
arXiv (CS.CV) 2026-06-11

CountZES: Counting via Zero-Shot Exemplar Selection

Object counting in complex scenes is particularly challenging in the zero-shot (ZS) setting, where instances of unseen categories are counted using only a class name. Existing ZS counting methods that infer exemplars from text often rely on off-the-shelf open-vocabulary detectors (OVDs), which in dense scenes suffer from semantic noise, appearance variability, and multi-instance proposals. Alternatively, random image-patch sampling is employed, which fails to accurately delineate object instances. Since counting is sensitive to exemplar quality, such selection strategies often yield poorly representative exemplars, leading to inaccurate count estimation. To address these issues, we propose CountZES, an inference-only approach for object counting via ZS exemplar selection. CountZES discovers diverse exemplars through three synergistic stages: Detection-Anchored Exemplar (DAE), Density-Guided Exemplar (DGE), and Feature-Consensus Exemplar (FCE). DAE refines OVD detections to isolate precise single-instance exemplars. DGE introduces a density-driven, self-supervised paradigm to identify statistically consistent and semantically compact exemplars, while FCE reinforces visual coherence through feature-space clustering. Together, these stages yield a complementary exemplar set that balances textual grounding, count consistency, and feature representativeness. Experiments on diverse datasets demonstrate CountZES superior performance among ZOC methods while generalizing effectively across domains.

10.
arXiv (CS.CL) 2026-06-17

DICE: Diffusion Large Language Models Excel at Generating CUDA Kernels

Diffusion large language models (dLLMs) have emerged as a compelling alternative to autoregressive (AR) LLMs, owing to their capacity for parallel token generation. This paradigm is particularly well-suited for code generation, where holistic structural planning and non-sequential refinement are critical. Despite this potential, tailoring dLLMs for CUDA kernel generation remains challenging, obstructed not only by the high specialization but also by the severe lack of high-quality training data. To address these challenges, we construct CuKe, an augmented supervised fine-tuning dataset optimized for high-performance CUDA kernels. On top of it, we propose a bi-phase curated reinforcement learning (BiC-RL) framework consisting of a CUDA kernel infilling stage and an end-to-end CUDA kernel generation stage. Leveraging this training framework, we introduce DICE, a series of diffusion large language models designed for CUDA kernel generation, spanning three parameter scales, 1.7B, 4B, and 8B. Extensive experiments on KernelBench demonstrate that DICE significantly outperforms both autoregressive and diffusion LLMs of comparable scale, establishing a new state-of-the-art for CUDA kernel generation.

11.
arXiv (quant-ph) 2026-06-16

Certified Finite-Shot Operating Windows for Virtual Distillation and Symmetry Verification

arXiv:2606.15464v1 Announce Type: new Abstract: Quantum error mitigation methods are usually compared through their infinite-shot bias, but on real devices the comparison is decided by finite sampling budgets, estimator instabilities, and per-shot resource costs. We develop a finite-shot operating-window theory that makes this comparison certifiable for virtual distillation (VD) and symmetry verification (SV): for each method we derive a mean-squared-error law with explicit, non-asymptotic remainder constants. For VD, the law captures the statistical bias and denominator instability of its quotient estimator, with a concentration certificate locating the sample size beyond which the quotient is trustworthy; for SV, it isolates the bias floor left by undetectable errors and the sampling penalty set by the acceptance probability. A selection trichotomy classifies any two-method comparison into a tie, uniform dominance, or a genuine tradeoff with a certified crossing window, including a self-consistency test that rejects spurious crossings. The theory makes falsifiable predictions – operating-window locations scaling as $p^{-2}$ or $p^{-1}$ in the noise rate, and the sign pattern of all pairwise comparisons – which exact white-box experiments confirm with fitted exponent $-1.97$ against the predicted $-2$ and with $300/300$ sign agreement, within a pre-registered analysis whose single failed gate, an over-strict all-instance criterion, is reported and audited in full. Gate-level simulation and archived runs on two IBM backends then test the windows under device conditions: idealized VD windows exist, but realistic interferometry overhead and denominator instability erase them, and calibrated SV is the practical winner in the tested QAOA instances. This absence of a universal winner is not a failure of mitigation; it is the regime structure that certified operating windows predict.

12.
arXiv (CS.AI) 2026-06-18

Equivariant Graph Neural Networks Improve Optical Spectra Prediction for Materials Screening

arXiv:2606.19133v1 Announce Type: cross Abstract: Scalable prediction of optical spectra is a critical component of high-throughput materials screening for optoelectronic applications such as solar cells. Existing surrogate models are trained on spectra computed from lower levels of theory or rely on rotation-invariant scalar features, limiting their geometric expressiveness. We explore the use of equivariant graph neural networks for optical spectra prediction, adapting GotenNet to this task and evaluating it on multiple datasets including a recently published collection of 10,533 structures with spectra computed at the level of the random phase approximation (RPA). The proposed model outperforms the current state of the art, with the largest gains in the 0-8 eV range and on predicting the static real permittivity, both of particular relevance for thin-film optics.

13.
Nature (Science) 2026-06-11

Daily briefing: Deep-sea whale graveyard is a treasure trove of fossils

作者:

Researchers have uncovered more than 400 fossilized whale bones in an ocean-floor chasm. Plus, the working lives of scientists, in pictures, and how AI could slow the pace of research publication for the better. Researchers have uncovered more than 400 fossilized whale bones in an ocean-floor chasm. Plus, the working lives of scientists, in pictures, and how AI could slow the pace of research publication for the better.

14.
medRxiv (Medicine) 2026-06-15

Wellbeing After Stroke-2 (WAterS-2): a feasibility study with process evaluation exploring inclusive, accessible, online psychological support after stroke

Objectives: Explore feasibility and acceptability of upskilling a workforce to deliver a co-developed intervention, based on Acceptance and Commitment Therapy (ACT), to support psychological adjustment post-stroke targeting underserved groups. Design: Multi-site, single-arm feasibility study with embedded mixed-methods process evaluation (ISRCTN17628580). Setting: Four NHS community stroke services across England. Participants: 1. Stroke survivors [≥]18 years of age, [≥]4 months post-stroke, reporting psychological difficulties adjusting to stroke, able to consent and access remote group sessions in English; 2. Group facilitators from NHS stroke services, not ACT specialists. Intervention: WAterS-2: an eight-session, remotely-delivered ACT-informed group intervention. Outcome measures: Recruitment, fidelity, safety, acceptability and perceived value were assessed using fidelity checklists, post-intervention surveys and semi-structured interviews with stroke survivors and facilitators. Clinical outcomes including mood (HADS), wellbeing (ONS4), psychological flexibility (AAQ-ABI), measured post-group and three-months later. Results: Nineteen stroke survivors recruited (mean 9.6 months post-stroke; n=5 (26%) minoritised ethnicities; n=10 (52%) with aphasia). Thirteen facilitators - including two peer support workers - delivered the intervention with fidelity following structured training across four services. Drop-out was low (2/19; 11%); with 15 (79%) attending [≥]5/8 sessions. Remote data collection was feasible (79% follow-up completion), with no adverse events recorded. Acceptability was high: survivors valued peer connection, grounding and mindfulness practices. ACT metaphors were helpful for some but challenging for others, including some with aphasia. Online delivery was suitable but limited informal connection. Facilitators reported increased capability, incorporating ACT skills into routine care. NHS workforce pressures and geographically-constrained referral pathways limited recruitment reach. Conclusions: WAterS-2 is feasible, safe, acceptable and inclusive. A mixed workforce, including NHS peer support workers, can be upskilled to deliver with fidelity. Inclusion of underserved groups is achievable but requires active strategies beyond standard NHS referral routes. Findings inform a provisional logic model and a future pragmatic trial.

15.
arXiv (quant-ph) 2026-06-24

Infinite-Level Hierarchy of Solvable Quantum Circuits

arXiv:2606.23803v1 Announce Type: new Abstract: Dual-unitary circuits have emerged as a paradigm of exactly solvable yet non-integrable quantum dynamics. Recently, a generalization of dual unitarity attempting to extend the phenomenology of exactly solvable circuits has been introduced through a hierarchy of conditions, with dual unitarity as the first level. However, beyond the second level the proposed generalized dual-unitary hierarchy ceases to be solvable in the whole spacetime. We present an infinite hierarchy of solvability conditions remedying this problem. These new conditions can be combined with the generalized dual-unitary hierarchy to obtain circuits for which correlation functions and entanglement dynamics can be analyzed exactly in the whole spacetime. We show that this novel hierarchy possesses non-trivial solutions at every level. Our results demonstrate that dual unitarity can be systematically extended while preserving solvability, opening up investigations of exactly solvable non-integrable systems with more general properties.

16.
arXiv (CS.CL) 2026-06-19

Leverage Is Not Reach: A Control-Window Law for Single-Neuron Steering in Language Models

Aligned language models gate behaviors such as refusal and language routing through sparse feed forward neurons, yet no theory predicts when a single neuron intervention controls a behavior coherently rather than collapsing the output. We develop a budget normalized control window framework for single neuron steering. A dose along one write direction reduces to one control coordinate: the alignment between the residual stream and the write, driven along a universal saturation curve in units of a coherence budget set by the residual norm divided by the write norm. Coherent control exists when a behavior trigger lies below the collapse ceiling. The same coordinate governs benign mode switches and refusal; the ceiling follows from weights and one generic forward pass, while triggers are measured at rollout. On fifteen held out neurons, the predicted ceiling has mean absolute error 0.14, about 0.07 in bulk layers, and the committed open or closed verdict holds on eleven against a ten of fifteen majority baseline. Closed cases expose three failure modes rather than violations: collapse before trigger, too little depth to propagate, or a normalization that caps how far one neuron can push. The law explains why local gradient attribution anti predicts control: true controllers write off the readout axis and carry a near zero first order gradient. A forward only contrastive screen made precise by the window recovers controllers that attribution misses. On refusal, the hardest case, intervention success is typed, not scalar: coherent bypass and strict actionable reach separate, so a neuron can flip refusal in fluent, on task text with no actionable content, and genuine actionable reach appears only for three of six audited Llama pivots and only at later rollout horizons. Single neuron steering is therefore a budgeted, typed audit of controllability rather than a fixed dose anecdote.

17.
arXiv (CS.CL) 2026-06-24

To Compare, or Not to Compare: On Methodological Practices in Evaluating Social Bias

As Large Language Models are increasingly deployed in critical applications, robustly evaluating their social biases is paramount. However, the current literature suffers from widespread methodological fragmentation, which yields contradictory conclusions. This stems largely from ignoring the structural framing of benchmark-level evaluations. To resolve this, we introduce a unified and controllable framework that standardizes heterogeneous benchmarks to systematically contrast isolated demographic assessments with forced-choice comparative settings. Crucially, this allows us to disentangle the confounding effects of Chain-of-Thought reasoning, neutral fallback options, and other structural artifacts in social bias evaluations. Our evaluation across multiple model families reveals a massive, systematic paradigm gap: while isolated assessments limit prejudice activation, comparative settings act as aggressive catalysts for latent discrimination, a shift primarily driven by underspecified contexts. Alarmingly, CoT reasoning exacerbates social biases under comparative settings, and this systemic bias persists as a deterministic prejudice even when models are provided neutral fallback options or claim to answer randomly. Finally, we demonstrate that this comparative prejudice is a generalized phenomenon that scales positively with model size. Ultimately, we offer a crucial methodological guideline: while researchers must leverage comparative settings to robustly audit hidden biases, practitioners cannot safely rely on comparative deployments in ambiguous real-world tasks.

18.
arXiv (CS.AI) 2026-06-16

Artificial Intelligence Index Report 2026

arXiv:2606.15708v1 Announce Type: new Abstract: Welcome to the ninth edition of the AI Index report. As AI continues to advance rapidly, the question becomes whether the systems built around it can keep up. Governance frameworks, evaluation methods, education systems, and the data infrastructure needed to track AI's impact are struggling to match the pace of the technology itself. That gap between what AI can do and how prepared we are to manage it runs through every chapter of this year's report. New in this edition, the report tracks how AI is being tested more ambitiously across reasoning, safety, and real-world task execution, and why those measurements are increasingly difficult to rely on. It also features new estimates of generative AI's economic value alongside emerging evidence of its labor market effects, an analytical framework on AI sovereignty, and a science chapter developed in collaboration with Schmidt Sciences. For the first time, the report features standalone chapters on AI in science and AI in medicine, reflecting AI's growing impact across these two domains.

19.
arXiv (CS.LG) 2026-06-17

BioArtlas: Computational Clustering of Multi-Dimensional Complexity in Bioart

arXiv:2511.19162v3 Announce Type: replace-cross Abstract: Bioart brings living material into artistic practice, where a single work can be at once an aesthetic object, a scientific instrument, and an ethical provocation. Traditional categories sort such works along one axis at a time, which flattens the very hybridity that defines the field and leaves curators no way to compare works across many dimensions together. I introduce BioArtlas, a computational atlas that represents each bioartwork along many curated dimensions at once and organizes the field by conceptual similarity rather than by medium or chronology. My method embeds the keywords of all 81 works on each of thirteen interpretive axes, groups related concepts into a shared codebook that tames inconsistent terminology, and then searches systematically for a clustering that is both statistically clean and interpretable. Among the methods that place every work on the map, agglomerative clustering separates the field far more cleanly than the usual k-means baseline (silhouette 0.664 versus 0.483), whereas density-based methods reach higher scores only by discarding most of the corpus as noise. By separating rigorous analysis from public storytelling, BioArtlas turns the tangled complexity of bioart into a navigable landscape, openly available as an interactive interface (https://www.bioartlas.com) and dataset (https://github.com/joonhyungbae/BioArtlas).

20.
arXiv (CS.LG) 2026-06-18

A Cross-Model VLM-Judge Protocol for Single-Image 3D Mesh Quality (and Why Cheap Proxies Fall Short)

arXiv:2606.18451v1 Announce Type: new Abstract: Single-image-to-3D generators are improving quickly, but there is no agreed, human-free way to tell whether one generated mesh is better than another. Practitioners commonly rely on cheap automatic proxies (render-space CLIP similarity and mesh geometry-validity statistics), yet how well these track perceived quality is unestablished. We make two contributions. First, we propose and validate a reproducible VLM-judge evaluation protocol: a fixed 24-view headless render rig, two independent vision-language judge families, and a mandatory position-bias correction that queries both presentation orders and keeps only order-consistent verdicts. The two judge families agree substantially with each other (Cohen's kappa = 0.66), well above the chance-agreement floor. Second, using this protocol as the reference, we show the cheap proxies do not substitute for it. Geometry validity is only a weak signal on average (because, as we show, it is bimodal) and stays below our pre-registered target, while render-CLIP is at chance. A learned Bradley-Terry head collapses onto a single manifoldness statistic (giving render-CLIP a negative weight) and matches geometry-only exactly, so learning the feature weights buys nothing. The proxy is also bimodal: it is significantly above chance on contrasts with visible geometric defects but at chance on ambiguous contrasts, consistent with geometry validity tracking the judge only when the defect is visually salient. We therefore recommend the VLM-judge protocol as a reliable, reproducible evaluator under the conditions tested (two feed-forward generators on Google Scanned Objects, with a face-drop degradation regime) and advise against geometry/CLIP proxies as optimization targets.

21.
arXiv (CS.CV) 2026-06-11

Detecting AI-Generated Content on Social Media with Multi-modal Language Models

Generative AI has enabled the creation of photorealistic images and videos that are increasingly disseminated on social media, often used for spam, misinformation, manipulation, and fraud. Existing AI-generated content (AIGC) detection methods face challenges including poor generalization to new generation models, reliance on single modalities, and lack of interpretable explanations. We present our pipeline that mitigates these issues by continuously curating diverse multi-modal social media data and training a compact vision-language model for detection and explanation. Our model achieves state-of-the-art detection performance on public benchmarks and demonstrates robust detection and explanation capabilities on internal social media datasets across multiple platforms. We deployed our model for post recommendation on social media platforms and observed positive downstream impacts on user engagement, demonstrating that it is feasible to perform effective AIGC detection in dynamic, real-world social media environments.

22.
arXiv (CS.AI) 2026-06-18

User as Engram: Internalizing Per-User Memory as Local Parametric Edits

作者:

arXiv:2606.19172v1 Announce Type: new Abstract: Personal memory in a language model is two problems: content and reasoning skill. The brain keeps the two apart (a sparse, local engram in the hippocampus for each episode, a slow neocortex for the shared skills that interpret it), so a new fact need not overwrite everything else. Most personalization today keeps a user's facts outside the weights, in a natural-language memory file or a retrieval index. When facts are written into the model instead, the standard recipe is the per-user LoRA adapter, which does the opposite of the brain, folding content and skill into one global weight delta. Writing a user's facts as a LoRA contaminates text unrelated to them; writing the same facts as local Engram rows leaves it mathematically untouched, resulting in a roughly 33,000x smaller memory footprint. We therefore propose User as Engram: store a user's content as surgical edits to the hash-keyed memory table of an Engram model, and carry the reasoning skill in one shared adapter. This layered design matches per-user LoRA's direct recall while delivering 5.6x higher indirect-reasoning accuracy on average, and never makes a single user worse at reasoning than the untouched base. The edit is a glass box: writing a fact switches on its lookup at exactly the trigger, adds the value the answer needs, leaves every other position unchanged to the last bit, and fails if written into the wrong layer. Because different users' facts land in disjoint hash slots, their edits compose: many users live in one shared table at once, stacking additively and losslessly, where a per-user LoRA, a single global weight delta, admits only one. Upon retrieval, a per-user Engram table does not grow with the population the retriever must search, so past ~100 facts it overtakes a retrieval pipeline on a 2.5x larger model.

23.
arXiv (CS.CL) 2026-06-15

The Culture Funnel: You Can't Align What isn't in the Data

Current cultural alignment approaches focus on inference-time interventions, assuming models already contain sufficient cultural knowledge. We argue modern LLM pipelines suffer from a cultural data funnel. Using a multidimensional tagging framework across pretraining, fine-tuning, alignment, and reasoning datasets, we show explicit cultural signals decline sharply during post-training, while geographically concentrated, task-specialized data dominates. Multilinguality enhances geographic diversity of cultural knowledge but does not ensure balanced representation. Our tags improve downstream cultural benchmark performance, demonstrating that advances require shifting focus in training data pipelines. To facilitate future research, we release our culturally tagged dataset with 5.6M samples at https://huggingface.co/datasets/CohereLabs/CultureMarkers.

24.
arXiv (CS.CL) 2026-06-16

T-Mem: Memory That Anticipates, Not Archives

Long-term memory is essential for conversational agents to remain coherent across extended dialogues, follow through on commitments made many sessions earlier, and adapt their behaviour to each user. Current LLM-backed long-term conversational memory, however, is reachability-bounded by the similarity between a query and stored content, both lexical and dense-vector. The approach is effective when query and memory share surface features such as wording or named entities (we call this descriptive). But it misses another, equally valuable class of cases, where query and memory do not share surface features and are tied only by a latent semantic arc (associative). On this regime prevailing long-term memory systems collectively fail. Covering this other half is what allows an assistant, for the first time, to actively draw on past dialogue as a semantic asset. On the memory side, this is the engineering counterpart of what cognitive science calls episodic future thinking: rehearsing past experience for the future contexts under which it will need to be found. We call these write-time rehearsals triggers. We propose T-Mem, the first long-term conversational memory architecture that covers both descriptive and associative recall. At each of two evidence granularities, single facts and full exchanges, T-Mem instantiates one descriptive trigger family and one associative trigger family, so that every memory remains reachable from both surface-similar and relevance-bound queries. As empirical validation, T-Mem reaches state-of-the-art on both LoCoMo and LoCoMo-Plus.

25.
arXiv (CS.AI) 2026-06-16

Surprise-Guided MergeSort: Budget-Efficient Human-in-the-Loop Ranking via Adaptive Comparison Scheduling

arXiv:2606.15623v1 Announce Type: cross Abstract: Pairwise comparison is the gold standard for subjective ranking tasks; however, exhaustive annotation requires a massive number of human comparisons ($O(n^2)$). While sorting-based methods have reduced this burden to $O(n\log n)$, they still require expensive human judgment for every single comparison. To further improve annotation efficiency, we propose leveraging a Vision-Language Model (VLM) not as an annotator replacement, but as a question prioritizer to identify which comparisons genuinely require human judgment. The proposed Surprise-Guided MergeSort (SGS) framework achieves this through three integrated components: (1) a bottom-up MergeSort scheduler that structures comparisons and exploits transitivity, (2) a composite Surprise Scorer – combining position-bias-cancelled VLM confidence, Elo gap, and vote entropy – to quantify comparison ambiguity, and (3) an adaptive budget allocator that routes high-surprise pairs to humans while automating low-surprise pairs via transitivity inference. Validation was conducted on six diverse benchmarks spanning text similarity (STS-B, BIOSSES, SICKR-STS) and image quality assessment (KonIQ-10k, TID2013, LIVE Challenge). SGS effectively identified and skipped up to 535 non-informative comparisons per session. Consequently, it achieved Kendall's $\tau{\times}100$ improvements of $+6$ to $+12$ over Active Elo under the same total budget. These results demonstrate that combining VLM-guided surprise metrics with algorithmic sorting provides a generally consistent accuracy-efficiency trade-off across diverse domains.