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 (quant-ph) 2026-06-19

On chip, multifunctional quantum sensing using single spins in a van der Waals crystal

arXiv:2606.19978v1 Announce Type: new Abstract: Nanoscale thermometry and magnetometry are in high demand across a wide range of scientific and technological applications. In this context, optically addressable spins in solids have emerged at the forefront of on-chip quantum sensing. However, simultaneous quantum sensing of multiple parameters (e.g., temperature and magnetic field) using the same spin sensor remains challenging due to cross-sensitivity to multiple physical quantities. Here, we demonstrate independent dual sensing of temperature and magnetic field using single quantum emitters in hexagonal boron nitride (hBN). We experimentally verify the independent response of the zero-phonon line (ZPL) position to temperature and of optically detected magnetic resonance (ODMR) to magnetic fields. Furthermore, we demonstrate local temperature sensing of a microcircuit while simultaneously measuring an external magnetic field. Our results establish quantum emitters in hBN as a robust platform for multifunctional quantum sensing under realistic operating conditions.

02.
arXiv (quant-ph) 2026-06-16

The Optimal Rate Function in Covariant Quantum State Tomography

arXiv:2606.16948v1 Announce Type: new Abstract: The problem of quantum tomography is to estimate an unknown quantum state $\rho$ from a measurement of $n$ copies of $\rho$. One can ask which tomography protocol, i.e.\ which choice of multi-copy measurement, gives the best possible estimate of $\rho$. To do so, we characterize tomography protocols by their rate function, which governs the exponential rate at which a protocol assigns probability to a particular estimate $\sigma$ of the true state $\rho$. This rate function is a quantum mechanical generalization of the classical relative entropy between the true state and its estimate, and depends on the choice of protocol. It is bounded by the quantum relative entropy, and we show that this bound is sharp: for any $\rho$ and $\sigma$ we construct a family of protocols whose rate functions converge to the quantum relative entropy $D(\sigma\|\rho)$. We consider the family of covariant tomography protocols; these are the basis independent state estimation schemes that assume no prior information about $\rho$ and $\sigma$. Keyl described a specific tomography protocol based on Schur sampling, and conjectured that among all covariant tomography protocols it has the largest possible rate function for all $\sigma$ and $\rho$. We prove this conjecture. The resulting rate function is an annealed version of quantum relative entropy, due to the cost of learning the eigenbasis in covariant quantum state tomography.

03.
arXiv (quant-ph) 2026-06-17

Creating squeezed and non-classical collective motional many-body states through stroboscopic Rydberg dressing

arXiv:2606.17849v1 Announce Type: cross Abstract: Realizing conditional quantum operations, e.g., quantum gates, for quantum computing and simulation requires controlled interactions between particles. Often, these interactions depend on the interparticle distance, and accordingly, an uncertainty of the relative particle position may translate into gate infidelities. We consider here a quantum computing platform based on an array of neutral atoms and present a method that allows to reduce the uncertainty of all interatomic distances. Our approach exploits the coupling between atomic motion and stroboscopically excited atomic Rydberg states. It allows to collectively squeeze the modes corresponding to interatomic displacements, thereby reducing distance fluctuations down to a fraction of the motional vacuum state. Furthermore, the method permits the creation of non-classical states with substantial Wigner negativity. These correlated states may allow reducing motional decoherence, increasing gate fidelity, and potentially yield a resource for quantum-enhanced metrology.

04.
medRxiv (Medicine) 2026-06-10

Global and local genetic overlap among ME/CFS, irritable bowel syndrome and psychiatric traits: a hypothesis-generating analysis

作者:

Background. Myalgic encephalomyelitis/chronic fatigue syndrome (ME/CFS) and irritable bowel syndrome (IBS) frequently co-occur following infection, yet shared genetic architecture at the locus level has not been systematically characterised. Aims. To estimate global and local genetic correlations between ME/CFS (including infection-onset subgroup), IBS, major depressive disorder (MDD) and loneliness/isolation, and characterise ME/CFS cell-type heritability enrichment. Method. GWAS summary statistics: DecodeME (15,579 ME/CFS; 9,738 infection-onset), FinnGen R9 (9,296 IBS), PGC MDD Wave 2 (45,396) and UK Biobank loneliness (N=455,364). LDSC for global correlations; LAVA for local correlations across 2,495 loci; MAGMA for cell-type enrichment (Descartes Human atlas); coloc.abf for colocalisation. Results. All pairwise global correlations were significant after Bonferroni correction, including ME/CFS-all-MDD (rg=0.598, 95% CI 0.46-0.74) and ME/CFS-all-IBS (rg=0.573, 0.39-0.75). Of 4,232 local tests, 16 reached FDR

05.
medRxiv (Medicine) 2026-06-22

Use of the Pharmacy First service in England in the first 12 months: geographic variation and health system context

Objectives: The Pharmacy First (PF) service was introduced across England from 31 January 2024 to expand the clinical role of community pharmacies and improve access to primary care. This paper describes use of PF in its first 12 months, in terms of uptake, access routes, consultation outcomes, geographic variations, service costs and antimicrobial supply. Methods: A descriptive analysis of all PF consultations submitted for payment to NHS Business Services Authority in England between 31 January 2024 and 31 January 2025. Pharmacy-level consultation data were linked to national data on population, location and pharmacy characteristics. PF use was examined using population-standardised consultation rates and consultations per pharmacy. Results: During the first year of implementation, 2,205,731 PF consultations were recorded as delivered across 11,349 pharmacies, with payment of GBP123 million to pharmacies. Uptake increased steadily over time. Most consultations were for acute sore throat (33%) and uncomplicated urinary tract infection (27%), with corresponding antibiotics, phenoxymethylpenicillin and nitrofurantoin being the most supplied. Most people self-referred (74%) into the service, with 95% of consultations managed without onward referral. Substantial geographic variation was observed. Northern regions had higher use based on the eligible population. The South East and Midlands had higher activity per pharmacy. London showed a distinct pattern, with higher self-referral into the service, lower medication supply and higher referral to other healthcare services. Higher consultation volume was weakly associated with pharmacy characteristics, including opening hours, pharmacy type and retail setting, and local context, in terms of socio-economic and geographic factors. Conclusions: PF had immediate uptake and is operating primarily as a direct-access model for common acute conditions. Findings suggest that PF is contributing to improved access to care and may shift demand away from general practice. However, the service uptake appears to be shaped by geographic location, proximity to other healthcare services and pharmacy characteristics.

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

ROSE: Benchmarking the Perception-to-Action Gap in Multimodal Models

arXiv:2606.19965v1 Announce Type: cross Abstract: Multimodal large language models (MLLMs) are increasingly expected to act on visual information, yet the same scene may require different actions under different task contexts. How reliably can a model turn the same visual evidence into the action required by the current context? To answer this question, we introduce \textsc{ROSE} (Reference-conditioned Oddity and Symbolic Execution), a controlled benchmark that holds the visual scene fixed while varying region constraints and required symbolic outputs. Through coupled counting and coordinate-action tasks, \textsc{ROSE} tests whether models can infer an implicit majority reference and act on the resulting fine-grained visual evidence under changing contexts. Across nine recent MLLMs, performance drops by as much as 44.5 percentage points from counting-oriented tasks to region-conditioned action, despite 98.8\% human performance. The gap persists on paired scenes and regions for which the same model returns the correct count, while global-click and matched local controls show that coordinate grounding explains only part of the loss, revealing a distinct, model-dependent bottleneck in turning shared visual evidence into context-specific actions.

07.
arXiv (math.PR) 2026-06-15

On a stochastic phase-field model of cell motility with singular diffusion

arXiv:2601.05881v2 Announce Type: replace Abstract: We study existence of solutions in the variational sense for a class of stochastic phase-field models describing moving boundary problems. The models consist of stochastic reaction-diffusion equations with singular diffusion forced by a phase-field. We investigate both the case of an independently evolving phase-field and of coupled phase-field evolution driven by a viscous Hamilton-Jacobi equation. Such systems are used in the modelling of single-cell chemotaxis, where the contour of the cell shape corresponds to a level set of the phase-field. The technical challenge lies in the singularities at zero level sets of the phase-field. For large classes of initial data, we establish global existence of probabilistically weak solutions in $L^2$-spaces with weights which compensate for the singularities.

08.
arXiv (CS.AI) 2026-06-15

Hybrid Open-Ended Tri-Evolution Makes Better Deep Researcher

arXiv:2606.13710v1 Announce Type: new Abstract: Deep research and agent evolution serve as de-facto tasks for AI agents in real-world applications toward artificial general intelligence. The former enables autonomous retrieval and integration of information in open-ended environments to tackle open-ended research tasks, yet it is constrained by the static parametric deep research capabilities of agent systems. The latter allows agents to autonomously interact with the environment to gain experiences that evolve model capabilities. However, its effectiveness has been widely validated only on verifiable tasks with standard answers, leaving a gap with open-ended research tasks. To bridge these two critical tasks, we propose the Hybrid Open-Ended Tri-Evolution (HOTE) framework, which leverages hybrid-mode reinforcement learning to facilitate the collaborative evolution of a proposer, solver and judge based on web-scale knowledge, moving toward autonomous evolving agents in open-ended tasks and environments. Extensive experiments on three long-form deep research benchmarks demonstrate that the 8B model trained via HOTE surpasses the strongest static open 8-32B models as well as those trained by state-of-the-art deep research training methods with less time overhead, and further verify that the evolution of all three modules in HOTE is indispensable.

09.
arXiv (CS.LG) 2026-06-18

Decomposing Prediction Mechanisms for In-Context Recall

arXiv:2507.01414v2 Announce Type: replace Abstract: We introduce a new family of toy problems that combine features of linear-regression-style continuous in-context learning (ICL) with discrete associative recall. We pretrain transformer models on sample traces from this toy, specifically symbolically-labeled interleaved state observations from randomly drawn linear deterministic dynamical systems. We study if the transformer models can recall the state of a sequence previously seen in its context when prompted to do so with the corresponding in-context label. Taking a closer look at this task, it becomes clear that the model must perform two functions: (1) identify which system's state should be recalled and apply that system to its last seen state, and (2) continuing to apply the correct system to predict the subsequent states. Training dynamics reveal that the first capability emerges well into a model's training. Surprisingly, the second capability, of continuing the prediction of a resumed sequence, develops much earlier. Via out-of-distribution experiments, and a mechanistic analysis on model weights via edge pruning, we find that next-token prediction for this toy problem involves at least two separate mechanisms. One mechanism uses the discrete symbolic labels to do the associative recall required to predict the start of a resumption of a previously seen sequence. The second mechanism, which is largely agnostic to the discrete symbolic labels, performs a "Bayesian-style" prediction based on the previous token and the context. These two mechanisms have different learning dynamics. To confirm that this multi-mechanism (manifesting as separate phase transitions) phenomenon is not just an artifact of our toy setting, we used OLMo training checkpoints on an ICL translation task to see a similar phenomenon: a decisive gap in the emergence of first-task-token performance vs second-task-token performance.

10.
arXiv (CS.CV) 2026-06-12

Masked and Predictive Self-Supervised Foundation Models for 3D Brain MRI

Self-supervised foundation models have shown strong promise in medical imaging. However, existing MRI foundation-model studies have primarily emphasized segmentation and dense prediction tasks, while systematic investigation of self-supervised foundation models for MRI-based disease detection remains limited. In this work, we investigate two major self-supervised pretraining paradigms for MRI-based disease detection: reconstruction-based learning via Masked Autoencoders (MAE) and predictive representation learning via Joint Embedding Predictive Architectures (JEPA). We study the role of auxiliary objectives by introducing a novel spectral-domain reconstruction loss for MAE to enhance sensitivity to fine-grained anatomical structure, and by integrating variance–covariance regularization (VCR) within our JEPA framework to encourage decorrelated latent representations. Our models are pretrained on heterogeneous single-contrast MRI volumes in a contrast-agnostic setting, without modality concatenation. Across five downstream disease detection tasks, our results highlight the importance of self-supervised objective design for medical foundation model pretraining, demonstrating that the downstream benefit of each objective is determined by its relevance to the task's structure. Specifically, spectral regularization yields the largest improvements when the downstream discriminative signal is characterized by strong high-frequency anatomical structures, while covariance regularization is most beneficial when discriminative information spans multiple decorrelated feature dimensions. MAE with spectral-domain supervision consistently achieves superior downstream performance for MRI-based disease detection. These findings suggest that self-supervised objectives in medical imaging encode specific biases, and their downstream benefit is fundamentally conditioned on the task's structure.

11.
arXiv (quant-ph) 2026-06-19

Sparse positive maps on qutrits with exact nondecomposability thresholds and PPT-entanglement transitions

arXiv:2606.19765v1 Announce Type: new Abstract: We study a family of sparse positive maps on qutrits for which positivity, decomposability, and PPT entanglement can all be analysed explicitly. The block structure of the associated Choi matrices reduces positivity to a Hermitian biquadratic form and leads to exact positivity boundaries for three representative parametric families. For the same families we determine the exact transition between decomposable and non-decomposable maps and construct associated PPT states of two classes. The first consists of witness-adapted deformations naturally tied to the non-decomposability analysis. The second consists of analytically tractable families whose full PPT-entangled branch is detected by fixed positive maps, yielding exact thresholds between separability and bound entanglement. For the trace-preserving subclass, we further compare positivity with a recent eigenvalue bound for 2-positive maps, thereby making the gap between positivity and higher-order positivity fully explicit within this family.

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

Select and Improve: Understanding the Mechanics of Post-Training for Reasoning

arXiv:2606.13125v1 Announce Type: cross Abstract: Reinforcement learning has rapidly emerged as a key component in the training of reasoning and coding models, yet it remains poorly understood from a mechanistic perspective. We study how and through what underlying processes capabilities are acquired or enhanced via reinforcement learning post-training. Our analysis, based on controlled math reasoning experiments with Qwen-2.5-1.5B, reveals two core mechanisms: strategy selection and strategy improvement. Our results highlight the role of SFT data and reinforcement learning data in activating these mechanisms, in particular showing how supervising the model on diverse reasoning strategies can enable strategy selection and how increasing difficulty in reinforcement learning data can enable strategy improvement. Taken together, our results provide mechanistic insight into RL training and suggest practical interventions to continue scaling reasoning capabilities.

13.
arXiv (CS.LG) 2026-06-12

Clipping Makes Distributed and Federated Asynchronous SGD Robust to Stragglers

arXiv:2606.13287v1 Announce Type: new Abstract: In modern machine learning, parallelization of training is an important strategy for increasing scale. Asynchronous stochastic gradient descent (ASGD), which maximizes the utilization of available hardware by avoiding waiting for slow workers. However, with constant step sizes, the convergence of ASGD is nonetheless affected negatively by slow workers due to large delays in updates. At the same time, it has been empirically observed in asynchronous training of deep learning models that gradient clipping "stabilizes" training. In this work, we provide a theoretical justification for this behavior, as we show that clipping removes the dependence of the maximum delay in the oracle complexity. We employ a sub-Weibull model of gradient noise which generalizes sub-Gaussian and sub-exponential distributions to more heavy-tailed distributions, motivated by empirical observations in deep learning. We show convergence in expectation, and the first time in asynchronous optimization, convergence with high probability.

14.
arXiv (CS.CV) 2026-06-11

TRON: Tracing Rays to Orchestrate a Neural Renderer for 3D Gaussian Reconstructions

We introduce TRON, a rendering framework that combines 3D Gaussian ray tracing with neural rendering to enable realistic and controllable rendering of real-world 3D scenes under novel lighting, dynamic object motion, object insertion, and material editing. Prior approaches that rely solely on physically based rendering (PBR) of Gaussian representations struggle to achieve realistic relighting due to imperfections in reconstructed geometry, material estimates, and light transport estimation. At the same time, neural rendering methods often lack an explicit scene representation, limiting their ability to support interactive editing with fine-grained manipulation. TRON bridges these two paradigms. We use intrinsic decomposition priors from a learned inverse rendering model to regularize the material properties of a Gaussian field, and repurpose a ray tracer to provide radiometric guidance rather than final pixels. By treating this output as a structured 3D scaffold, we empower a lightweight neural renderer to bridge the domain gap between shading-model constrained estimates and photorealistic output. Our key insight is that the combination of explicit 3D knowledge with robust material priors provides speed and controllability, while neural rendering enables the synthesis of photorealistic images. To support real-world scenarios, we train our neural renderer with a multi-stage strategy consisting of large-scale pretraining and targeted fine-tuning on a newly constructed dataset of 2.1M rendered synthetic and real-world frames from 3D reconstructions. TRON outperforms Gaussian-based relighting methods in realism, and prior neural renderers in editability and speed. To the best of our knowledge, TRON is the first method to enable practical interactive applications in captured 3D environments, offering realistic appearance under dynamic geometric, lighting and material conditions.

15.
arXiv (CS.CV) 2026-06-17

Flash-GRPO: Efficient Alignment for Video Diffusion via One-Step Policy Optimization

Group Relative Policy Optimization has emerged as essential for aligning video diffusion models with human preferences, but faces a critical computational bottleneck: training a 14B parametered model typically demands hundreds of GPU days per experiment. Existing efficiency methods reduce costs through sliding window subsampling training timesteps, but fundamentally compromise optimization, exhibiting severe instability and failing to reach full trajectory performance. We present Flash-GRPO, a single-step training framework that outperforms full trajectory training in alignment quality under low computational budgets while substantially improving training efficiency. Flash-GRPO addresses two critical challenges: iso-temporal grouping eliminates timestep-confounded variance by enforcing prompt-wise temporal consistency, decoupling policy performance from timestep difficulty; temporal gradient rectification neutralizes the time-dependent scaling factor that causes vastly inconsistent gradient magnitudes across timesteps. Experiments on 1.3B to 14B parameter models validate Flash-GRPO's effectiveness, demonstrating substantial training acceleration with consistent stability and state-of-the-art alignment quality.

16.
arXiv (CS.LG) 2026-06-16

An Integrable Token Mixing Layer from the Generalized Yang Baxter Equation

arXiv:2606.15085v1 Announce Type: new Abstract: The YB Mixer is a sequence token mixing layer derived from free fermion and generalized Yang Baxter structures. It applies a core principle from integrable systems where a local algebraic constraint guarantees global computational stability. By using the Ising exchange algebra the mixer creates a free fermionic structure that acts as an exactly norm preserving orthogonal map. This algebra also produces commuting transfer matrices which allow inference to be order free and adaptable to any variable budget. To ensure the model can generalize to longer sequence lengths it uses a spectral circulant generator. This generator maintains the crucial orthogonal and commuting properties of the system. The result is a highly stable and mathematically grounded architecture for sequence processing.

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

Target-Side Paraphrase Augmentation for Sign Language Translation with Large Language Models

Sign language translation (SLT) remains constrained by the limited availability of paired sign-video/text corpora and by the heavy-tailed vocabularies typical of real-world datasets. We study a target-side augmentation strategy in which a large language model (LLM) generates controlled paraphrase variants of the reference spoken-language sentence while the sign input remains unchanged. Concretely, we use GPT-4o to produce semantically faithful variants of the training targets and train a Signformer-style pose-based Transformer under a two-stage schedule: pre-training on the augmented corpus followed by fine-tuning on the original references. We evaluate this strategy on three datasets that span complementary challenges: PHOENIX14T (German Sign Language), a real-world corpus with moderate lexical diversity; the Greek Sign Language Dataset with highly controlled, repetitive recordings; and LSA-T (Argentinian Sign Language), a naturalistic corpus with a large vocabulary and severe long-tail sparsity. This range allows us to characterize precisely when and why target-side augmentation is beneficial. On PHOENIX14T, augmentation improves BLEU-4 from 9.56 to 10.33, demonstrating that paraphrastic exposure helps the decoder generalize beyond memorized reference phrasing. The near-saturated GSL baseline and the extremely sparse LSA-T setting reveal the limits of the approach: in both cases, single-reference lexical overlap metrics are insufficient to capture the full picture, motivating a complementary semantic evaluation. To our knowledge, this is the first study to examine LLM-generated target-side paraphrases as an augmentation mechanism for SLT, and the first to apply an LLM-as-a-Judge evaluation protocol to SLT. This complementary evaluation reveals gains in semantic fidelity that lexical overlap metrics understate.

19.
arXiv (CS.CL) 2026-06-11

To Intervene or Not: Guiding Inference-time Alignment with Probabilistic Model Blending

The wide deployment of LLMs has made model alignment necessary to make newly trained models safely and effectively respond to user instructions. Among different methods, inference-time alignment is often cheaper as it intervenes (i.e., offers guidances) only during output generation. Existing proposals apply guidances extracted from certain aligned models without properly assessing their reliability. Nonetheless, our systematic evaluation reveals that guidance effectiveness varies drastically across models; since ineffective guidances lead to further confusion and thus further interventions, the resulting excessive interventions typically indicate poor performance. To make interventions more effective and thus more efficient, we introduce BlendIn, an inference-time alignment framework that shifts from binary decisions to creating hybrid distributions integrating both models' knowledge. BlendIn stabilizes inference-time alignment by performing quality-aware alignment and proportionally weighting each model's contribution based on reliability. Compared with existing works, it preserves beneficial guidance while downweighting unreliable suggestions. BlendIn provides both diagnostic signals and mitigation strategies for misaligned guidance, achieving consistent and up to 50% performance improvement on challenging model pairs. Our code is available at: https://github.com/DecayingSeart/BlendIn.

20.
arXiv (math.PR) 2026-06-11

Construction of ergodic IDLA forests in $\mathbb{Z}^d$

arXiv:2506.10476v2 Announce Type: replace Abstract: We prove the existence of infinite-volume IDLA forests in $\mathbb{Z}^d$ , with $d \geq 2$, based on a multi-source IDLA protocol. Unlike IDLA aggregates, the laws of the IDLA forests studied here depend on the trajectories of particles, and then do not satisfy the famous Abelian property. Their existence is due to a stabilization result (Theorem 1.1, our main result) that we establish using percolation tools. Although the sources are infinitely many, we also prove that each of them play the same role in the building procedure, which results in an ergodicity property for the IDLA forests (Theorem 1.2).

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

The Latent Color Subspace: Emergent Order in High-Dimensional Chaos

Text-to-image generation models have advanced rapidly, yet achieving fine-grained control over generated images remains difficult, largely due to limited understanding of how semantic information is encoded. We develop an interpretation of the color representation in the Variational Autoencoder latent space of FLUX.1 [Dev], revealing a structure reflecting Hue, Saturation, and Lightness. We verify our Latent Color Subspace (LCS) interpretation by demonstrating that it can both predict and explicitly control color, introducing a fully training-free method in FLUX based solely on closed-form latent-space manipulation. Code is available at https://github.com/ExplainableML/LCS.

22.
arXiv (CS.AI) 2026-06-17

SEAGym: An Evaluation Environment for Self-Evolving LLM Agents

arXiv:2606.17546v1 Announce Type: new Abstract: Self-evolving LLM-based agents improve mainly by changing their agent harness: the structured execution layer around a base model, including prompts, memory, tools, middleware, runtime state, and the model-tool interaction loop. Existing evaluations often reduce this process to isolated task scores or a single sequential curve, obscuring whether an update produces reusable improvement, overfits recent tasks, increases cost, or harms older behavior. We introduce SEAGym, an evaluation environment for measuring agent harness updates across training, validation, test, replay, and cost records. SEAGym turns Harbor-compatible benchmarks into dynamic self-evolution task sources with train batches, frozen update-validation, held-out ID and OOD transfer views, replay diagnostics, and saved snapshot and metric records. Instantiating SEAGym on Terminal-Bench 2.0 and HLE, we compare ACE, TF-GRPO, and AHE under a shared epoch/batch protocol. The results show that these evaluation views provide complementary signals about the evolution process: frequent updates may fail to improve held-out performance, useful intermediate snapshots may collapse later, and source diversity and model backend can affect harness reliability.

23.
arXiv (CS.AI) 2026-06-17

Understanding LLMs in Title-Abstract Screening: From Disagreements to Recommendations

arXiv:2606.17588v1 Announce Type: cross Abstract: Several studies have examined the use of large language models (LLMs) for title-abstract screening in systematic reviews (SRs), reporting mixed accuracy. However, questions of reliability remain largely unaddressed. In this study, we go beyond quantitative LLM-human agreement metrics and qualitatively investigate how and why LLMs fail. We also propose actionable recommendations. We analyzed disagreements between LLMs and researchers across six software engineering SRs and over 1,000 primary study papers. For each SR, papers were screened independently by human experts and LLMs in zero-shot mode, resulting in Kappa values ranging from 0.52 to 0.77. Qualitative analysis suggests that human-LLM disagreement results from recurring, identifiable causes, such as boundary ambiguity in key terms, keyword overemphasization, and incorrect topic inference. Based on these findings, we propose recommendations such as validating semantic understanding before deployment, running multiple LLMs, and focusing validation efforts on borderline cases. Future studies are needed to validate the impact of our recommendations, and community efforts are needed to develop normative guidelines on LLM usage in SRs.

24.
arXiv (CS.AI) 2026-06-19

Uncertainty-Aware Reward Modeling for Stable RLHF

arXiv:2606.19818v1 Announce Type: cross Abstract: Reinforcement learning from human feedback (RLHF) aligns large language models by training reward models on preference data and optimizing policies to maximize predicted rewards. However, this pipeline faces two fundamental challenges: (1) reward models cannot signal when their predictions are unreliable, since they usually act as deterministic point estimators; and (2) modern group-based policy optimization can amplify unreliable reward signals, as exemplified by GRPO's uniform treatment of rewards during advantage computation. As policies explore increasingly diverse responses, these two limitations create a critical vulnerability: unreliable reward estimates may be granted disproportionate influence, triggering severe reward hacking. We propose Uncertainty-Aware Reward Modeling (UARM), which equips reward models with calibrated uncertainty via quantile-based conformal prediction and reweights GRPO advantages through heteroscedastic variance decomposition. Experiments across HelpSteer, UltraFeedback, and PKU-SafeRLHF demonstrate that UARM significantly improves reward model calibration, reduces reward hacking, and enhances downstream alignment quality compared to standard GRPO and uncertainty-agnostic baselines.

25.
arXiv (CS.AI) 2026-06-12

Can I Buy Your KV Cache?

arXiv:2606.13361v1 Announce Type: new Abstract: Right now, across the world, AI agents are repeating the same absurd act: to read one document, they each recompute it from scratch. Every agent re-runs prefill, the most compute-intensive step a large model takes, over identical text, only to rebuild a key-value (KV) cache identical to the one the agent before it just built. The same answer, computed a million times. We make a proposal that is almost offensively simple: compute it once. Let a publisher precompute a document's KV cache, and let every other agent buy the right to load it and skip prefill. It works, and it is token-exact: loading a precomputed KV and continuing matches prefilling from scratch (24/24 greedy tokens, and at the logits level), with no accuracy cost. On Qwen3-4B, reuse is 9-50x cheaper in compute than prefill, and the gap widens with length (prefill's attention scales with L^2), so a single reuse already pays it back. Then the part that matters: where the KV lives. Shipping it fails, because KV is nearly incompressible, so per-load egress costs more than the prefill it saves. Hosting it provider-side, exactly as production prompt-caching works, removes egress entirely. The size of the prize is set by our measured compute saving: serving one hot 3774-token document to 80M agents costs ~$1.5M to re-prefill but only ~$0.03M of reuse compute (49.7x less). The 0.1x cache-read tariff APIs charge passes a 10x discount to users while sitting inside this measured envelope, so the 10x is a floor that the measured ~50x compute saving clears, and the gap to the physical ~50x is provider margin: millions of dollars per popular document. We frame the resulting agent-native prefill CDN and leave lossless KV compression and a cross-party payment layer as the open problems.