中 EN
Sign In Register

Academic Intelligence · Curated Daily

Explore the Frontier of Global Academia

AcademicHub aggregates real-time literature from top journals and preprint platforms. Build your personal research radar and let large language models compile cross-disciplinary analysis briefings automatically.

Sign In Register
01.
arXiv (CS.AI) 2026-06-15

Where's the Plan? Locating Latent Planning in Language Models with Lightweight Mechanistic Interventions

Authors:
Nicole Ma↗Nick Rui↗

arXiv:2605.07984v2 Announce Type: replace-cross Abstract: We study planning site formation in language models – where internal representations of structurally-constrained future tokens form during the forward pass, and whether they causally drive generation. Using rhyming-couplet completion as a clean test of forward-looking constraint, we apply two lightweight methods (linear probing and activation patching) across Qwen3, Gemma-3, and Llama-3 at more than ten scales. Probing shows that future-rhyme information is linearly decodable at the line boundary, with signal that strengthens with scale in all three families. Activation patching reveals that only Gemma-3-27B causally relies on this encoding, exhibiting a handoff in which the causal driver migrates from the rhyme word to the line boundary around layer 30. Every other model we test conditions on the rhyme word throughout generation, with near-zero causal effect at the line boundary despite strong probe signal. We localize the Gemma-3-27B handoff to five attention heads through two-stage path patching that recover ~90% of the rhyme-routing capacity at the newline.

Read & Discuss → View Source →
02.
arXiv (CS.AI) 2026-06-16

HAMON: Passive Optical Sequence Mixing for Long-Horizon Forecasting

Authors:
Alper Y{\i}ld{\i}r{\i}m↗

arXiv:2606.17028v1 Announce Type: cross Abstract: Simple linear and frequency-domain models remain surprisingly competitive in long-horizon time-series forecasting, and recent mechanistic evidence suggests that standard forecasting benchmarks may not require the dense superposed representations that make transformers powerful in other domains. This raises a substrate-level question: if the core forecasting operator is often low-complexity and approximately linear, does it need to be implemented as learned digital temporal mixing? We introduce HAMON, a passive diffractive optical forecasting core in which historical values are encoded onto an optical aperture, future positions are left dark, and cascaded trainable phase masks with free-space diffraction shape the forecast directly in the output field. At inference, prediction is performed by a single passive optical propagation pass with no trainable digital sequence-mixing layer. Across standard benchmarks, HAMON outperforms the strongest digital baselines considered on ETTm2 at all horizons and on ETTh2 at all but the longest horizon, improving MSE by up to 14\% and doing so consistently across horizons rather than at isolated points. It is competitive on Weather and trails the strongest baselines on the remaining ETT settings and on the high-channel-count Traffic and Electricity datasets. Phase encoding, intensity-compatible readout, and phase-scrambling ablations, together with a TorchOptics cross-simulator check, indicate that the forecasts arise from the data-bearing optical field rather than from a digital forecasting head. Because the passive core uses standard Fourier optics, HAMON defines a concrete target for optical hardware and for passive physical sequence mixing.

Read & Discuss → View Source →
03.
arXiv (CS.CL) 2026-06-12

PRISM: Prosody-Integrated Multi-Agent Reasoning Framework for Empathetic Spoken Dialogue

Authors:
Wen Zhang↗Xiaocui Yang↗Zhuoyue Gao↗Shi Feng↗Daling Wang↗Yifei Zhang↗

Empathetic spoken dialogue systems require not only semantically appropriate responses but also emotionally aligned prosodic expression. However, cascade pipelines often discard acoustic cues during speech-to-text conversion, while end-to-end speech models lack interpretable control over emotion and knowledge integration. To address these challenges, we propose PRISM, a multi-agent framework for empathetic spoken dialogue that decouples speech perception, response generation, and speech synthesis into coordinated components. PRISM introduces a prosody-to-language translation mechanism to stabilize large language model reasoning and enables on-demand invocation of external knowledge tools for empathetic dialogue generation. Experimental results demonstrate that PRISM achieves consistent improvements in empathy, prosodic appropriateness, and text response generation quality across objective and subjective metrics. Our code is available at: https://github.com/Bxzfrm/PRISM.

Read & Discuss → View Source →
04.
medRxiv (Medicine) 2026-06-16

Language fMRI lateralization success and head motion in pediatric epilepsy patients with ADHD, and improvements based on fMRI task training

Authors:
Alexander↗Santamaria↗Genc↗Barton↗Kean↗Wray↗Maixner↗Macdonald-Laurs↗Yang↗J. Y. Y.- M↗

Introduction Language functional MRI (fMRI) is a valuable tool for presurgical planning in epilepsy. Functional MRI can be challenging in children, and head motion can compromise its utility. The candidacy of patients with ADHD for fMRI is sometimes queried regarding concerns about possible head motion. In 2020, we implemented an fMRI task training program, via telehealth and/or mock MRI. We aimed to determine whether training increased language lateralisation success and/or reduced head motion in all patients, and in those with ADHD. We also aimed to determine whether patients with ADHD exhibited more head motion during fMRI than those without ADHD. Methods We retrospectively identified 223 epilepsy (85%) and other neurosurgery patients, (241 scans including repeats) with language fMRI at Royal Children's Hospital, Melbourne, Australia, 2016-2024. There were 24 individuals with ADHD listed in the Electronic Medical Record, five of whom had diagnoses of both ADHD and autism; and nine with autism. Language lateralisation success was determined by clinician description recorded as left/right/bilateral in the medical record. 99 patients were provided the training including fMRI task practise. Head motion was quantified by maximum Framewise Displacement (FDmax; mm). Results ADHD was associated with lower language lateralisation success. Training was associated with greater language lateralisation success, across all patients, and in those with ADHD. Regarding ADHD and head motion, outliers in FDmax were seen in 5 young patients with ADHD. Data were trimmed to allow separate investigation of FDmax for the sample with and without extremes of head motion. In untrimmed data, FDmax was significantly higher in patients with ADHD than in those without. In trimmed data, FDmax was on average lower in patients with ADHD than those without, however this was not statistically supported. Regarding training and head motion, across all patients, FDmax was significantly lower for scans with training than without. In patients with ADHD, FDmax was on average lower for scans with training, however training was not associated with FDmax. Conclusions Language fMRI training was associated with higher language lateralization success, particularly in patients with ADHD. Training was associated with reduced head motion across all patients. Although some young patients with ADHD had substantial head motion, most in our sample did not move more than those without ADHD. We conclude that the training program increases success of language fMRI, and that an ADHD diagnosis should not be a contraindication to language fMRI.

Read & Discuss → View Source →
05.
arXiv (CS.CV) 2026-06-19

Hierarchical mutual distillation for multi-view fusion: Learning from all possible view combinations

Authors:
Jiwoong Yang↗Haejun Chung↗Ikbeom Jang↗

Multi-view learning often struggles to effectively leverage images captured from diverse angles and locations. Learning methods for unstructured multi-view images remain largely underexplored. We propose a novel Hierarchical Mutual Distillation for Multi-View Fusion (HMDMV) method, which can handle both structured and unstructured multi-view scenarios. It makes predictions utilizing all possible view combinations: single view, partial multi-view, and full multi-view. The method generates predictions for each view combination and then applies hierarchical mutual distillation to enhance inter-view consistency. An uncertainty-based weighting mechanism further refines the fusion process by adjusting the influence of each view combination according to its prediction confidence, reducing the impact of low-confidence views. Extensive experiments on large-scale structured and unstructured datasets demonstrate that HMDMV consistently achieves state-of-the-art classification accuracy. Another unique advantage of HMDMV is that it provides improved flexibility in inference, allowing for more or fewer view counts in inference than those used in training without additional processing. We also provide a light version with reduced training cost by designing an efficient strategy that randomly samples subsets of view combinations during each training iteration. These results highlight HMDMV's robustness in real-world settings where view availability is variable or incomplete. The code is available at https://github.com/labhai/HMDMV.

Read & Discuss → View Source →
06.
arXiv (CS.CV) 2026-06-16

EyeMVP: OCT-Informed Fundus Representation Learning via Paired CFP–OCT Pretraining

Authors:
Zhuo Deng↗Ruiheng Zhang↗Ziheng Zhang↗Weihao Gao↗Yitong Li↗Qian Wang↗Lei Shao↗Jiaoyue Dong↗Zhixi Zeng↗Lijian Fang↗Haibo Wang↗Xiaobin Lin↗…

Color fundus photography (CFP) is the mainstay for large-scale retinal screening, yet its diagnostic capacity is constrained by the lack of depth-resolved structural information. Optical coherence tomography (OCT) provides cross-sectional retinal anatomy, but is less accessible in population-level screening. Here, we present EyeMVP, a cross-modal retinal foundation model that uses paired CFP–OCT pretraining to learn OCT-informed CFP representations. EyeMVP is pretrained on 674,893 strict same-eye same-day paired CFP–OCT image triples from 112,642 patients across eight hospitals in China. The model uses cross-modal masked reconstruction to enrich CFP representations with OCT-associated supervision, while requiring only CFP images at inference. To accommodate the non-aligned imaging geometry between en-face CFP and cross-sectional OCT, EyeMVP combines source-constrained cross-attention with CFP-derived structural masks. Across 16 downstream tasks, including classification, segmentation, few-shot adaptation, and cross-modal retrieval, EyeMVP outperforms representative retinal foundation models and shows consistent gains on tasks involving macular and optic nerve structure. For CFP-challenging macular diseases, EyeMVP achieves an AUROC of 0.948 for macular edema (vs.~0.852 for EyeCLIP) and 0.825 for myopic macular schisis. In an exploratory reader study, EyeMVP exceeds junior and intermediate ophthalmologist groups but does not reach senior ophthalmologist performance on macular edema, while showing numerically higher balanced accuracy than all reader groups on myopic macular schisis. These results suggest that pixel-level cross-modal reconstruction can enrich CFP representations with OCT-associated supervision, providing a practical route toward stronger CFP-based retinal analysis in screening settings.

Read & Discuss → View Source →
07.
arXiv (CS.AI) 2026-06-25

Agentic Knowledge Tracing: A Multi-Agent LLM Architecture for Stealth Assessment of Financial Literacy in Serious Games

Authors:
Gabriel Santos↗Rita Julia↗Marcelo Nascimento↗

arXiv:2606.25358v1 Announce Type: new Abstract: Assessing financial literacy during gameplay without disrupting the learning experience remains a key challenge in serious games for education. We present the Agentic BKT pipeline, a multi-agent large language model architecture for stealth assessment of financial competencies from open-ended gameplay events. The pipeline processes events from a 2D platformer serious game aligned with the OECD/INFE financial literacy framework through four phases: (1) the game captures every player decision as a structured event log; (2) an LLM event classifier labels each action on a four-point rubric validated against three domain experts (Fleiss kappa = 0.624, substantial agreement); (3) four domain-specific agents specializing in risk mitigation, investing, spending, and credit management perform session-level reasoning over behavioral trajectories, feeding per-competency Bayesian Knowledge Tracing that estimates mastery within each domain; and (4) an expert judge agent synthesizes the domain-level estimates into an overall mastery score. Evaluated with 193 K-12 participants across 264 game sessions, the Agentic BKT pipeline yields mastery estimates significantly correlated with learning gain (r = 0.276, p = 0.0001) and post-test scores (r = 0.333, p < 0.0001) while showing no correlation with pre-test scores, providing both convergent and discriminant validity. The multi-agent approach approximately triples the predictive validity of a single-LLM baseline (r = 0.095, not significant) in this study, demonstrating that domain decomposition and session-level reasoning play a central role in capturing the multidimensional nature of financial literacy from gameplay

Read & Discuss → View Source →
08.
medRxiv (Medicine) 2026-06-24

Durability and Seasonal Variation in the Effectiveness of Nirsevimab over Three Seasons in Connecticut

Authors:
Xu↗Aparicio-Llorente↗Warren↗Kennedy-Shaffer↗Pitzer↗V. E↗Weinberger↗D. M↗Oliveira↗C. R↗PROVE-ID Group Authors↗

Background Nirsevimab has been widely administered in the United States since 2023 to protect infants and young children from severe disease caused by respiratory syncytial virus (RSV). Although early post-licensure studies have shown high effectiveness against medically attended RSV infection, uncertainty remains about the durability of protection, effectiveness beyond the first RSV season, and the extent to which changing RSV seasonality influences real-world effectiveness. Objective To estimate the effectiveness of nirsevimab against medically attended RSV infection across three consecutive RSV seasons and to examine how effectiveness varies by season and time since immunization. Methods We conducted a test-negative case-control study utilizing electronic health records of infants and young children tested for RSV by polymerase chain reaction in outpatient and inpatient settings within the Yale New Haven Health System between October 1, 2023, and March 1, 2026. Effectiveness of nirsevimab was estimated using multivariable logistic regression, adjusting for age, weekly RSV activity, pre-existing risk factors, and other potential confounders. Variation in effectiveness was examined by season, encounter setting, and time since immunization up to 24 months. Results Overall, 17,755 infants and young children were tested for RSV infection, of whom 2,388 (13.4%) were cases and 15,367 (86.6%) were controls. The overall effectiveness of nirsevimab was 67.3% (95% confidence interval [CI]: 59.8, 73.3%) against all medically-attended RSV infections, 60.2% (95% CI: 49.6, 68.5%) against RSV-associated outpatient visits, and 88.9% (95% CI: 82.3, 93.0%) against RSV-associated hospitalization. Effectiveness against medically attended RSV infection declined across seasons, from 76.7% (95% CI: 60.5, 86.3%) in 2023/24 to 54.4% (95% CI: 33.0, 68.9%) in 2025/26. Lower season-specific effectiveness in later seasons corresponded with progressively delayed RSV activity over. Protection against RSV-associated hospitalization declined with increasing time since immunization, from 92.5% (95% credible interval [CrI]: 85.9, 96.4%) at 1 month, to 77.2% (95% CrI: 60.4, 87.6%) at 6 months, and 39.9% (95% CrI: 2.4, 63.3%) at 12 months post-immunization, after which effectiveness plateaued. Conclusions Nirsevimab remained effective against RSV-associated hospitalization through 6 to 12 months after immunization. Delayed RSV activity was associated with lower effectiveness, highlighting the importance of aligning administration with local RSV circulation.

Read & Discuss → View Source →
09.
arXiv (CS.CL) 2026-06-16

Agentic Reinforcement Learning for Search Misaligns Instruction-Tuning

Authors:
Yushi Yang↗Shreyansh Padarha↗Sarah Ball↗Andrew Lee↗Adam Mahdi↗

Agentic reinforcement learning (RL) trains large language models to use tools, but its impact on alignment is poorly understood. We study how agentic RL for search affects the alignment of instruction-tuned (IT) models. We find that RL-trained models inherit refusal reasoning by deflecting harmful requests into benign search queries, but this breaks down under a simple diagnostic trigger that elicits a search call before refusal can occur. Under this condition, RL models produce multi-step unsafe search actions and reasoning, reducing search query safety by up to 68.6% in Qwen and Llama models relative to their IT counterparts. The effect generalises across model families, scales, and RL algorithms. To understand why, we identify linear directions in the residual stream that control search query safety, and show that RL training progressively shifts search behaviour toward the harmful end of this direction. We thus propose representation-guided RL training, which adds a reward penalty based on projection toward the harmful search direction. Training on benign data alone, it restores IT-level alignment without reducing task accuracy and requires no additional training data. Together, our work provides the first framework for diagnosing, mechanistically analysing, and mitigating alignment degradation in agentic RL for search.

Read & Discuss → View Source →
10.
arXiv (CS.AI) 2026-06-12

Prefill Awareness in Large Language Models

Authors:
Andy Wang↗Parv Mahajan↗David Demitri Africa↗Alexandra Souly↗Jordan Taylor↗Robert Kirk↗

arXiv:2606.12747v1 Announce Type: new Abstract: Safety-relevant studies of language models, including alignment and jailbreaking evaluations and AI control protocols, often rely on prefilling model outputs. If AI models can recognize and act on the fact their prior assistant messages have been inserted or edited, the effectiveness and validity of these methods could be compromised. We investigate whether frontier language models can distinguish between tampered and untampered assistant-side context, a capability we call prefill awareness. To do so, we construct a binary preference benchmark across three prefill mechanisms, filtering for cases where models show consistent stances. We find that frontier models show substantial prefill awareness: Claude Opus 4.5 detects prefills opposing its preferences in 9-35% of cases with a 0% false positive rate when prompted; additionally, models often revert towards baseline behavior without explicitly reporting that the prefill was foreign. Controlled ablations later also show that detection and resistance rely on different cues, where stylistic mismatch mainly affects whether models flag a prefill as foreign, while preference mismatch mainly affects whether they revert toward their baseline answer. We also examine more realistic agentic settings such as misalignment-continuation evaluations and SWE-bench trajectories, where frontier models sometimes disavow prefilled assistant turns in ways that depend strongly on dataset, task success, and hidden formatting artifacts. Our results indicate that prefill awareness is already a substantial confound for some prefill-based methods. We recommend that model developers track this capability in frontier systems.

Read & Discuss → View Source →
11.
arXiv (CS.CV) 2026-06-16

Post-Launch Capability Expansion of Vision-Language Models via Prompting for On-Orbit Spacecraft Inspection

Authors:
Nicholas A. Welsh↗Lennon J. Shikhman↗Monty Nehru Attazs↗Seemanthini K. Putane↗Van Minh Nguyen↗Ryan T. White↗

Spaceborne inspection systems often deploy perception models prior to launch, after which updating model weights or expanding fixed label sets becomes operationally impractical. While supervised models can be integrated pre-flight, adding new semantic capabilities in orbit requires retraining and re-uploading parameters. We investigate whether prompt-driven vision–language models can enable post-launch semantic expansion, allowing new spacecraft components to be specified via natural-language prompts without modifying onboard weights. We evaluate zero-shot instance segmentation of spacecraft components under a strictly frozen, single-pass inference protocol on a test set of $129$ images of previously unseen satellites. Under fixed global thresholds and no post-processing, SAM3 achieves $0.385$ mAP@$0.5$ and $0.267$ mAP@$0.5{:}0.95$. Performance is strongly scale-dependent: large structural elements like spacecraft bodies ($0.639$ AP@$0.50$) and solar arrays ($0.598$ AP@$0.5$) localize reliably, while relatively small appendages like antennas ($0.221$ AP@$0.5$) and thrusters ($0.081$ AP@$0.5$) remain difficult. Prompt formulation influences performance, with structured prompts incorporating spatial and geometric descriptors yielding up to $82%$ improvement over short category-name prompts. The model operates within the memory and compute envelope of contemporary embedded GPUs, suggesting prompt-driven grounding can provide a practical mechanism for post-launch semantic extension of dominant spacecraft structures while highlighting limitations of zero-shot localization for fine-scale components under orbital domain shift.

Read & Discuss → View Source →
12.
arXiv (CS.LG) 2026-06-16

MIRAGE: Auditing Anti-Muslim Bias in Frontier LLMs Across Reasoning, Agentic, and Time-Coupled Conditions

Authors:
Noor Islam S. Mohammad↗Tamim Sheikh↗

arXiv:2606.16562v1 Announce Type: new Abstract: Five years after the discovery of persistent anti-Muslim bias in large language models, most evaluations remain confined to single-turn prompt completion, a setting that no longer reflects how frontier LLMs are deployed. We introduce MIRAGE (Muslim-Identity Reasoning and Agentic Generation Evaluation), a benchmark of 1{,}200 prompts spanning three deployment-realistic conditions: direct completion, chain-of-thought reasoning, and simulated agentic decision-making across content moderation, lending triage, refugee claim summarization, and hiring screens. Across six frontier models, we find that (i) chain-of-thought reasoning amplifies rather than suppresses Muslim-violence associations by 12–34\% relative to direct completion, (ii) agentic decisions exhibit a 9–22 percentage-point asymmetry between Muslim and matched non-Muslim cases on identical evidence, and (iii) bias is sharply time-coupled to retrieved news context, increasing 18–27\% under recent-conflict retrieval. Existing prompt-based mitigations transfer poorly across our three conditions, suppressing direct-completion bias while leaving agentic asymmetry largely intact. We release MIRAGE and an open evaluation harness to support targeted mitigation research.

Read & Discuss → View Source →
13.
arXiv (CS.CV) 2026-06-11

ARGUS: Stacked Multi-View Identity Mosaic Injection for Subject-Preserving Video Generation

Authors:
Zijie Meng↗Jiwen Liu↗Yufei Liu↗Chengzhuo Tong↗Xiaoqiang Liu↗Yuanxing Zhang↗Yulong Xu↗Pengfei Wan↗

Subject-preserving video generation is not solved by frontal-face similarity alone: a generated person must remain recognizable across motion, large viewpoint changes, expression shifts, occlusion, scale variation, and conflicts among text, first-frame, and identity references. We argue that the central bottleneck is the point-reference paradigm, which collapses identity into a single static observation entangled with pose, accessories, lighting, background, and camera statistics. We introduce Argus, a Wan-based framework centered on Stacked Multi-View Identity Mosaic Injection (SMII). SMII converts MLLM-selected image/video identity evidence into a 3*3 stacked mosaic, synchronizes the mosaic with the current diffusion time, and injects it as negative-time read-only memory in Wan's native token space. This turns identity from an external clean adapter or a single reference image into a compact dynamic distribution. Around SMII, an MLLM Identity Director selects informative identity moments and resolves condition conflicts, while no-cross-pair counterfactual training, Temporal Identity Annealing, and Adaptive Self-Likeness Guidance improve robustness without paired subject-video supervision. We further release HardID-Celeb, a public-figure identity-stress benchmark, and introduce YawScore and OccScore to probe large-yaw and first-frame-occlusion robustness. Argus achieves state-of-the-art results on OpenS2V-Eval Human-Domain, reaching 64.38 Total Score, 71.86 FaceSim, 51.62 NexusScore, and 79.14 NaturalScore. On HardID-Celeb, Argus obtains 76.80 FaceSim and improves YawScore and OccScore by 12.60 and 15.10 points over the strongest baselines, demonstrating that dynamic identity memory and large-scale counterfactual self-supervision are highly effective for subject-preserving video generation.

Read & Discuss → View Source →
14.
arXiv (CS.LG) 2026-06-25

Lifelong In-Context Learning with Transformers Requires Parametric Forms of Attention

Authors:
Luke McDermott↗Robert W. Heath jr↗Rahul Parhi↗

arXiv:2606.25342v1 Announce Type: new Abstract: Lifelong continual learning remains an obstacle on the path to human-like intelligence. Modern transformers show sparks of intelligence with in-context learning. The quadratic nature of attention, however, prohibits transformers from performing this process on arbitrarily long sequences. In this work, we argue that extending in-context learning to lifelong settings is a practical solution for continual learning in AI agents. In particular, we argue that parametric forms of attention are needed to understand a lifetime of context with transformers on a fixed hardware budget. These attention mechanisms learn the relationship between keys and their associated values at test-time with parametric regression. Our generalization of parametric approaches (linear attention, state-space models, fast weight programmers, and test-time training layers) contrasts with nonparametric counterparts like softmax attention. They replace the ever-growing key-value cache with an online-trainable neural network, maintaining a constant memory footprint. We highlight how parametric attention currently fall short of lifelong learning due to limited memory capacity or costly online updates. To address these issues, we pose a set of open questions with novel insights to guide the field toward long-horizon agents.

Read & Discuss → View Source →
15.
arXiv (CS.CV) 2026-06-25

An Improved Variational Method for Image Denoising

Authors:
Jing-En Huang↗Jia-Wei Liao↗Ku-Te Lin↗Yu-Ju Tsai↗Mei-Heng Yueh↗

The total variation (TV) method is an image denoising technique that aims to reduce noise by minimizing the total variation of the image, which measures the variation in pixel intensities. The TV method has been widely applied in image processing and computer vision for its ability to preserve edges and enhance image quality. In this paper, we propose a Mixed-norm TV (MixTV) model for image denoising and the associated numerical algorithm to carry out the procedure, which is particularly effective in removing several types of noise and their combinations. Our MixTV admits a unique solution and the associated numerical algorithm guarantees convergence. Numerical experiments are demonstrated to show improved effectiveness and denoising quality compared to other TV models. Such encouraging results further enhance the utility of the TV method in image processing. Our project page is available at https://angusbb.github.io/MixTV.

Read & Discuss → View Source →
16.
arXiv (CS.AI) 2026-06-25

Elo-Disentangled Player-Style Embeddings for Human Chess via Rating-Conditioned Residual Move Model

Authors:
Jason Carlson↗

arXiv:2606.25176v1 Announce Type: new Abstract: We study representation learning for individual human chess style: a per-player embedding learned from a player's move history such that inner products measure stylistic similarity, while being approximately disentangled from playing strength (Elo). Our key design is a residual formulation: a rating-conditioned base move model (Maia-3 policy logits plus Stockfish-derived features, scored over Maia-2-proposed candidates) captures what a typical player of a given strength would play, and a frozen copy of it anchors a learned move encoder and a per-player vector z, so that z explains only deviations from rating-typical play. The base model improves move prediction over the strong Maia-3 policy by 27-37% relative NLL across the rating spectrum, with the largest gains at the top (2800+); Stockfish's marginal value grows monotonically with Elo (negligible at 900-1200, +0.085 nats at 2800+). On a shared Elo-stratified benchmark of 22,620 held-out decisions, top-1 move-matching rises monotonically from Maia-2 to Maia-3 to the Stockfish-augmented base (0.51 -> 0.57 -> 0.68): the base is +33% relative top-1 over Maia-2 and +19% over Maia-3 (30% lower NLL), with the engine-feature lift largest at high Elo. The player embedding adds little to raw move-matching on top of this base – its marginal top-1 gain falls within the 95% confidence interval – and its value is instead representational: z generalizes to held-out decisions without overfitting, re-identifies players from disjoint games above chance, and a linear probe recovers rating from z with only R^2 = 0.06 (no better nonlinearly), evidence it captures style on an Elo-orthogonal axis. We argue that a strong rating-conditioned base plus a compact, Elo-disentangled embedding – separating typical play from individual deviation – is an economical, interpretable model of individual style, an alternative to per-player preference fine-tuning.

Read & Discuss → View Source →
17.
arXiv (CS.AI) 2026-06-25

Internal Data Repetition Destroys Language Models

Authors:
Jessica Chudnovsky↗Joshua Kazdan↗Noam Levi↗Rylan Schaeffer↗Yegor Denisov-Blanch↗Bo He↗Mehmet Donmez↗Sanmi Koyejo↗David Donoho↗

arXiv:2606.24998v1 Announce Type: cross Abstract: Language models are running out of high-quality training data, and even aggressively deduplicated corpora retain some amount of repetition. Earlier controlled studies predated Chinchilla-style scaling laws and could only measure the cost of repetition indirectly. We revisit repetition in the Chinchilla era, using a fitted no-repetition scaling law to report Compute-Equivalent Gain and Compute-Equivalent Loss. We show that under this modernized paradigm, repetition damage is systematic in three ways. First, holding compute allocated to repeated data constant, eval loss peaks at an intermediate repeat count $\Rep$; repeating a moderately sized subset a moderate number of times damages performance more than repeating a large subset a few times or a small subset many times. Second, the location of this peak is well-fit by a power law in model size; this scaling law reveals that the most damaging number of repeated data grows more quickly than compute. Finally, when repeated documents consume 10\% of the FLOPs budget in a controlled exact-document repetition setting, the compute-equivalent loss can be large: on FineWeb-Edu-Dedup, the most damaging repeat count for a Qwen3-style 344M-parameter model at $\operatorname{OT}=1$ matches the loss of a no-repetition run using 67% of the FLOPs. We demonstrate that these phenomena are not language-model-specific, and can be analytically understood in a simple statistical model: a misspecified linear regression with verbatim duplicates reproduces the same qualitative loss peak, quantifying how such peaks can arise from a statistical tradeoff between memorization and generalization. Our findings add precision to the study of duplication in language models, allowing practitioners to quantify the wasted compute incurred by the presence and repeat structure of duplicates in pretraining corpora.

Read & Discuss → View Source →
18.
arXiv (CS.AI) 2026-06-25

What Does It Mean to Break a Distillation Defense?

Authors:
Lena Libon↗Pura Peetathawatchai↗Michael Aerni↗Daniel Paleka↗Florian Tram\`er↗

arXiv:2606.25059v1 Announce Type: cross Abstract: Black-box LLMs (accessible only via API) are vulnerable to distillation attacks, in which an attacker queries the model and trains a student on its outputs. A recent line of work proposes output perturbation defenses that modify the teacher's output to reduce student performance while preserving utility for legitimate users. As a relatively new family of approaches, output perturbation defenses lack a shared threat model, making it difficult to compare them, reason about composing them with other attacks, or evaluate their robustness against realistic adversaries. This underspecification matters beyond technical evaluation: when defenses are deployed to protect intellectual property or justify regulatory compliance, an imprecise threat model can create a false sense of security. We propose a threat model framework that describes attackers along three dimensions: a query budget, a data budget, and an interface profile that captures how attackers interact with the API. Using antidistillation sampling as a case study, we show that whether the defense is considered effective depends on the assumed threat model. We argue that future work on distillation defenses, along with any governance or policy frameworks built around them, should explicitly specify and stress-test attacker capabilities along our three dimensions.

Read & Discuss → View Source →
19.
arXiv (quant-ph) 2026-06-17

Quantum conditional entropies from convex trace functionals

Authors:
Roberto Rubboli↗Milad M. Goodarzi↗Marco Tomamichel↗

arXiv:2410.21976v4 Announce Type: replace Abstract: We study geometric properties of trace functionals that generalize those in [Zhang, Adv. Math. 365:107053 (2020)], arising from a novel family of conditional entropies with applications in quantum information. Building on new convexity results for these functionals, we establish data-processing inequalities and additivity properties for our entropies, demonstrating their operational significance. We further prove completeness under duality, chain rules, and various monotonicity properties for this family. Our proofs draw on tools from complex interpolation theory, multivariate Araki–Lieb and Lieb–Thirring inequalities, variational characterizations of trace functionals, and spectral pinching techniques.

Read & Discuss → View Source →
20.
arXiv (CS.CV) 2026-06-19

SAFE-Cascade: Cost-Adaptive Vision-Language Routing for Chart Question Answering

Authors:
Ayush Dwivedi↗Qixin Wang↗Ashvi Soni↗Ruoteng Wang↗Han Li↗Animesh Mahapatra↗Neeraj Agrawal↗Xintao Wu↗

Vision-language models (VLMs) are powerful for chart question answering, but invoking a VLM for every query can be unnecessarily expensive when many questions are answerable from OCR text and lightweight language reasoning. We demonstrate SAFE-Cascade, an interactive system for cost-adaptive chart question answering. Given a chart image and a natural-language question, SAFE-Cascade first extracts chart text with OCR, obtains a provisional answer from a text-only language model, and then uses a learned router to decide whether to accept the text answer or escalate to a VLM. The demo exposes this decision process to users: OCR evidence, text-only answer, routing probability, escalation decision, final answer, estimated cost, and estimated latency are shown side by side. SAFE-Cascade is designed as a transparent interface for understanding when visual grounding is actually needed. Users can upload or select charts, ask questions, inspect the evidence used by each pathway, compare text-only and VLM answers, and adjust the escalation threshold to explore the accuracy-cost frontier. The system is implemented with Azure Document Intelligence for OCR, gpt-5-mini as the text-only model, gemini-2.5-flash-image as the VLM, and a Random Forest router trained on inference-time features. On a held-out ChartQA test split of 375 examples from a 2,500-example experiment, SAFE-Cascade achieves 69.1% unified accuracy with 73.1% VLM invocation, compared with 67.7% accuracy and 100% VLM invocation for the full-VLM baseline. The observed +1.4 percentage-point difference is statistically uncertain, so we interpret SAFE-Cascade as matching full-VLM performance while reducing VLM calls by 26.9% and estimated cost by 9.3%. The demonstration shows how selective modality routing can make multimodal knowledge systems more transparent, tunable, and cost-aware.

Read & Discuss → View Source →
21.
arXiv (CS.CL) 2026-06-19

NEST: Narrative Event Structures in Time for Long Video Understanding

Authors:
Ali Asgarov↗Kaushik Narasimhan↗Najibul Haque Sarker↗Hani Alomari↗Chia-Wei Tang↗Anushka Sivakumar↗Zaber Ibn Abdul Hakim↗Shaurya Mallampati↗Chris Thomas↗

Recent progress in vision-language models has enabled the processing of increasingly long video sequences, but the ability to handle extended token streams does not translate to understanding of narrative structure in long videos. Existing long video benchmarks focus on needle-in-a-haystack retrieval rather than evaluating how low-level actions form events, how events interact across time, and how narratives progress, for example, whether a model can connect an early setback, such as a job loss to a later relationship breakup, despite long gaps, intervening scenes, or flashbacks that reframe what occurred. We introduce NEST (Narrative Event Structures in Time for Long Video Understanding), a dataset of 1005 full-length movies (avg. 98 minutes), each annotated with 102 multimodal narrative events grounded in visual content, dialogue, and audio. NEST captures multimodal narrative events with structured annotations grounded in visual content, dialogue, and audio, and links them through relations that reflect narrative structure, including temporal ordering, hierarchical composition, and long-range dependencies. We introduce baselines for event trigger detection (ETD), event localization (EL), event argument extraction (EAE), and event relation extraction (ERE). The benchmark is highly challenging for grounded event discovery, with ETD below 8%, EL under 6%, and EAE below 11%. In contrast, ERE is more tractable once events are given, reaching 35.45% F1 zero-shot and 44.42% F1 after fine-tuning.

Read & Discuss → View Source →
22.
arXiv (CS.AI) 2026-06-25

Specifying AI-SDLC Processes: A Protocol Language for Human-Agent Boundaries

Authors:
Ylli Prifti↗

arXiv:2606.20615v2 Announce Type: replace Abstract: AI agents now participate as first-class team members across the software development lifecycle, yet no specification language exists for expressing the human-agent responsibility boundaries, approval gates, and governance constraints this collaboration requires. Existing approaches encode process in agent prompts (subject to drift), target adjacent domains (workflow management, business processes), or address only fragments (access control, approval gates). We propose a domain-specific language for specifying AI-SDLC processes as protocols, with formal syntax, well-formedness conditions, operational semantics, and enforcement invariants. The language distinguishes policy (declared intent) from mechanism (structural enforcement), enabling implementations to bound process non-determinism through primitives such as validation tokens and capability boundaries. Three results follow. A failure rate analysis shows that structural enforcement bounds system failure rates at a weighted product of agent and validator rates, while behavioral compliance permits cumulative or near-saturating growth. The 2+N team pattern (two human-in-control roles plus N specialized agent members) formalizes classical Separation of Duties for AI-SDLC. Kleene closure of orchestration loops and reflexive protocol-adherence validation emerge as design properties rather than special-case constructs. We position the contribution against multi-agent frameworks (MetaGPT), workflow specification (FlowAgent, BPMN extensions), and capability-based security (SAGA): the novelty lies in the specific integration, not any single primitive. A working implementation demonstrates feasibility; empirical evaluation is future work.

Read & Discuss → View Source →
23.
arXiv (CS.CL) 2026-06-12

When Iterative RAG Beats Ideal Evidence: A Diagnostic Study in Scientific Multi-hop Question Answering

Authors:
Mahdi Astaraki↗Mohammad Arshi Saloot↗Ali Shiraee Kasmaee↗Hamidreza Mahyar↗Soheila Samiee↗

Retrieval-Augmented Generation (RAG) extends large language models (LLMs) beyond parametric knowledge, yet it is unclear when iterative retrieval-reasoning loops meaningfully outperform static RAG, particularly in scientific domains with multi-hop reasoning, sparse domain knowledge, and heterogeneous evidence. We provide the first controlled, mechanism-level diagnostic study of whether synchronized iterative retrieval and reasoning can surpass an idealized static upper bound (Gold Context) RAG. We benchmark eleven state-of-the-art LLMs under three regimes: (i) No Context, measuring reliance on parametric memory; (ii) Gold Context, where all oracle evidence is supplied at once; and (iii) Iterative RAG, a training-free controller that alternates retrieval, hypothesis refinement, and evidence-aware stopping. Using the chemistry-focused ChemKGMultiHopQA dataset, we isolate questions requiring genuine retrieval and analyze behavior with diagnostics spanning retrieval coverage gaps, anchor-carry drop, query quality, composition fidelity, and control calibration. Across models, Iterative RAG consistently outperforms Gold Context, with gains up to 25.6 percentage points, especially for non-reasoning fine-tuned models. Staged retrieval reduces late-hop failures, mitigates context overload, and enables dynamic correction of early hypothesis drift, but remaining failure modes include incomplete hop coverage, distractor latch trajectories, early stopping miscalibration, and high composition failure rates even with perfect retrieval. Overall, staged retrieval is often more influential than the mere presence of ideal evidence; we provide practical guidance for deploying and diagnosing RAG systems in specialized scientific settings and a foundation for more reliable, controllable iterative retrieval-reasoning frameworks.

Read & Discuss → View Source →
24.
arXiv (CS.AI) 2026-06-24

Sol Video Inference Engine: Agent-Native Full-Stack Acceleration Framework for Efficient Video Generation

Authors:
Yitong Li↗Junsong Chen↗Haopeng Li↗Haozhe Liu↗Jincheng Yu↗Ligeng Zhu↗Ping Luo↗Song Han↗Enze Xie↗

arXiv:2606.23743v1 Announce Type: cross Abstract: Modern video diffusion models achieve higher generation quality through scaling, but this also increases inference cost. Although many acceleration methods have been proposed, a central challenge is that the most effective acceleration strategy is highly instance-specific: a recipe that works well for one combination of model, hardware, and inference configuration often does not transfer to another. Different models vary in architecture, numerical sensitivity, and attention concentration patterns. Inference settings differ in spatial and temporal resolution and video duration, while hardware platforms differ in memory hierarchy, supported numerical formats, and kernel throughput. These factors create a large tuning space, making manual performance engineering costly. We present Sol Video Inference Engine, an agentic, native, training-free acceleration framework for video diffusion models. It organizes five broadly applicable techniques, cache, sparse attention, token pruning, quantization, and kernel fusion, into an agentic acceleration stack for instance-specific optimization. For a concrete deployment target defined by a model, hardware platform, and serving configuration, parallel skill agents optimize the implementation of each technique, an agent integrator composes them into a global acceleration stack, and a human validator provides feedback on generation quality. We instantiate this workflow on three video models with different sizes and architectures: 64B Cosmos3-Super, 22B LTX-2.3, and 2B SANA-Video. With little human effort, the full stack achieves more than 2x end-to-end acceleration while maintaining near-lossless VBench quality, demonstrating the effectiveness of the agent framework for video diffusion acceleration.

Read & Discuss → View Source →
25.
arXiv (CS.AI) 2026-06-16

Multi-Granular Node Pruning for Causal Circuit Discovery

Authors:
Muhammad Umair Haider↗Hammad Rizwan↗Hassan Sajjad↗A. B. Siddique↗

arXiv:2512.10903v2 Announce Type: replace Abstract: Circuit discovery aims to identify minimal subnetworks that are responsible for specific behaviors in large language models (LLMs). Existing approaches primarily rely on iterative edge pruning, which is computationally expensive and limited to coarse-grained units such as attention heads or MLP blocks, overlooking finer structures like individual neurons. We propose a node-level pruning framework for circuit discovery that addresses both scalability and granularity limitations. Our method introduces learnable masks across multiple levels of granularity, from entire blocks to individual neurons, within a unified optimization objective. Granularity-specific sparsity penalties guide the pruning process, allowing a comprehensive compression in a single fine-tuning run. Empirically, our approach identifies circuits that are smaller in nodes than those discovered by prior methods; moreover, we demonstrate that many neurons deemed important by coarse methods are actually irrelevant, while still maintaining task performance. Furthermore, our method has a significantly lower memory footprint, 5-10x, as it does not require keeping intermediate activations in the memory to work.

Read & Discuss → View Source →