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.

01.
arXiv (quant-ph) 2026-06-17

Quantum Computing Algebra (QCA), the theory and implementation

arXiv:2606.17621v1 Announce Type: new Abstract: We present a real geometric algebra framework designed for the direct translation of the Dirac formalism into geometric algebra representations. Unlike previous approaches based on positive-definite signatures, QCA employs a split-signature construction that enables a natural realization of quantum states and operators while simplifying computational implementation. We further present an implementation of QCA using the GAALOP software and show how quantum gates and multi-qubit systems can be efficiently represented and generated computationally. As an application, we demonstrate the use of QCA in quantum game theory, where the real-algebraic formulation provides computational advantages for modeling entangled strategies and quantum interactions. The proposed framework establishes a practical bridge between the abstract formalism of quantum computation and efficient geometric algebra implementations.

02.
Nature (Science) 2026-06-24

Genetic diversity of late Neanderthals in northwestern Europe

Archaeological, osteological and genetic evidence suggests that Neanderthals lived in small groups1,2; however, less is known about whether these groups were part of isolated communities or belonged to larger, well-connected populations3. The dense concentration of broadly contemporaneous Neanderthal sites in the Meuse Basin, Belgium4, provides a rare opportunity to study regional populations at high resolution. Here we generated genetic data from 27 Neanderthals who lived less than approximately 52,500 years ago from ten archaeological sites in Belgium and France, including a high-coverage genome from a 45,000-year-old individual from Goyet, Belgium. We show that most of these individuals are more closely related to one another than to other contemporaneous late Neanderthals in Europe. Further, some of these individuals carry DNA from a Neanderthal lineage predating the split of late Neanderthals. Although these Neanderthals overlapped temporally with early modern humans in northwestern Europe from around 47,000 years ago, we find no evidence of recent gene flow from modern humans. They also do not show the genetic signatures of mating among close relatives found in Altai Neanderthals, suggesting that they lived in larger or better-connected groups. Moreover, genetic load did not accumulate over time, arguing against progressive genetic deterioration as a driver of Neanderthal extinction. Genetic sequencing of multiple late Neanderthals living less than 52,500 years ago provides an overview of genetic diversity and demonstrates that even low-coverage nuclear genome data can increase resolution of within-Neanderthal diversity.

03.
arXiv (CS.CL) 2026-06-25

Constraint Tax in Open-Weight LLMs: An Empirical Study of Tool Calling Suppression Under Structured Output Constraints

Tool Calling and Structured Output are two core capabilities of modern Agent systems, yet their interaction under joint deployment conditions remains insufficiently understood. This paper reports a reproducible phenomenon observed in a production Agent system: when Tool Calling and JSON Schema constraints are simultaneously enabled, multiple open-weight models cease invoking tools despite maintaining high schema compliance. We refer to this behavior as Tool Suppression. Through controlled experiments across multiple model families and deployment settings, we consistently reproduce Tool Suppression under joint constraints, while tool execution and schema compliance remain functional when evaluated independently. Further analysis reveals that JSON Schema constraints are compiled into grammar-based token masks, causing tool-call tokens to become unreachable during decoding. This provides an implementation-level explanation for the observed behavior. To interpret the phenomenon, we formulate the Constraint Priority Inversion (CPI) hypothesis, which suggests that schema satisfaction may dominate action-selection behavior under multiple simultaneous constraints. We present CPI as a behavioral hypothesis consistent with the observed evidence rather than a verified internal mechanism. To mitigate the problem, we propose Transparent Two-Pass Execution, an inference-time strategy that decouples tool execution from schema-constrained response generation. Experimental results show that this approach restores tool invocation while preserving structured output guarantees without requiring model retraining. These findings suggest that evaluating tool use and structured output separately may overlook important reliability issues in production Agent systems. Code, data, and docs will be released at https://github.com/Fzsama/Constrain-Tax-26-06.git.

04.
arXiv (CS.CL) 2026-06-19

Efficiently Representing Algorithms With Chain-of-Thought Transformers

The increasing popularity of reasoning models – language models that output a series of reasoning or thought tokens before producing an answer – is justified, in part, by theoretical results showing that chain-of-thought (CoT) transformers can simulate Turing machines, and thus perform arbitrary computation. However, the Turing machine, while suitable for complexity-theoretic analysis, is not convenient, intuitive, or efficient for discussing algorithms. Algorithms are typically designed and analyzed at a higher level of abstraction, captured by the Word RAM model with random-access memory and unit-cost operations on $\bigO(\log n)$-bit words. As a result, Word RAM algorithms can be substantially more efficient than their Turing machine counterparts, raising the question: Can CoT transformers efficiently simulate Word RAM algorithms? For instance, can they sort $n$ items in $\bigO(n \log n)$ steps or run Dijkstra's algorithm in $\bigO(E + V \log V)$ steps? We answer affirmatively, up to poly-logarithmic overhead. We first establish this for finite-precision transformers with poly-logarithmic width and rightmost unique hard attention, then strengthen the result to two more practical settings with finite width and log-precision: continuous CoT, where reasoning takes the form of vectors rather than tokens, and a hybrid architecture in which transformer layers sit atop a recurrent (linear RNN) layer. In all three cases, we find that CoT can efficiently simulate any Word RAM algorithm with only a poly-logarithmic overhead in $n$. This overhead reduces to log-square when the Word RAM has a ``flat'' instruction set, and only logarithmic for multiplication-free flat instructions – in stark contrast to known CoT simulations of Turing machines, which require quadratic overhead over Word RAM.

05.
arXiv (CS.LG) 2026-06-16

deFOREST: Fusing Optical and Radar satellite data for Enhanced Sensing of Tree-loss

arXiv:2510.14092v2 Announce Type: replace-cross Abstract: In this paper we develop a deforestation detection pipeline that incorporates optical and Synthetic Aperture Radar (SAR) data. A crucial component of the pipeline is the construction of anomaly maps of the optical data, which is done using the residual space of a discrete Karhunen-Lo\'{e}ve (KL) expansion. Anomalies are quantified using a concentration bound on the distribution of the residual components for the nominal state of the forest. This bound does not require prior knowledge on the distribution of the data. This is in contrast to statistical parametric methods that assume knowledge of the data distribution, an impractical assumption that is especially infeasible for high dimensional data such as ours. Once the optical anomaly maps are computed they are combined with SAR data, and the state of the forest is classified by using a Hidden Markov Model (HMM). We test our approach with Sentinel-1 (SAR) and Sentinel-2 (Optical) data on a $92\,km \times 92\,km$ region in the Amazon forest. The results show that both the hybrid optical-radar and optical only methods achieve high accuracy that is superior to the recent state-of-the-art hybrid method. Moreover, the hybrid method is significantly more robust in the case of sparse optical data that are common in highly cloudy regions.

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

Order Is Not Control

AI alignment, interpretability, steering, and neural perturbation studies identify order-inducing objects. We argue that order is not control. Control requires a receiver-gated response law: a denominator-indexed operator mapping material state, action/drive, bath, and receiver state to response displacement, sinks, effort, and basin projection. We identify it across biological, LLM, adapter, and stochastic-operator panels. The laws are local: an intervention can be admitted, saturated, sign-changing, leaky, or overdriven depending on medium, bath, receiver state, action port, and comparator. Control is assigned when finite effort moves a target or outcome-readout class under the same denominator while damage, null/evasive, invalid format, overdrive, and unnecessary effort stay bounded. Mouse ALM, C. elegans, and zebrafish panels provide physical response-operator evidence while excluding coordinate identity and controller conclusions. LLM panels show generated-output response laws: across four material conditions, response vectors are predictable at 72.8-73.7% component-sign accuracy, rising to 84.3-84.8% on nonzero components; held-out observers predict system-effect and target/oracle families at 93.6% and 91.7% accuracy. Constitution-conditioned adapters reshape susceptibility as prepared media, and stochastic-operator panels separate measured opportunity from deployable action policies. This gives a driven-dissipative response-system account at the mesoscopic control level: drives act through prepared media, baths, and receivers, producing admitted movement, impedance, sinks, or overdrive. The evidence supports local admitted control and measurable stochastic response operators, while leaving deployable pre-generation control, hidden/logit causal sufficiency, biological-to-LLM coordinate identity, and literal thermodynamic quantities outside scope.

07.
arXiv (CS.AI) 2026-06-16

EMS: Multi-Agent Voting via Efficient Majority-then-Stopping

arXiv:2604.02863v2 Announce Type: replace Abstract: Majority voting is the standard for aggregating multi-agent responses into a final decision. However, traditional methods typically require all agents to complete their reasoning before aggregation begins, leading to significant computational overhead, as many responses become redundant once a majority consensus is achieved. In this work, we formulate efficient multi-agent voting as a reliability-aware agent scheduling problem and propose Efficient Majority-then-Stopping (EMS) to improve reasoning efficiency. EMS first estimates a Task-Conditioned Reliability Ordering (TCRO) for each agent by retrieving its historical consensus evidence on semantically similar queries, and then invoking agents in descending reliability order. Next, Adaptive Incremental Voting (AIV) terminates the process once the current leading answer cannot be overturned by any possible votes from the remaining agents, and returns this answer. Finally, Reliability History Updating (RHU) updates only the invoked agents according to their consensus with the final decision. Extensive evaluations across five benchmarks show that EMS preserves the accuracy of Majority Voting while reducing the average number of invoked agents by 35% and token consumption by 44%, respectively. The code is available at https://github.com/fuyu66/EMS.

08.
arXiv (CS.CL) 2026-06-16

Spectro-Temporal Interference Confounds Phase Encoding in Spatial Audio Foundation Models

Recent spatial self supervised audio models achieve high performance on localization tasks, raising questions about their encoding of microsecond interaural phase fine structures. We propose a psychoacoustic benchmark based on the binaural masking level difference to evaluate this. Using an equalization cancellation baseline and a GCC PHAT positive control we evaluate nine frozen audio models spanning binaural SSL, monaural SSL, and neural audio codecs. Four monaural negative controls yield zero BMLD confirming binaural specificity. Two general purpose binaural SSL models exhibit minimal phase sensitivity while dedicated binaural spatial SSL models achieve BMLD comparable to the analytical baseline. Progressive physical ablations show that general purpose binaural SSL models rely on spectro temporal interference textures rather than cross channel phase computation. High detection rates in speech reflect a confounding reliance on broadband envelopes rather than genuine phase encoding.

09.
Nature Medicine 2026-06-08

Apitegromab for lean mass preservation during tirzepatide-induced weight loss: a randomized, double-blind, placebo-controlled phase 2 trial

Loss of lean mass in proportion to total weight loss is observed with incretin mimetic therapies such as tirzepatide and has the potential to adversely affect health and function. Apitegromab is an investigational, fully human monoclonal antibody that selectively inhibits myostatin activation and is, thereby, capable of increasing muscle mass. In the randomized, double-blind, placebo-controlled phase 2 EMBRAZE study, adults with overweight or obesity (n = 102) were randomized 1:1 to receive tirzepatide plus apitegromab (10 mg kg−1) or tirzepatide plus placebo. At week 24, apitegromab resulted in a least square mean (80% confidence interval (CI)) of 1.9 (1.2−2.7) kg less lean mass loss than placebo (P = 0.001), despite similar total body weight loss between groups, representing a 54.9% retention of lean mass relative to placebo. In participants receiving apitegromab, trough concentrations of apitegromab and total latent myostatin, a pharmacodynamic marker, both increased over time and reached a plateau after approximately 16 weeks. Incidence of adverse events (AEs) (% (95% CI)) was generally similar across apitegromab-treated participants and placebo-treated participants, with 39 of 51 (76% (63−86%)) and 36 of 51 (71% (57−81%)) participants experiencing an AE, respectively. Serious adverse events (SAEs) were balanced and experienced by one of 51 (2% (0−10%)) participants in each arm. In summary, this proof-of-concept study demonstrated that selective targeting of myostatin by apitegromab was well tolerated and effective in preserving lean mass when combined with tirzepatide. ClinicalTrials.gov identifier: NCT06445075 . In the phase 2 EMBRAZE study, participants receiving tirzepatide and apitegromab lost less lean mass compared to participants receiving tirzepatide and placebo.

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

One Transit Is All You Need: Detecting Exoplanets Through Learned Stellar Behaviour with EXOVEIL

arXiv:2606.02778v3 Announce Type: replace-cross Abstract: I present EXOVEIL, a transit detection system that learns what a star's brightness should look like and flags when reality disagrees. Unlike existing systems that require phase-folded input, EXOVEIL operates on raw flux time series and can detect planets that transit only once.A Transformer world model, trained on 16,499 Kepler light curves with transit-masked self-supervised learning, predicts expected stellar flux. A matched-filter detector with variance weighting extracts transit signals from the prediction residuals. A learned classifier (XGBoost) separates planets from false positives, achieving AUC 0.938 on Kepler DR25. Applied to single-transit injection-recovery, EXOVEIL recovers 32% of transits at 1000 ppm depth a task where all classification-based systems score 0% by construction. A blind search of 3,737 Kepler stars yields 179 new transit-like signals not present in the DR25 TCE catalogue, including 46 monotransit candidates. Applied withoutretraining to 47 confirmed TESS planets in the PLATO LOPS2 field, EXOVEIL achieves 100% recovery, demonstrating zero-shot cross-mission transfer. At PLATO's 25-second cadence, detection reaches 100 ppm – approaching the Earth-analog regime. I provide the first application of conformal prediction to transit detection (95.9% empirical coverage) and release the system as pip install exoveil with pretrained weights and a candidate catalogue.

11.
PLOS Medicine 2026-05-13

On the evolution of the company we keep: Implications for infectious disease modeling

Authors:

by Joël Mossong Whom we meet shapes how infections spread. Where earlier focus of mathematical epidemiology was on incorporating age, more recent work has begun to reveal the importance of socioeconomic aspects for understanding and managing future epidemics. In this Perspective, Joël Mossong discusses the importance of understanding social contacts and how they have evolved for infectious disease modeling, and the need to factor in additional considerations such as ethic and socioeconomic backgrounds.

12.
arXiv (CS.CL) 2026-06-16

DoubtProbe: Black-Box Jailbreak Defense via Structural Verification and Semantic Auditing

As large language models (LLMs) are increasingly deployed in user-facing systems, black-box jailbreak defense has become an important practical problem. Existing defenses often rely on known-attack coverage, prompt-level semantic judgment, or local runtime control, yet these paths can become unstable under evolving prompt packaging, expression rewriting, and structure manipulation. We observe that many black-box jailbreaks do not remove the harmful goal, but reorganize the information needed to express and execute it, thereby evading safety alignment while remaining recoverable during generation. Motivated by this observation, we propose DoubtProbe, a dual-branch inference-time defense framework that combines structural verification with semantic auditing and formulates black-box jailbreak defense as consistency checking under controlled transformation. The structural branch extracts a structured representation from the original request, reconstructs the request under representation constraints, and detects information-preservation failures between the original and reconstructed requests; the semantic branch audits the original prompt directly. We evaluate DoubtProbe against representative black-box defenses on jailbreak and benign-request benchmarks, and further test backbone transfer from Qwen2.5-72B to Llama-3.1-70B. Results show that DoubtProbe achieves a stronger and more stable defense-utility trade-off: on Qwen2.5-72B, it reduces the JBB attack success rate from 0.293 to 0.100 and the CodeAttack attack success rate from 0.152 to 0.001, while maintaining false positive rates of 0.022 and 0.016 on AlpacaEval and OR-Bench; the same pattern remains stable on Llama-3.1-70B. These findings show that structural inconsistency signals provide a practical and generalizable basis for black-box jailbreak defense, especially when combined with semantic auditing.

13.
arXiv (CS.CL) 2026-06-19

Connect the Dots: Training LLMs for Long-Lifecycle Agents with Cross-Domain Generalization Via Reinforcement Learning

This work presents a general framework for training large language models (LLMs) to "Connect the Dots" (CoD), a meta-capability required by long-lifecycle agents: as an LLM-based AI agent gets deployed in an environment, it solves a long sequence of tasks while continuously exploring the environment, learning from its own experiences, and iteratively self-updating its context about the environment, thereby achieving progressively better performance on future tasks conditioned on the updated context. Major components of the CoD framework include: (1) algorithm design and infrastructure for end-to-end reinforcement learning (RL) with long rollout sequences interleaving solve-task and update-context episodes; (2) tasks and environments for incentivizing and eliciting the targeted meta-capability in LLMs during training, as well as for faithfully measuring progress during evaluation. We present proof-of-concept implementations of the CoD framework, including a GRPO-style RL algorithm with fine-grained credit assignment, as well as tasks and environments tailored to the targeted meta-capability (rather than domain-specific LLM capabilities or standard task-by-task RL). Empirical results validate the efficacy of end-to-end RL training in the CoD setting, and demonstrate the potential for out-of-distribution generalization – within the training domains, across different domains, and from CoD to Ralph-loop settings – of the elicited meta-capability. Our investigation of CoD connects several lines of prior works, and opens up new opportunities for advancing LLMs and AI agents. To facilitate further research and applications, we release our implementations at \url{https://github.com/agentscope-ai/Trinity-RFT/tree/research/cod/examples/research_cod}.

14.
arXiv (CS.CV) 2026-06-15

Hybrid Classical-Quantum (HCQ) Alzheimer's Classification via Supervised $\beta$-VAE and Quantum Kernels

This paper presents a two-stage Hybrid Classical-Quantum (HCQ) pipeline for binary Alzheimer's disease (AD) classification from 3D T1-weighted structural MRI volumes, where the classical and quantum components are designed to complement each other rather than operate independently. A supervised 3D $\beta$-variational autoencoder (VAE) is trained end-to-end under voxel-wise reconstruction, KL-divergence, and focal classification losses that compress each 3D MRI volume (resized from 152 x 184 x 152 to 96 x 96 x 96) into a 64-dimensional latent code. Partial Least Squares (PLS) regression selects the six components in the latent code that best separate Alzheimer's Disease (AD) from cognitively normal (CN) subjects and rescales them into rotation angles, which are encoded onto a six-qubit register using the ZZ quantum feature map to give us the respective quantum states. The input to a precomputed-kernel Support Vector Machine (SVM) is an N x N Gram matrix (N = 308), created by calculating the overlap between every pair of quantum states. The novelty of this work lies in the fact that the quantum kernel operates directly on disease-aware features that are learned end-to-end by a supervised autoencoder, rather than on pre-extracted inputs. On 308 ADNI-1 subjects, consisting of 137 AD and 171 CN subjects, the baseline achieved 67.2% accuracy and 0.759 AUC, while the stability-enhanced variant reached 72.1% accuracy and 0.799 AUC with cross-fold variance halved. 3D Grad-CAM further helped validate our model's focus on brain regions linked to Alzheimer's. The HCQ pipeline could serve as a general-purpose framework for diagnostic classification across biomedical imaging domains that present similar challenges for classical approaches.

15.
arXiv (CS.CV) 2026-06-24

Latent Visual States for Efficient Multimodal Reasoning

The integration of visual evidence has significantly enhanced the capabilities of large multimodal models. However, this integration predominantly relies on generating discrete outputs (etc., code or box coordinates) to invoke external tools, a process that introduces rigid dependencies and substantial latency. To overcome these limitations, we propose {EVA} (LatEnt Visual StAtes), a novel framework that natively generates continuous latent visual representations. These internal representations manifest as an adaptive sequence of Latent\_slot tokens, serving as intermediate visual thoughts during the reasoning process. These Latent\_slot tokens are then trained end-to-end with the discrete text tokens. This co-optimization, notably, causes extreme policy deviation in the 'transition window' following the Latent\_slot tokens. We develop D-GSPO (Decouple-GSPO) to target this root cause by decoupling the optimization of latent and discrete components. To support SFT, we construct EVA-230K, a high-quality text-image interleaved CoT dataset encompassing a diverse range of real-world scenes, documents, charts and OCR tasks. Extensive experiments across multiple benchmarks confirm that EVA achieves significant performance gains while enhancing inference efficiency.

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

Ensuring Trustworthy Online A/B Testing: Addressing Five Key Questions on CUPED

arXiv:2606.18750v1 Announce Type: cross Abstract: A/B testing has become the gold standard for data-driven decision-making in large-scale online experimentation, providing critical guidance for feature launch, pricing optimization, and user experience enhancement. To maximize statistical sensitivity, many technology companies routinely employ Controlled-experiment Using Pre-Experiment Data (CUPED), a technique that achieves substantial variance reduction while preserving the unbiasedness of estimating the average treatment effect. Despite its widespread adoption, several critical methodological and practical nuances of CUPED remain underexplored. This paper systematically addresses five frequently encountered yet overlooked questions regarding the application of CUPED. First, we provide a comparative analysis of various post-CUPED estimators to identify the optimal adjustment specification. Second, we evaluate the validity of regression-based adjustments and delineate robust variance estimation methods tailored for such frameworks. Finally, we extend our investigation to complex but common scenarios, including multi-arm experiments and two-stage sampling designs. Our findings reveal that in these settings, naive reliance on standard variance estimators can lead to severely misleading inferences. By offering rigorous theoretical insights and extensive experimental validation, this work deepens the conceptual understanding of CUPED. Notably, the recommended methodologies have been successfully deployed and integrated into ByteDance's experimentation platform.

17.
arXiv (CS.CL) 2026-06-16

From ASR to ASP: Evaluating Prompt Attack Vulnerabilities Against Open-Source LLMs

Recent studies demonstrate that Large Language Models (LLMs) are vulnerable to attacks that generate harmful or sensitive outputs. As open-source LLMs are increasingly adopted in high-impact applications such as finance, law, and healthcare, systematically investigating their security risks is becoming increasingly important towards trustworthy LLM era. This paper comprehensively studies effective prompt injection attacks against 14 widely used open-source and three closed-source LLMs on five attack benchmarks. Moreover, existing evaluation metrics mostly only consider the attack success rate, overlooking uncertainty in model responses. Our proposed Attack Success Probability (ASP) additionally captures uncertain behaviors for evaluation, where the model may initially refuse a harmful request but subsequently provide harmful guidance or vice versa, reflecting inconsistency and ambiguity in attack feasibility. By systematically analyzing the effectiveness of prompt injection attacks, we propose a straightforward and effective hypnotism attack; results show that this attack causes aligned language models, including Stablelm2, Mistral, Openchat, and Vicuna, to generate objectionable behaviors, achieving around 90% ASP. They also indicate that ignore prefix attacks can break all 14 open-source LLMs, achieving over 60% ASP on a multi-categorical dataset. We find that moderately well-known LLMs exhibit higher vulnerability to prompt injection attacks, highlighting the need to raise public awareness and prioritize efficient mitigation strategies.

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

AgentRivet: an automated system for producing Rivet routines from journal publications

arXiv:2606.13535v1 Announce Type: cross Abstract: Particle physics collider experiments provide Rivet routines as part of the analysis preservation strategy for model-independent measurements. Rivet is a C++ toolkit that allow new theoretical models to be compared to the measurements, thus aiding the development and tuning of Monte Carlo event generators as well as searches for physics beyond the Standard Model. However, analysis coverage is known to be incomplete, with only 39% of measurements having documented and publicly available Rivet routines. In this article, we design and implement an automated workflow based on Large Language Models with the goal of providing the missing routines. This multi-step workflow, referred to as AgentRivet, extracts the physics analysis information from published papers and writes the missing Rivet routines, with intermediate code- and physics- reviews as part of an autonomous quality control. We report the results obtained using commercial Large Language Models, provided by OpenAI, Anthropic, and Google, for two recent measurements from the ATLAS and CMS experiments. We find that AgentRivet produces competent Rivet routines with few syntax errors. The physics fidelity of the routines is reasonable and follows the explanations given in the relevant publications. Nevertheless, physics-implementation issues do arise and are investigated using the artefacts produced by AgentRivet. The majority of physics implementation issues arise from subtle-but-ambiguous definitions in the given publication, although some models struggle to implement complex observables even when clear definitions are given.

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

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

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.

20.
arXiv (CS.AI) 2026-06-16

Training and Evaluating Diffusion Policies with Long Context Lengths

arXiv:2606.16447v1 Announce Type: cross Abstract: Imitation learning has enabled highly-dexterous robotic manipulation from RGB observations. Policies trained with these methods, however, typically condition robot actions on only a short history of observations. These policies cannot solve tasks that require memory and can get stuck repeatedly executing the same failing motions. In this work, we first benchmark policy performance as context length is incrementally increased from short to long, across a spectrum of tasks with varying local stability and memory requirements, and in multiple data regimes. To our knowledge, this is the first study to investigate context length in imitation learning at this level of detail. Our results challenge prior claims: naively scaling context length is not as brittle as advertised in literature. With an appropriate conditioning method and denoising backbone (UNet+Cross-Attention), single-task policies achieve high success rates on many tasks in the usual data regime even with naive scaling. Next, we propose a training algorithm to jointly train policies at multiple context lengths, further reducing the sample complexity of long-context learning. Finally, we apply our findings to re-evaluate some previously proposed solutions to long-context imitation learning.

21.
arXiv (CS.CL) 2026-06-16

Replay What Matters: Off-Policy Replay for Efficient LLM Reinforcement Unlearning

LLM unlearning has emerged as a cost-effective alternative to full retraining for removing hazardous knowledge from pretrained models while preserving general utility. Recent RL-based methods such as RULE reformulate unlearning as learning a refusal behavior, but their on-policy optimization repeatedly samples from the same forget and retain/boundary prompts throughout training. We identify a critical inefficiency in this process: easy cases quickly converge and provide little useful gradient signal, while hard cases near the forget/retain boundary continue to produce low-reward rollouts that are discarded after a single use. To address this issue, we propose ReRULE, an off-policy replay enhancement for reinforcement unlearning. ReRULE stores low-reward hard-case rollout groups in a replay buffer during early GRPO training and reuses them in later stages through importance-sampled off-policy updates, redirecting computation toward boundary cases that still require learning. Theoretically, we show that ReRULE yields a tighter hard-case convergence bound than pure on-policy RULE. Empirically, ReRULE improves MUSE-Books Retain Quality from 46.3 to 56.2 while adding only 5–11% training time across benchmarks. Its limited improvement on the simpler TOFU setting further supports the intended conditional behavior: replay is most beneficial when the hard/easy disparity is pronounced.

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

Self-CTRL: Self-Consistency Training with Reinforcement Learning

arXiv:2606.18327v1 Announce Type: cross Abstract: Language models (LMs) that faithfully describe their own behavior can more easily be audited, understood, and trusted by users. This paper describes Self-Consistency Training with Reinforcement Learning (Self-CTRL), a method that optimizes for consistency between a LM's self-explanations and behavior on related inputs by updating explanations to better predict behavior or updating behavior to better match explanations. We apply our method in two domains. First, we study a formal probabilistic reasoning task in which LMs must learn to imitate a family of biased samplers and evaluated on their ability to report the associated biases. We find that consistency training improves the correlation between self-reported and behaviorally-measured latent biases from $R^2=0.24$ to $R^2=0.64$ on a set of held-out distributions, matching the generalization of direct ground-truth supervision. Second, we study a constitutional AI domain in which LMs must describe when they will refuse or comply with user requests. Here, Self-CTRL produces rules that faithfully describe the model's behavior on held-out requests, improving the refusal predictions of a third-party auditor model from $36\%$ to $92\%$. In the other direction, behavior updates improve alignment, reducing HarmBench failure rate from $15.0\%$ to $0.5\%$ without substantially increasing refusal on harmless prompts. By aligning explanations and behavior, our work provides a general recipe for training AI models to be safer, more transparent, and more controllable.

23.
arXiv (CS.AI) 2026-06-12

The Containment Gap: How Deployed Agentic AI Frameworks Fail Public-Facing Safety Requirements

arXiv:2606.12797v1 Announce Type: new Abstract: Agentic large language model systems that autonomously invoke tools, maintain persistent memory, and execute multi-step plans are increasingly deployed in public-facing domains, including government services, healthcare triage, and financial advising. We ask whether the frameworks used to build these systems provide architectural-level structural safety guarantees. Applying six containment principles derived from a compositional model of agentic architectures, we audit three dominant frameworks (LangChain, AutoGPT, and OpenAI Agents SDK) and find no native compliance in any of them. Memory integrity, a defense against one of the most prevalent vulnerability classes, is not observed in any of the three evaluated frameworks. We validate these findings empirically: in a simulated government benefits agent built on LangChain, a single memory-poisoning write induces persistent targeted corruption across all tested seeds and backends, increasing the wrongful denial rate for targeted applicants to 88.9%. Under a complex five-factor policy, the same attack preserves aggregate accuracy while increasing targeted wrongful denials by 3.5x, rendering the corruption difficult to detect through standard monitoring. We then introduce two lightweight containment mechanisms: a memory integrity validator and a policy gate, which eliminate both attack vectors with sub-millisecond overhead (

24.
arXiv (CS.CV) 2026-06-25

Benchmarking Vision-Language Models for Microscopic Plant Image Understanding

Microscopic imaging provides essential visual evidence for studying plant biology and pathology at the cellular and subcellular levels. However, existing benchmarks on vision-language models primarily focus on macroscopic plant imagery, while the microscopic domain remains underexplored. To address this gap, we present PlantMicro, a comprehensive benchmark for evaluating vision-language models (VLMs) in microscopic plant imagery. PlantMicro integrates more than 5,000 images collected across diverse hosts, biological domains, and imaging modalities. Building on this diversity, we design a set of complementary tasks that capture different facets of microscopic image understanding. To support these tasks, we construct over 9,000 VQA pairs that systematically evaluate the capabilities of VLMs. Experiments on PlantMicro show that current VLMs struggle with fine-grained recognition and biologically grounded reasoning. For example, GPT-5 achieves 34.93% accuracy on the pathogen classification task, which is only modestly above the random-guessing baseline. The results highlight a significant gap in current VLMs' ability to comprehend plant microscopic images. PlantMicro provides a standardized foundation for advancing VLMs toward reliable and comprehensive microscopy-level plant understanding.

25.
arXiv (CS.AI) 2026-06-24

video-SALMONN-R$^3$: Learning to ReWatch, ReAsk, and ReAnswer for Efficient Video Understanding

arXiv:2606.24477v1 Announce Type: cross Abstract: Video large language models (LLMs) are often constrained by computation and memory budgets, leading them to use reduced frame rates and spatial resolutions, which may cause them to miss critical information for question answering (QA). A practical and efficient solution is a two-stage paradigm: first perform coarse video understanding to localize relevant segments, and then re-watch these segments at higher temporal or spatial fidelity. In this paper, we present video-SALMONN-R$^3$, the first end-to-end video-LLM that enables re-watch through reinforcement learning without relying on chain-of-thought (CoT) cold-start. This design removes the need for costly CoT data annotations and avoids CoT-based supervised fine-tuning (SFT), which can otherwise degrade the pretrained video understanding abilities. To address the mismatch between the reasoning-first behavior induced by re-watch and the answer-first tendency of pretrained video-LLMs, we propose a re-answer strategy, in which the model first produces a direct answer in the first watch and then refines it after re-watching. Finally, to improve question adherence during re-watching, we propose a re-ask mechanism that re-injects the query when revisiting localized segments. Experimental results show that video-SALMONN-R$^3$ consistently outperforms both the base model and the QA-SFT baseline, while surpassing prior re-watch-based approaches with significantly lower computational cost. Code, models, and data will be publicly released upon acceptance.