Masked Diffusion Models (MDMs) are well known for fast text generation via parallel decoding. However, independent parallel sampling can commit to inconsistent predictions, and less is known about how to selectively revise these early token commitments. In this work, we introduce $\underline{\mathrm{P}}$lanned $\underline{\mathrm{U}}$nmasking-and-$\underline{\mathrm{R}}$evision Joint $\underline{\mathrm{L}}$earning (PURL) for MDMs, a training-based framework that seamlessly unifies parallel generation with selective token revision. We derive a normalized joint transition whose likelihood factorizes into edit selection, unmask selection, and token prediction. These components are trained jointly on partially masked sequences constructed according to the model's own unmasking policy. During inference, all three decisions are supported by a single shared forward pass, enabling the model to decode masked tokens and revise noisy tokens. Across logic puzzles, mathematical reasoning, and code generation, PURL improves accuracy at matched unmask budgets, with relative gains up to 47.4% accuracy improvement and up to an $8\times$ reduction in scheduled decoding rounds. Beyond these gains, our analysis further reveals general principles of revisable decoding: token edits become more effective as context accumulates; early edits can steer unfinished trajectories, whereas later edits tend to remain localized; and effective edits therefore depend on both calibrating when to intervene and learning where and how to revise. Together, these suggest that effective parallel generation depends not on revising more tokens, but on making the right commitments revisable at the right time.
Machine-learned circuit parameters could remove one of the main bottlenecks of variational quantum eigensolvers: the classical optimization loop, with its typical barren plateaus, local minima, and costly energy and gradient evaluations, is replaced by a direct prediction. Prior work showed that machine-learning models paired with a structured circuit design can transfer across system size and structure within hydrogenic systems. The next step: Extending them to organic molecules would seem to require re-training on use-case-specific training sets. In this work, we show that this is not necessary by establishing an alchemical approach. A geometric projection places surrogate hydrogen atoms along the bond directions of the target molecule and so maps its local bonding environment onto small hydrogenic fragments. Existing hydrogenic solvers can then be reused as they are. The only new quantity is a single projection radius per bond type, which is fitted once on a small reference molecule and reused unchanged thereafter. We demonstrate the approach on the separable-pair circuit designs that have been used in prior work and extend the approach to multi-circuit approaches that span effective bases for quantum subspaces.
Salient and camouflaged object detection (SOD and COD) both recover foreground support, yet most joint methods output only a binary mask and therefore cannot distinguish the type. USC12K distinguishes the two types, but its baseline USCNet still binds foreground existence to the type within a single normalization, so the failure source cannot be attributed. Our cross-task study shows that foreground localization transfers between SOD and COD, whereas type discrimination does not: a COD-only model recovers foreground on SOD images but its type judgement follows the training distribution rather than the foreground-background relation, so the two decisions require separate mechanisms. SegDis factorizes the output into an explicit foreground probability and a conditional type probability given foreground: FLR provides high-quality task-agnostic support, and RTA refines the SOD/COD assignment from the foreground-background relation without a task identifier. Because support is explicit, each ground-truth foreground pixel is classified as correct-type, type-swapped, or missed, so foreground-recovery and type-assignment errors are diagnosed separately. With PVT-v2-B2, SegDis attains 68.98% FG-mIoU and 78.35% mIoU on USC12K at about one-eighth of USCNet's parameters; CSCS is 5.75.
AI agents deployed on the open web are becoming more capable of performing actions that may breach contractual provisions and violate computer security laws. In this paper, we introduce ScrapeBench, the first agentic benchmark for empirically evaluating the legal compliance of AI agents with website scraping policies, namely terms of service (ToS) and $\texttt{robots.txt}$ directives. Across widely used closed-weight and open-weight AI agents, we find that most agents exhibit low compliance rates (< 5%) when instructed to scrape websites in our sample of 1,626 popular websites that explicitly prohibit scraping. In addition, we observe that even where AI agents actively check the applicable ToS and $\texttt{robots.txt}$, they continue to exhibit low compliance rates. Conversely, in our sample of 269 websites that do not prohibit scraping, we observe that agents sometimes refuse to scrape despite it being lawful, such as Claude Opus 4.6 exhibiting an over-refusal rate of 98.5%. Our human baseline experiments (n = 180 participants) suggest that humans exhibit compliance rates comparable to AI agents but over-refuse at much higher rates.
This paper contains examples of stereotypes about veterans and displaced people that may be offensive or distressing. Social bias benchmarks for language models commonly examine categories such as gender, race, and religion, often drawn from lists of legally protected characteristics. Veterans and internally displaced people (IDPs) fall outside these lists, and stereotypes about them, which can undermine access to employment, housing, and social participation, remain largely absent from language model evaluation. We introduce WarBias, a Ukrainian-English benchmark of stereotypes about these two groups. Built from a taxonomy of 120 stereotype claims grounded in research and monitoring reports, WarBias contains 1,007 stereotype-counter-stereotype-unrelated sentence triplets and 240 multiple-choice questions per language. A Ukrainian language model selected in a blinded human pilot generated the Ukrainian triplets. At least two independent annotators reviewed each triplet in the annotation cohort. The authors revised items in response to the reviews, and the final release was machine-translated into English. Each claim carries a human severity rating. In a baseline evaluation of six models, four prefer stereotypical Ukrainian sentences about at least one group, most strongly about IDPs, while English scores are mostly balanced. WarBias supports complementary evaluations of stereotypical preferences and context-sensitive judgments, and provides paired forget/retain material for unlearning-based bias mitigation. We document the benchmark's construction, intended uses, and limitations, and offer its construction method as a template for analogous benchmarks in other conflict-affected societies.
Multimodal LLMs have made remarkable progress in perceiving the physical world, while their perceptual scope remains largely human-centric, focusing on what humans can directly see or hear. This progress raises the prospect of perception beyond human limits, such as extending hearing to sounds that are inaudible to humans. More broadly, the physical world is full of waves, vibrations, and oscillations, only a small fraction of which are directly audible to humans. This invites the extension of multimodal LLMs toward a broader form of machine hearing, which remains largely underexplored. To investigate this direction, we introduce the Machine Hearing Benchmark (MHB), comprising 6,128 instances across 13 tasks and seven domains. MHB spans highly heterogeneous input regimes, with sampling rates from approximately $10^{-5}$ to $10^6$ Hz, sequence lengths from $10^1$ to $10^6$ samples, and both single- and multi-channel inputs. Evaluations of current open-source and proprietary Audio LLMs (ALLMs) reveal limited zero-shot transfer to machine hearing. Supervised adaptation on machine-hearing data alone does not close this gap, suggesting that the challenge extends beyond data exposure to a systematic mismatch between existing audio front ends and machine-hearing inputs. To address this mismatch, we introduce MachEar, a lightweight encoder extension that complements the pretrained Audio LLM for machine-hearing inputs. The results show that MachEar equips a pretrained Audio LLM with stronger machine-hearing capabilities while largely preserving its existing audio understanding performance.
Continual learning suffers from catastrophic forgetting, leading to the loss of past knowledge and unstable adaptation across task streams. Existing approaches typically address this issue through replay buffers, architectural changes, or additional regularization terms. We argue that forgetting is a consequence of unstable learning dynamics, which may benefit from an optimizer-level solution. The main challenge is that learning dynamics are time-dependent, as demonstrated by empirical phenomena such as the stability gap. First, we formally prove that, for distance-based classifiers, bounded parameter drift guarantees bounded margin erosion under a local score-stability condition. Based on this insight, we propose STABLE, a primal-dual optimizer with adaptive drift control. STABLE enforces a drift budget by increasing the dual variable when the budget is violated and decreasing it otherwise. This results in high stability without penalizing plasticity. Experiments on long-horizon unimodal streams and CLIP-based multimodal streams show that vanilla PEFT baselines trained with STABLE match or outperform specialized continual learning methods. In the multimodal setting, STABLE further preserves zero-shot transfer, matching or exceeding dedicated multimodal continual learning methods. Code is available at https://github.com/atlas-luiss/STABLE.
The design of manufacturing systems relies on discrete-event simulation (DES) models that experts still build in tool-specific interfaces. AI agents can now write simulation code, but every model they write from scratch is new, unverified software. We present SimuLean, an open-source 3D DES simulator on the Unity engine designed to be programmed by AI agents: its modelling semantics are verified once, in the platform, and the agent builds and studies one model through domain-level commands, in the 3D scene or headless; physically simulated durations are calibrated for fast headless studies. We evaluated it on 18 manufacturing modelling tasks, from single machines to published production lines. A frontier agent built valid models in 93 of 108 attempts (86%), in about two minutes each; on the benchmark's interface version, two frontier agents built 105 of 108 models through domain commands alone, so that each comes with a record of how it was built. Every residual failure was silent - a model that builds and runs but answers wrongly - and only comparison with reference models revealed them all. In an industrial case, one agent working alone rebuilt, verified and studied a robotised welding workshop from a published study, in SimuLean and in SimPy. Comparing them with the published, expert-built model exposed an error in it; against the corrected model, the agent-built models agree within 1%. One agent can already build complete, realistic and traceable models for the design of manufacturing systems; independent reference models are what make them trustworthy.
Multimodal Large Language Models (MLLMs) inherit strong textual reasoning capabilities from their language backbones and text pretraining, but often fail to apply the same level of competency to visual inputs. We show that this degradation cannot be attributed to only perception failures, by evaluating models on semantically-equivalent text and rendered-text image inputs: models can accurately read text in the images and recover near-text performance when provided their own transcriptions, but just cannot reason well on the images directly. We characterize this discrepancy as a modality reasoning-access gap, where models fail to route perceived visual content into existing reasoning machinery accessible with text. To bridge this, we propose TIGER (Text-to-Image Gap-targeted training for Enhanced Reasoning), which first mines synthetic modality counterfactuals where a model succeeds on a text problem but fails when provided the corresponding rendered image, and post-trains the model on the image input via Group Relative Policy Optimization (GRPO), along with SFT and DPO variants. Importantly, although trained exclusively on synthetic rendered-text images with no multimodal annotations, TIGER results in model performance improvements across diverse real-world visual benchmarks including geometric diagrams (MathVerse), multi-disciplinary sciences (EMMA, MMMU), complex charts (ChartQAPro), and embodied spatial reasoning (RoboSpatial-Home). TIGER also complements existing state-of-the-art RLVR visual reasoning methods, yielding further performance gains. Analysis involving reasoning-subspace activation and activation patching provide support indicating that TIGER helps to align visual-to-reasoning pathways toward text reasoning states. Our implementation is available at https://github.com/minhngh/Tiger.
Hardware executions of the quantum approximate optimization algorithm (QAOA) are often validated by comparing the device's mean cost with a simulation, a check that can miss what matters for optimization. Under global depolarization, the mean normalized gap and the floor-corrected probability of every set of bitstrings, the optimum included, retain the same fraction of their noiseless improvement over uniform sampling. We test this prediction with depth-2, transferred-angle QAOA circuits for binary subproblems of an alternating-direction decomposition on the IBM Heron processor \texttt{ibm\_marrakesh}, with and without dynamical decoupling and twirling. At 12 qubits, the main-family circuits retained 29\% to 46\% of their mean-gap improvement but only 5\% to 15\% of their probability of the optimum, lower on at least nine of ten instances per run. Near-optimal bitstrings were lost about as much as the optimum, whereas the avoidance of poor bitstrings was retained best. In the first submission, a device-matched noise model reproduced the pattern but overestimated the retention of the optimum five- to eightfold. On a rugged planted family, the device found the optimum in 62\% of 768-shot blocks (noiseless: 98\%), within the range of local heuristics. Validations of optimization circuits should report retention of the optimum and of near-optimal sets.
Visual data are often expressed through multiple complementary modalities (e.g., images and segmentation masks), each capturing different aspects of the same underlying structure. Most existing multimodal methods adopt a top-down paradigm, learning a shared latent variable through joint training from scratch, which limits scalability. We propose Split and Bridge (SnB), a bottom-up framework for multimodal generation that couples pretrained unimodal diffusion models at sampling time. Rather than learning a unified latent space, SnB generates coherent multimodal samples by following a shared diffusion trajectory up to a split point and then branching into modality-specific trajectories guided by a bridge module. This design enables plug-and-play reuse of powerful pretrained models without joint training, making it scalable and easily adaptable. Across standard benchmarks (PolyMNIST, CelebAMask-HQ) and more complex settings (PIE-Bench++, ROCO), SnB achieves a strong balance between sample quality and coherence. We show that SnB naturally extends to downstream zero-shot tasks such as interleaved co-generation, image editing, and style coupling/decoupling. Our results demonstrate that bottom-up generation provides a practical framework that complements existing top-down multimodal approaches. Code is available at https://github.com/AhmadArrabi/SnB
Large language models (LLMs) are increasingly explored for resource management across the Cloud-Edge continuum. However, embedding LLMs directly inside latency-sensitive control loops places inference latency, execution jitter, and hallucination risks on the critical path. This study argues for a different use of LLMs: a continually instruction-tuned intent compiler that runs once per management intent or declared context change, above a fast, deterministic tactical controller that handles all low-latency runtime actions. We formulate intent compilation as a continual instruction tuning problem, because management intents arrive sequentially and accumulate while infrastructure and workload contexts drift. Unlike recent LLM-based management frameworks, our design scores the model's outputs by execution, so that catastrophic forgetting can be measured in the executed objective of the underlying controller. On a continuum simulator with a 2B backbone and parameter-efficient adapters, preliminary results show that naive sequential tuning forgets earlier intents in the executed objective, that replay of stored examples nearly prevents this forgetting, and that the tuned compact compiler performs on par with simpler translators while being more robust than a grammar parser to rephrased intents.
Large language model (LLM) recommenders can produce semantically different recommendations when a user's protected attribute changes, even if the interaction history remains fixed. Controlling this disparity is challenging when model weights and full-vocabulary logits are inaccessible. We propose FAIDER (Fairness-Aware Item-level DEcoding for Recommendation), which incorporates fairness into list construction by selecting complete item candidates rather than manipulating token-level logits or repeatedly repairing finished outputs. The central idea is to separate the operational criterion for a violation from how strongly generation is constrained. Conformal-inspired empirical calibration fixes a cross-group semantic-disparity threshold, while online decoding applies either hard filtering or a soft penalty adjusted through dual-inspired feedback to target a prescribed violation rate. Separately, we establish a pathwise guarantee: under a uniform enforceability assumption, the soft decoder's empirical violation rate exceeds its target by at most $O(1/T)$. Experiments across recommendation benchmarks and LLM backbones show that Soft FAIDER improves both measured semantic fairness and ranking quality over the evaluated prompt-repair baselines. It also generally achieves a more favorable fairness-utility balance than the hard variant, indicating that stricter per-recommendation constraints do not necessarily yield lower measured disparity.
Probabilistic neural surrogates for PDEs enable ensemble forecasting, but often exhibit miscalibration when used for uncertainty quantification, especially in compute-constrained, few-step inference regimes. Existing generative approaches typically rely on fixed global stochasticity, which cannot adapt predictive uncertainty to the local dynamical state. We propose TORNADO, a latent probabilistic forecasting framework that formulates next-step prediction as stochastic transport between VAE encoder posteriors. Unlike existing approaches with globally prescribed stochasticity, TORNADO learns a SDE with state-dependent diffusion, enabling the predictive uncertainty to adapt to the local latent dynamics. The model is trained using analytic conditional endpoint targets derived from a Gaussian interpolant, enabling simulation-free learning. An additional energy distance regularization term improves distributional alignment. Empirically, TORNADO improves uncertainty calibration over representative generative baselines, including stochastic interpolants, diffusion models, and flow matching, while maintaining comparable predictive accuracy. Our results show that adapting stochasticity to the latent state, rather than prescribing it globally, is an effective mechanism for improving calibration in practical neural PDE forecasting settings. Project page: https://tornado-sde.github.io.
Independent verification organizations (IVOs) have emerged as a popular, bipartisan solution for facilitating high-quality independent evaluations of AI technologies. But momentum is outpacing design: there is emerging consensus on the necessary operating conditions for robust IVOs, but little discourse on institutional and policy specifications to ensure those conditions are met. IVOs are currently envisioned primarily as a mechanism operating at the AI frontier and model level to evaluate catastrophic risks. Little work has yet been done on how to establish or enforce an IVO ecosystem that is equipped to handle myriad risks (i.e., beyond catastrophic ones) and operate at the system and application level across the deployment and use stages of the AI lifecycle. This paper argues that IVO design should ensure the capacity for evaluating AI models, systems, and applications in context for both frontier companies and enterprise companies & AI-adopting organizations. Additionally, this paper provides early considerations to advance IVO discourse from normative to operative, and from existential to everyday risks. Drawing from existing institutional constructs in financial services, education, public health, and science, we outline key considerations on whom IVOs serve, what access they require, how they interact with external governing structures, how they are funded, and how they fit into the wider assurance ecosystem. Much of the structure and knowledge necessary to holistically design IVOs and their ecosystem does not yet exist, even as companies rush to ink agreements with their chosen evaluators.
Speckle is an inherent characteristic of synthetic aperture radar (SAR) imagery that complicates image interpretation and the extraction of reliable information. Although deep-learning-based despeckling methods have demonstrated promising performance, their applicability across different polarimetric configurations is often limited by the need for specialized architectures or separate training procedures. This paper presents a unified deep-learning framework for despeckling single-, dual-, and full-polarization SAR data using a single trained convolutional neural network. The proposed approach adopts a common nine-channel representation of the polarimetric covariance matrix and a polarization-aware masking strategy to accommodate different configurations without modifying the network architecture. Qualitative and quantitative comparison on three different scenes from different sensors demonstrate competitive performance across the considered experiments and support the feasibility of a unified, polarization-flexible approach to SAR despeckling across different polarimetric configurations.
Large language models (LLMs) offer a scalable approach for annotating mental health and crisis-related social media content, but their reliability in safety-critical tasks remains unclear. This study evaluates LLM-based crisis-severity annotation of veteran-related social media posts against annotations from eight Veteran Subject Matter Experts (SMEs). Using 149 tweets and five severity categories, we evaluate 15 open-source and 10 proprietary LLMs across eight prompt variants, examining annotation performance, inter-model agreement, prompt sensitivity, and high-severity crisis detection. Human reviewers showed higher ordinal than nominal agreement ({\alpha} = 0.494), indicating that disagreements often occurred between neighboring severity levels. In contrast, selected LLMs demonstrated substantially higher inter-model consistency, with ordinal agreement ({\alpha}) reaching 0.8336 under the baseline prompt. Proprietary models generally showed better agreement with expert annotations and greater stability than open-source models. GPT-5.6-Terra achieved the best overall performance, while the open-source Qwen2.5-7B showed the strongest performance in identifying Acute Crisis with Suicidal Ideation cases. However, higher model consistency did not necessarily correspond to agreement with human judgments or reliable detection of rare high-severity cases. Label-flipped demonstration experiments also revealed model-specific sensitivity to misleading in-context examples, while explanatory demonstrations did not consistently improve performance. Overall, LLMs show promise as scalable complements to expert annotation, but require careful evaluation before use in high-stakes mental health applications.
When several answers fit a riddle, exact-match scoring can penalise a valid response. CulturalCrosswords tests how answer constraints affect large language models (LLMs) solving cultural riddles, whose solutions draw on knowledge specific to a community. We constructed 53 crosswords from crowdsourced riddles in 42 languages and evaluated 11 LLM configurations on five versions of each clue set: plain batches, batches with lengths, batches with lengths and partial-answer hints, and crosswords requesting an answer list or a completed grid. On 962 clues with alternative answers, lengths improve agreement with the reference by 22.8 percentage points, compared with 8.4 points when accepted alternatives also receive credit. Part of the gain therefore appears to come from selecting a required answer form. Crossword input increases the share of all crossings with compatible answers by 9.7 points; requiring a completed grid reveals large differences between models. The benchmark distinguishes agreement with a reference from the ability to produce answers that fit together and occupy the correct cells.
Pretrained vision language action (VLA) models show strong zero-shot performance but they are rarely deployed directly and are often first post-trained on demonstrations from the target setting. Yet VLAs are typically compared by zero-shot or full-data fine-tuned success which reveals little about the data and compute required for downstream adaptation. We study adaptation efficiency for six VLAs ($\pi_0$, $\pi_{0.5}$, and GR00T N1-N1.7) on four LIBERO suites, varying both demonstrations and training steps. This reveals differences hidden by full-data fine-tuning: models which are within two points on the full LIBERO-Spatial dataset differ by up to 3$\times$ in success with 50 demonstrations. $\pi_{0.5}$ needs 2-5$\times$ fewer demonstrations than $\pi_0$ and is the most sample-efficient model at every budget, with the largest gap on the long-horizon suite. Within GR00T, N1.5 adapts more efficiently than N1.6 and N1.7 on three of four suites, despite their stronger zero-shot performance. We also study one pretraining factor behind these differences by freezing the VLM during finetuning: this severely degrades the $\pi$ models whose VLM is jointly trained with the action expert during pretraining, but barely affects GR00T models.
Video LLMs fail conspicuously at counting how many times an object crosses the screen, a failure attributed to missing object permanence. Using construction-grounded synthetic testbeds with validated labels, we show this attribution is incomplete: the count signal is present but unread - linear probes decode per-window presence at 93% while global count decoding collapses to a duration confound. Concurrent work reports the same representation-output dissociation for repetition counting; we show that in exit-and-reappear event counting it is repairable: two readout-directed repairs (episode-localization supervision and a discrete event-symbol side channel) raise same-object counting from 46.7% to 65.0% (oracle 66.7%) on a chance-33% benchmark across 7B-30B systems. Transfer to an independent factory costs nothing, and a measurement audit exposes five data-integrity failures that had obscured the boundary. A real-video audit (REALCOUNT, n=18, blind-audited) shows the boundary persists - best 55.6% at chance 33.3% - and symbol transfer fails zero-shot. A 1-epoch in-domain adaptation bench is negative (six levers at 28-50%), but 3 epochs reverse this for the symbol channel: with oracle timelines the reader reaches 94.4% sequence-disjoint (and 100% strict on the full set), while the deployed 2-epoch head reaches 61.1% in all three seeds (42-44), no better than a shuffled-symbol control (61.1% seed 42, 66.7% seed 43), and one clip above the unadapted reader (55.6%). The residual gap is detector evidence, not reader capacity: the same reader reaches 72.2-88.9% on GT timelines. The failure is repairable with informative timelines, but the deployed head remains the bottleneck; a frozen-feature boundary probe is weak (AUC 0.62 vs. 0.51 permutation) and non-transferring (0.77 within-clip vs. 0.62 across windows).