arXiv:2507.03502v3 Announce Type: replace
Abstract: Markov games with coupling constraints model constrained dynamical decision-making involving self-interested agents, where the feasibility of an individual agent's strategy depends on the joint strategies of the others. Such games arise in numerous real-world applications involving safety requirements and budget caps, for example, in environmental management, electricity markets, and transportation systems. In unconstrained dynamical decision-making, the correlated equilibrium has emerged as a desired solution concept due to its computational tractability and amenability to learning algorithms. Understanding how coupling constraints shape correlated equilibria is a crucial step towards computing solutions in constrained Markov games. In this paper, we formalize and characterize the notion of constrained correlated equilibria for Markov games, defined as feasible joint policies where any unilateral deviation is either unprofitable or infeasible. Building on this characterization, we further study existence conditions for constrained correlated equilibria. In particular, we provide a novel existence proof of such equilibria in Markov games with coupling constraints.
Science Journals
arXiv:2506.17803v3 Announce Type: replace
Abstract: Non-signaling correlations, which (strictly) include quantum correlations, provide a tractable path to explore the potential impact of quantum nonlocality on the capacity of classical communication networks. Motivated by a recent discovery that certain wireless network settings benefit significantly from non-signaling (NS) correlations, various generalizations are considered. First, it is shown that for a point to point discrete memoryless channel with a non-causal channel state information at the transmitter (CSIT), the NS-assisted Shannon capacity matches the classical (without NS assistance) capacity of the channel for the setting where the state is also made available to the receiver. The key insight is summarized as 'virtual signaling of CSIT via NS-assistance' and is supported by further results as follows. For a discrete memoryless 2-user broadcast channel (BC), the Shannon capacity region with NS-assistance available only between the transmitter and User 1, is found next. Consistent with the aforementioned key insight, the result matches the classical capacity region for the setting where the desired message of User 2 is made available in advance as side-information to User 1. The latter capacity region is known from a result of Kramer and Shamai. Next, for a semi-deterministic BC, the Shannon capacity region with full (tripartite) NS-assistance is shown to be the same as if only bipartite NS-assistance was available between the transmitter and the non-deterministic user. Bipartite NS-assistance between the transmitter and only the deterministic user, does not improve the capacity region relative to the corresponding classical setting. Finally, the analysis is extended to a K-user BC with full NS-assistance among all parties.
arXiv:2605.26821v1 Announce Type: cross
Abstract: The Lund plane offers a physics-motivated, hierarchical representation of QCD radiation within jets, while transformer-based taggers have reached state-of-the-art performance by learning directly from raw particle constituents and their pairwise relations. We investigate whether transformers implicitly capture hierarchical QCD structure from constituent-level inputs, or whether explicit physics representations remain complementary. To test this, we introduce PLuM, a multimodal architecture that projects particle constituents and Lund plane splittings into a shared latent space, processing both jointly with a unified transformer. Cross-attention allows the model to probe whether structured QCD information provides discriminating power beyond what particles alone encode. We observe systematic gains for top-quark and $\mathrm{H}\to\mathrm{b}\bar{\mathrm{b}}$ tagging, while finding no comparable improvement for $\mathrm{H}\to\mathrm{c}\bar{\mathrm{c}}$ or $\mathrm{H}\to 4\mathrm{q}$ topologies. This selective enhancement suggests that explicit hierarchical information about b-jet formation remains complementary to raw particle representations even in highly expressive architectures, while other topologies are already well-captured at constituent level. For high-impact LHC analyses such as Lorentz-boosted di-Higgs searches in the four $\mathrm{b}$ quark final state ($\mathrm{H}\mathrm{H}(4\mathrm{b})$), the gains are substantial: at a $25\%$ di-Higgs efficiency working point, PLuM achieves $25\%$ higher background rejection than the baseline. Our results indicate that physically structured representations of QCD radiation retain discriminating value in the transformer era, motivating further study into how different aspects of jet dynamics are encoded by deep learning algorithms.
arXiv:2605.26136v1 Announce Type: new
Abstract: Audio deepfakes have improved rapidly recently, yet their effect on human trust in real speech remains unstudied. We present the largest listening study on audio deepfake perception to date, collecting 35,532 judgments from 1,768 participants across 138 text-to-speech and voice conversion systems. Our central finding is a skepticism shift: compared to a 2021 baseline, human accuracy on fake samples barely changed (72.9% to 71.2%), but accuracy on real samples dropped from 72.7% to 64.1%. Participants are not worse at detecting synthesis artifacts; rather, they increasingly distrust authentic speech. Samples generated by commercial and autoregressive language model systems proved hardest to detect (61.3 - 65.9%), while those from traditional seq2seq and flow-matching models remain easier to spot (75.4 - 76.8%). An ML detector that served as a reference point maintained over 94.5% accuracy across all conditions. Our results suggest that the primary threat posed by modern deepfakes may not be mere deception, but the erosion of trust in genuine audio.
arXiv:2605.17774v2 Announce Type: replace
Abstract: Large language models are increasingly used as planning components in agentic systems, but current tool-use pipelines often require full tool schemas to be included in every prompt, creating substantial token overhead and limiting the practicality of smaller models. This paper investigates whether tool-use knowledge can be internalized into small language models through parameter-efficient fine-tuning, enabling structured planning without explicit tool descriptions at inference time. Using AssetOpsBench as the primary benchmark, we fine-tune Gemma 4 E4B and Qwen3-4B with 8-bit QLoRA on approximately 1,700 tool-use examples spanning tool knowledge, question-to-plan mappings, and execution-style traces. We evaluate the resulting models under description-free inference, where the prompt omits the tool catalog entirely. The fine-tuned models outperform an informed unfine-tuned baseline that receives full tool descriptions, reducing input length by 82.6\% while improving structural and LLM-judge planning scores. In the best Gemma run, the model achieves an AT-F1 of 0.65 and an overall judge score of 3.88, compared with 0.47 and 2.88 for the informed baseline. Qwen3-4B achieves a strong overall judge score of 3.78 while using 62\% less memory and running 2.5$\times$ faster than Gemma, though it also exhibits greater catastrophic forgetting on general multiple-choice benchmarks. Additional ablations show that LoRA rank controls a quality--retention trade-off, with $r=32$ maximizing planning quality and smaller ranks preserving more general knowledge. These results suggest that, for fixed tool catalogs, QLoRA fine-tuning can shift tool knowledge from prompt context into model weights, substantially reducing inference overhead while maintaining or improving tool-planning quality.
arXiv:2605.17617v2 Announce Type: replace
Abstract: Complex operational workflows coordinating personnel, tools, and information are central to system operations, yet end-to-end automation remains challenging due to extensive human input requirements and limited ability to adapt over time. We present GraphMind, a system that constructs, executes, and evolves action-centric workflow graphs with minimal human effort. The system operates in three phases. First, a scalable offline pipeline extracts structured workflow graphs from large volumes of human resolution traces, capturing problems, actions, and their causal relationships. Second, an online multi-agent traversal engine navigates the graph to dynamically construct and execute workflows, combining graph-guided retrieval with LLM-driven reasoning at each step. Third, Adaptive Traversal Reinforcement (ATR) reinforces successful traversal paths, enabling execution-informed graph adaptation. GraphMind has been deployed across four production cloud database services for incident investigation. Evaluated on 93 held-out incidents and validated via blind expert review, the system outperforms an Agentic Summary-RAG baseline in mitigation reach, hallucination rate, and diagnostic throughput while requiring 8x less retrieval context. The ATR layer reduces hallucination rate by 26%, demonstrating that workflow graphs can learn from execution feedback. A 12-week field study confirms practical value: 97% of scored conversations yield actionable results within interactive latency.
arXiv:2605.17036v3 Announce Type: replace
Abstract: This paper studies autonomous generative AI agents in multi-echelon supply chains using the MIT Beer Game. We identify four inference-time levers that shape performance: model selection, policies and guardrails, centralized data sharing, and prompt engineering. Model capability is the dominant factor: an out-of-the-box reasoning model exceeds human-level performance, and optimized reasoning models reduce costs by up to 67% relative to human teams. However, strong average performance masks substantial reliability risks. We introduce agent bullwhip: the amplification of run-to-run decision instability in autonomous multi-echelon systems. A central component is decision bullwhip, the portion of order variability generated by stochastic agent decisions rather than by changes in customer demand. We show that decision instability can amplify both across facilities at a fixed point in time and within the same facility over time, even when the demand path is held fixed. Repeated sampling, a natural test-time remedy, fails to meaningfully reduce this instability, suggesting that reliability requires changing the underlying decision policy rather than merely averaging over model outputs. To address this limitation, we propose a Group Relative Policy Optimization (GRPO)-based reinforcement-learning post-training framework that trains a shared base LLM using system-level supply-chain rewards. Post-training substantially reduces tail events, curtails agent bullwhip, and improves the reliability of autonomous supply-chain agents.
arXiv:2506.11253v2 Announce Type: replace
Abstract: Machine unlearning removes certain training data points and their influence from AI models (e.g., when a data owner revokes their consent to allow models to learn from the data). In this position paper, we propose to lift data-tracing machine unlearning to knowledge-tracing for foundation models (FMs). We support this position based on practical needs and insights from cognitive studies. Practically, tracing data cannot meet the diverse unlearning requests for FMs, which may be from regulators, enterprise users, product teams, etc., who have no access to FMs' massive training data. Instead, it is convenient for these parties to issue an unlearning request about the knowledge or capability FMs (should not) possess. Cognitively, knowledge-tracing unlearning aligns with how the human brain forgets more closely than tracing individual training data points does. We further discuss the nontrivial challenges in the knowledge-tracing machine unlearning paradigm. Finally, we provide a concrete case study about a vision-language FM to illustrate how an unlearner might instantiate the knowledge-tracing machine unlearning paradigm. Code is available at: https://1yuwen.github.io/Knowledge-Tracing-MU-Page.
arXiv:2506.09532v5 Announce Type: replace
Abstract: We present Athena-PRM, a multimodal process reward model (PRM) designed to evaluate the reward score for each step in solving complex reasoning problems. Developing high-performance PRMs typically demands significant time and financial investment, primarily due to the necessity for step-level annotations of reasoning steps. Conventional automated labeling methods, such as Monte Carlo estimation, often produce noisy labels and incur substantial computational costs. To efficiently generate high-quality process-labeled data, we propose leveraging prediction consistency between weak and strong completers as a criterion for identifying reliable process labels. Remarkably, Athena-PRM demonstrates outstanding effectiveness across various scenarios and benchmarks with just 5,000 samples. Furthermore, we also develop two effective strategies to improve the performance of PRMs: ORM initialization and up-sampling for negative data. We validate our approach in three specific scenarios: verification for test time scaling, direct evaluation of reasoning step correctness, and reward ranked fine-tuning. Our Athena-PRM consistently achieves superior performance across multiple benchmarks and scenarios. Notably, when using Qwen2.5-VL-7B as the policy model, Athena-PRM enhances performance by 10.2 points on WeMath and 7.1 points on MathVista for test time scaling. Furthermore, Athena-PRM sets the state-of-the-art (SoTA) results in VisualProcessBench and outperforms the previous SoTA by 3.9 F1-score, showcasing its robust capability to accurately assess the correctness of the reasoning step. Additionally, utilizing Athena-PRM as the reward model, we develop Athena-7B with reward ranked fine-tuning and outperforms baseline with a significant margin on five benchmarks.
arXiv:2509.26619v3 Announce Type: replace
Abstract: Many static benchmarks are beginning to saturate: as models rapidly improve, they achieve near-perfect scores on fixed test sets, leaving little headroom to expose genuine model weaknesses -- and even expert-curated challenge sets quickly saturate after hillclimbing. We present a fully automatic framework that searches the Internet at scale to construct challenging benchmarks without human curation. The key insight is to model the Internet as a vast space of topics and formalize the search as a multi-armed bandit problem, where each topic's difficulty is revealed only through expensive sample-and-evaluate queries. Our epsilon-greedy strategy identifies the most challenging topics while exploring only 6% of the search space -- a 100 times cost reduction over exhaustive evaluation. We validate on machine translation and knowledge question answering, confirming that discovered difficulty is robust across independent metrics (GEMBA-SQA and MetricX), languages, and models.
arXiv:2605.18110v3 Announce Type: replace
Abstract: We consider the problem of computing sample points in each connected component of a semi-algebraic set defined by the non-vanishing or the positivity of an n-variate polynomial of degree d, with rational coefficients of bit size bounded by $\tau$. Such a problem is a basic routine in effective real algebraic geometry, used in higher-level algorithms for solving polynomial systems over the reals and finds many applications in sciences. We design a probabilistic algorithm for solving this problem, which is based on reductions to different routines for solving zero-dimensional polynomial systems. It assumes that the input polynomial satisfies sufficiently generic properties (namely, smoothness of its defining hypersurface). This is done through the computations of critical points of well-chosen maps to capture the connected components of the semi-algebraic set under study. We derive a bit complexity estimate for the cost of this algorithm, which is, in terms of the B{\'e}zout bound d(d -1)^{n-1}, essentially cubic for obtaining parametrisations of the sought-for real points. Moreover, we also consider the case of obtaining rational approximations of those points, which are precise enough to lie in the same connected components as their exact counterparts, which yields a cost that is essentially quartic in the B{\'e}zout bound. In these complexity estimates, we take into account the degree structure of the input polynomial and its partial derivatives, allowing for a more refined bit complexity when the partial derivative of the input polynomial have degree lower than expected. We also analyse the probability of success of those algorithms. We report on practical experiments, benchmarking with random dense input polynomials as well as polynomials coming from applications, which were out of reach of the state-of-the-art implementations, and hence illustrate the practical efficiency of these new algorithms.
arXiv:2505.18728v2 Announce Type: replace
Abstract: The recent success of State-Space Models (SSMs) in sequence modeling has motivated their adaptation to graph learning, giving rise to Graph State-Space Models (GSSMs). However, existing GSSMs operate by applying SSM modules to sequences extracted from graphs, often compromising core properties such as permutation equivariance, message-passing compatibility, and computational efficiency. In this paper, we introduce a new perspective by embedding the key principles of modern SSM computation directly into the Message-Passing Neural Network framework, resulting in a unified methodology for both static and temporal graphs. Our approach, MP-SSM, enables efficient, permutation-equivariant, and long-range information propagation while preserving the architectural simplicity of message passing. Crucially, MP-SSM enables an exact sensitivity analysis, which we use to theoretically characterize information flow and evaluate issues like vanishing gradients and over-squashing in the deep regime. Furthermore, our design choices allow for a highly optimized parallel implementation akin to modern SSMs. We validate MP-SSM across a wide range of tasks, including node classification, graph property prediction, long-range benchmarks, and spatiotemporal forecasting, demonstrating both its versatility and strong empirical performance.
arXiv:2505.18603v2 Announce Type: replace
Abstract: Document understanding aims to perform question answering and information extraction over document images, where the visual content is highly information-dense and most queries rely on only a few relevant layout regions. However, existing methods either adopt a one-pass strategy that implicitly assumes all layouts are equally important, or focus excessively on small regions at the cost of losing critical layout information. To address these limitations, we introduce Doc-CoB (Chain-of-Boxes), a simple-yet-effective framework that integrates coarse-to-fine layout-aware visual reasoning into multimodal large language models. Instead of directly zooming into small regions, Doc-CoB progressively focuses on query-relevant layouts while preserving global document information. Specifically, it first selects key layout boxes and then focuses on them for further understanding with visual prompting. To support this paradigm, we introduce two reasoning tasks for box recognition and box reasoning, with an automatic pipeline that constructs 249k training samples with intermediate visual supervision. Extensive experiments on seven benchmarks with four popular models show that Doc-CoB significantly improves performance, demonstrating its effectiveness and wide applicability.
arXiv:2605.14664v2 Announce Type: replace
Abstract: Reference-guided video editing takes a source video, a text instruction, and a reference image as inputs, requiring the model to faithfully apply the instructed edits while preserving original motion and unedited content. Existing methods fall into two paradigms, each with inherent limitations: decoupled encoders suffer from modality gaps when processing instructions and visual content independently, while unified vision-language encoders lose fine-grained spatial details by relying solely on final-layer representations. We observe that VLM layers encode complementary information hierarchically -- early layers capture localized spatial details essential for precise editing, while deeper layers encode global semantics for instruction comprehension. Building on this insight, we present MiVE (Multiscale Vision-language features for reference-guided video Editing), a framework that repurposes VLMs as multiscale feature extractors. MiVE extracts hierarchical features from Qwen3-VL and integrates them into a unified self-attention Diffusion Transformer, eliminating the modality mismatch inherent in cross-attention designs. Experiments demonstrate that MiVE achieves state-of-the-art performance by ranking highest in human preference, outperforming both academic methods and commercial systems.
arXiv:2605.26626v1 Announce Type: new
Abstract: Stochastic model-predictive control (SMPC) has evolved to a powerful framework for the control of stochastic dynamical systems. SMPC utilizes a probabilistic uncertainty description to provide a systematic trade-off between the control objective and constraint satisfaction in a statistical sense. However, the majority of existing SMPC methods face challenges related to computational tractability due to the need for stochastic inference. Approaches that apply accurate inference are computationally demanding, which can lead to serious limitations in the implementability of these methods. Hence, in practice, the uncertainty propagation and the resulting distributions are typically approximated, e.g., by Gaussian distributions. These approximations promote computational efficiency, but are often too conservative, becoming a limiting factor in the representation of stochastic state evolution and the implied guarantees. To overcome this fundamental limitation of SMPC approaches, we propose a novel formulation based on the meta-state-space (MSS) representation of stochastic dynamical systems. The proposed MSS-based SMPC scheme offers a computationally efficient way to forward propagate the uncertainty with a flexible and highly accurate approximation of the probabilistic system description. With the presented method, the entire output probability density function can be directly shaped, which is unprecedented among existing SMPC techniques. Finally, we provide a detailed theoretical analysis and demonstrate the effectiveness of the proposed methodology via an extensive simulation study.
arXiv:2605.11867v2 Announce Type: replace
Abstract: Structural MRI-to-amyloid PET synthesis has been proposed as a non-invasive alternative for amyloid assessment in Alzheimer's disease (AD). However, reported performance of identical models varies widely across studies, and increasingly complex architectures have not led to consistent gains. This inconsistency is thought to be caused by a fundamental biological ambiguity: MRI captures neurodegeneration, while PET measures amyloid pathology - two processes that are often temporally decoupled in AD. As a result, similar MRI patterns may correspond to different amyloid states, creating ambiguous one-to-many mappings. MRI-to-amyloid PET synthesis may therefore be intrinsically ill-posed; however, this idea has yet to be tested scientifically. The aim of this work is to test this hypothesis through two controlled experiments. We first control the training distribution by stratifying paired MRI-PET data by amyloid and neurodegeneration status. Using two standard synthesis models under a controlled design, we show that biologically unambiguous mappings are learnable in isolation, but performance collapses when data ambiguity is introduced. This demonstrates that ambiguity in the data distribution, rather than architectural capacity, constrains performance. Second, we show that introducing orthogonal biological information in the form of plasma biomarkers resolves this ambiguity. When multimodal inputs are incorporated, performance improves and stability is restored. Together, these findings suggest that limited and inconsistent performance in MRI-to-amyloid PET synthesis is explained by intrinsic biological ambiguity, and that stable, meaningful progress requires multimodal integration rather than architectural complexity.
arXiv:2605.05774v2 Announce Type: replace
Abstract: Most production ERC-4337 Paymasters rely on Process-Oriented Abstraction (POA): a centralized off-chain server signs each sponsorship request, acting as a potential censorship bottleneck. We propose Asset-Oriented Abstraction (AOA), encapsulating payment capability in a persistent, user-owned on-chain asset -- the Gas Card -- rather than an off-chain signing process. Following the Design Science Research (DSR) methodology, we implement SuperPaymaster on Optimism Mainnet, anchoring sponsorship validity in on-chain Soulbound Token state and deterministic policy rules, removing the off-chain signer as a validity gate. We evaluate gas costs via single-UserOp ERC-20 transfers on Optimism Mainnet (n = 50 per system). In pure L2 execution gas (txGasUsed; actualGasUsed = txGasUsed + PVG), SuperPaymaster (167,830) is lower than both evaluated POA baselines: Alchemy Gas Manager (205,951) and Pimlico ERC-20 paymaster (328,937). It still pays a ~32,000-gas on-chain verification overhead versus Alchemy, but reduces gas by 49% versus Pimlico by replacing on-chain token liquidation with an internal balance update. In total billed gas, SuperPaymaster (286,818) exceeds Alchemy (257,299) due to higher bundler PVG overhead, not paymaster architecture. Code structural analysis and on-chain Mainnet evidence confirm that sponsorship validity requires no off-chain signing server: validatePaymasterUserOp reads only on-chain state. These findings suggest that AOA can mitigate the usability-decentralization-efficiency trade-offs in gas payment.
arXiv:2605.26628v1 Announce Type: new
Abstract: This report describes Tail-Aware HiFloat4, our submission to the low-bit text-to-video generation quantization challenge. Our method adapts the public ViDiT-Q post-training quantization pipeline to Wan2.2 under the HiFloat4 numerical format. We quantize the main linear layers in both Wan2.2 transformer modules with W4A4 HiFloat4 fake quantization, keep numerically sensitive boundary modules in high precision, and introduce an activation-tail-aware percentile calibration module for channel-mask construction. Together with compact PTQ-state restoration, this design reduces the influence of rare calibration outliers while keeping the runtime HiFloat4 arithmetic and sampling pipeline unchanged.
arXiv:2505.17163v2 Announce Type: replace
Abstract: Recent advancements in multimodal slow-thinking systems have demonstrated remarkable performance across various visual reasoning tasks. However, their capabilities in text-rich image reasoning tasks remain understudied due to the absence of a dedicated and systematic benchmark. To address this gap, we propose OCR-Reasoning, a novel benchmark designed to systematically assess Multimodal Large Language Models on text-rich image reasoning tasks. Specifically, OCR-Reasoning comprises 1,069 human-annotated examples spanning 6 core reasoning abilities and 18 practical reasoning tasks in text-rich visual scenarios. Unlike existing text-rich image understanding benchmarks that only provide a final answer, this benchmark additionally provides a detailed step-by-step reasoning process. This dual annotation enables the evaluation of both the models' final answers and their reasoning processes, thereby offering a holistic assessment of text-rich reasoning capabilities. By leveraging this benchmark, we conducted a comprehensive evaluation of the latest MLLMs. Our results demonstrate that even the most advanced MLLMs exhibit substantial difficulties in text-rich image reasoning tasks, with none achieving an accuracy above 50\% on our benchmark, indicating that the challenges of text-rich image reasoning are an urgent issue to be addressed. The benchmark and evaluation scripts are available at https://github.com/SCUT-DLVCLab/OCR-Reasoning.
arXiv:2605.26629v1 Announce Type: new
Abstract: Novel-view synthesis and 3D reconstruction from sparse posed images are central to robotics and AR/VR. Yet, feed-forward 3D Gaussian reconstruction fails under lowlight due to noise, color shifts, and unreliable correspondence. We propose DelowlightSplat, a lowlight-aware feed-forward Gaussian splatting framework for clean novel-view rendering. We build a controllable multi-view lowlight benchmark by degrading only context views while keeping target views clean. We introduce a lightweight Lowlight Adapter for residual enhancement to improve matchability, and couple it with cost-volume-based multi-view inference to directly predict clean 3D Gaussians. Experiments show that DelowlightSplat significantly outperforms previous feed-forward method and two-stage pipeline under lowlight conditions.
arXiv:2505.16942v2 Announce Type: replace
Abstract: Recent optical flow estimation methods often employ local cost sampling from a dense all-pairs correlation volume. This results in quadratic computational and memory complexity in the number of pixels. Although an alternative memory-efficient implementation with on-demand cost computation exists, this is significantly slower in practice and therefore many prior methods process images at downsampled resolutions, missing fine-grained details. To address this, we propose an algorithm for both memory and compute-efficient implementation of the all-pairs correlation volume sampling, still matching the exact mathematical operator as defined by RAFT. Our approach outperforms on-demand sampling by up to 92% while maintaining equally low memory usage, and performs at least on par with the default implementation with up to 99% lower memory usage. As cost sampling makes up a significant portion of the overall runtime, this can translate to up to 63% savings for the total end-to-end model inference on high-resolution inputs. Our evaluation of existing methods includes an 8K ultra-high-resolution dataset and an inference-time extension of the SEA-RAFT method. With this, we achieve state-of-the-art results at high resolutions both in accuracy and runtime.
arXiv:2505.13775v4 Announce Type: replace
Abstract: Recent impressive results from large reasoning models have been interpreted as a triumph of Chain of Thought (CoT), and especially of the process of training on CoTs sampled from base LLMs in order to help find new reasoning patterns. While these traces certainly seem to help model performance, it is not clear how they influence it, with some works ascribing semantics to them and others cautioning against relying on them as transparent and faithful proxies of the model's internal computational process. To systematically investigate the role of end-user semantics of derivational traces, we set up a controlled study where we train transformer models from scratch on formally verifiable reasoning traces and the solutions they lead to. We notice that, despite gains over the solution-only baseline, models trained on entirely correct traces can still produce invalid reasoning traces even when arriving at correct solutions. More interestingly, our experiments also show that models trained on corrupted traces, whose intermediate reasoning steps bear no relation to the problem they accompany, perform similarly to those trained on correct ones, and even generalize better on out-of-distribution tasks. We also study the effect of GRPO-based RL post-training on trace validity, noting that while solution accuracy increases, this is not accompanied by improvements in trace validity. Finally, we examine whether reasoning-trace length reflects inference-time scaling and find that trace length is largely agnostic to the underlying computational complexity of the problem being solved. These results challenge the assumption that intermediate tokens or ``Chains of Thought'' reflect or induce predictable reasoning behaviors and caution against anthropomorphizing such outputs or over-interpreting them (despite their mostly seemingly forms) as evidence of human-like or algorithmic behaviors in language models.
arXiv:2505.02974v3 Announce Type: replace
Abstract: Machine learning-based surrogate models have emerged as a powerful tool to accelerate simulation-driven scientific workflows, but their adoption is limited by the lack of large-scale, diverse, and standardized datasets for physics-based simulations. Existing benchmarks often focus on narrow domains or rely on simplified data models, and fail to capture the heterogeneity arising from variable geometries, meshes, and topologies, which is critical for assessing generalization in realistic settings. We introduce PLAID (Physics-Learning AI Data model), a unified and extensible data layer for heterogeneous physics simulations. It preserves the full complexity of simulation data while enabling efficient and scalable machine learning workflows, together with a library for dataset construction and manipulation~(\href{https://github.com/PLAID-lib/plaid}{github.com/PLAID-lib/plaid}). We release six datasets covering structural mechanics and computational fluid dynamics, designed to reflect realistic industrial scenarios and provide standardized benchmarks. The framework includes reproducible evaluation protocols and is integrated with Hugging Face to enable open, community-driven benchmarking with active user participation (\href{https://huggingface.co/PLAIDcompetitions}{huggingface.co/PLAIDcompetitions}).
arXiv:2602.19206v3 Announce Type: replace
Abstract: Zero-shot 3D Anomaly Detection is an emerging task that aims to detect anomalies in a target dataset without any target training data, which is particularly important in scenarios constrained by sample scarcity and data privacy concerns. While current methods adapt CLIP by projecting 3D point clouds into 2D representations, they face challenges. The projection inherently loses some geometric details, and the reliance on a single 2D modality provides an incomplete visual understanding, limiting their ability to detect diverse anomaly types. To address these limitations, we propose the Geometry-Aware Prompt and Synergistic View Representation Learning (GS-CLIP) framework, which enables the model to identify geometric anomalies through a two-stage learning process. In stage 1, we dynamically generate text prompts embedded with 3D geometric priors. These prompts contain global shape context and local defect information distilled by our Geometric Defect Distillation Module (GDDM). In stage 2, we introduce Synergistic View Representation Learning architecture that processes rendered and depth images in parallel. A Synergistic Refinement Module (SRM) subsequently fuses the features of both streams, capitalizing on their complementary strengths. Comprehensive experimental results on four large-scale public datasets show that GS-CLIP achieves superior performance in detection. Code can be available at https://github.com/zhushengxinyue/GS-CLIP.
arXiv:2602.17605v2 Announce Type: replace
Abstract: In environmental monitoring, data collection is often costly, sparse, and shaped by urgent public-health needs. This is particularly true for cancer-causing PFAS (Per- and polyfluoroalkyl substances) contamination, where discussions with domain experts and environmental organizations highlight the need to strategically identify high-risk, under-observed regions under tight sampling budgets. More broadly, similar challenges arise in disaster response and public health settings, where dynamic environments make it essential to efficiently uncover hidden targets from limited ground truth. Yet sparse and biased geospatial labels limit the applicability of existing learning-based methods, such as reinforcement learning. To address this, we propose a unified geospatial discovery framework that integrates active learning, online meta-learning, and concept-guided reasoning. Our approach introduces two key innovations built on a shared notion of *concept relevance*, capturing how domain-specific factors influence target presence: a *concept-weighted uncertainty sampling strategy*, where uncertainty is modulated by learned relevance from readily available concepts such as land cover and source proximity; and a *relevance-aware meta-batch formation strategy* that promotes semantic diversity during online-meta updates, improving generalization in dynamic environments. We evaluate our framework on PFAS contamination discovery as a real-world inspired environmental monitoring task, demonstrating robust target discovery under limited data and changing conditions.