arXiv:2606.16142v1 Announce Type: new Abstract: The Landau--Lifshitz--Gilbert (LLG) equation is a fundamental model in micromagnetics for the magnetisation dynamics of ferromagnets at temperatures well below the Curie temperature. In chiral magnetic materials, the bulk Dzyaloshinskii--Moriya interaction (DMI) contributes a first-order term to the effective field, induces a natural chiral boundary condition, and plays an important role in the dynamics of skyrmions. In this paper, we analyse a mass-lumped midpoint finite element scheme for the LLG equation with exchange interaction and bulk DMI. The fully discrete scheme preserves the nodal unit-length constraint exactly and satisfies a discrete energy law. Under suitable regularity assumptions on the exact solution, we prove an optimal-order convergence estimate in the energy norm, first order in space and second order in time. To our knowledge, this is the first such a priori error estimate for the mass-lumped midpoint finite element method, even in the exchange-only case. Since the midpoint scheme is nonlinear, its practical implementation requires an algebraic solver that is compatible with the geometric structure of the method. To this end, we propose a structure-preserving fixed-point iteration which preserves the nodal unit-length constraint at every iterate. We further prove that, under an appropriate residual stopping criterion, the resulting inexact scheme retains the same convergence rate as the ideal midpoint scheme, up to the contribution of the solver tolerance. Numerical experiments support the theoretical results and illustrate the robustness and structure-preserving properties of the method.
Science Journals
arXiv:2606.13710v2 Announce Type: replace Abstract: Deep research and agent evolution serve as de-facto tasks for AI agents in real-world applications toward artificial general intelligence. The former enables autonomous retrieval and integration of information in open-ended environments to tackle open-ended research tasks, yet it is constrained by the static parametric deep research capabilities of agent systems. The latter allows agents to autonomously interact with the environment to gain experiences that evolve model capabilities. However, its effectiveness has been widely validated only on verifiable tasks with standard answers, leaving a gap with open-ended research tasks. To bridge these two critical tasks, we propose the Hybrid Open-Ended Tri-Evolution (HOTE) framework, which leverages hybrid-mode reinforcement learning to facilitate the collaborative evolution of a proposer, solver and judge based on web-scale knowledge, moving toward autonomous evolving agents in open-ended tasks and environments. Extensive experiments on three long-form deep research benchmarks demonstrate that the 8B model trained via HOTE surpasses the strongest static open 8-32B models as well as those trained by state-of-the-art deep research training methods with less time overhead, and further verify that the evolution of all three modules in HOTE is indispensable.
arXiv:2606.16526v1 Announce Type: new Abstract: The Faraday effect is a hallmark of nonreciprocal light-matter interactions and traditionally requires magnetic bias or intrinsically chiral media. Here we introduce a temporal chiral metamaterial in which an effective chiral response is generated entirely by Floquet modulation, without magnetic fields or structurally chiral constituents. The medium is realized by periodically rotating the principal axes of the permittivity and permeability tensors in time. Using a nonlocal temporal effective medium theory derived from Hamiltonian homogenization, we show that the resulting chiral parameter is an odd function of the wavevector, giving rise to intrinsic nonreciprocity despite Onsager-symmetric constitutive relations. This Floquet-induced chirality produces a temporal Faraday effect, in which the polarization plane of a linearly polarized wave rotates continuously in time. The direction and magnitude of the rotation are programmable through the modulation sequence and remain invariant under both spatial and temporal reversal. Our work establishes Floquet-induced chirality as a fundamentally new mechanism for nonreciprocal light control and opens a route to reconfigurable polarization manipulation in time-modulated photonic systems.
arXiv:2606.16960v1 Announce Type: new Abstract: Modern autonomous driving depends on accurate metric 3D understanding for perception, reconstruction, and planning, which in turn requires reliable multi-camera depth prediction. However, the outward-facing nature of vehicle-mounted surround-view camera rigs inherently limits visual overlap across views, challenging the correspondence-based assumptions that underpin conventional multi-view geometry. To bridge this gap, we present SurroundNEXO, named after the Spanish word nexo for a geometric link, a low-overlap multi-camera metric depth framework that grounds cross-view reasoning in ego-centric geometry rather than dense visual correspondences. Instead of directly enforcing early global fusion, SurroundNEXO first assigns image tokens globally comparable ego-frame viewing directions through Ego-Ray Positional Encoding, then uses sparse LiDAR measurements as metric anchors to propagate absolute scale cues, and finally expands feature interaction progressively from view-local modeling to decomposed spatio-temporal reasoning and global integration. This design enables metric-scale depth prediction with improved spatial consistency across weakly overlapping cameras. Across low-overlap autonomous driving benchmarks, including NuScenes, Waymo and DDAD, SurroundNEXO reduces single-view error by 33.2%, improves cross-view consistency by 10.5%, and enhances metric reconstruction quality by 25.6% compared with SOTA methods. It further remains robust under extremely sparse depth prompts and exhibits strong zero-shot generalization to unseen camera layouts.
arXiv:2602.04525v2 Announce Type: replace Abstract: Rapid urban expansion has fueled the growth of informal settlements in major cities of low- and middle-income countries, with Lahore and Karachi in Pakistan and Mumbai in India serving as prominent examples. However, large-scale mapping of these settlements is severely constrained not only by the scarcity of annotations but by inherent data quality challenges, specifically high spectral ambiguity between formal and informal structures and significant annotation noise. We address this by introducing a benchmark dataset for Lahore, constructed from scratch, along with companion datasets for Karachi and Mumbai, which were derived from verified administrative boundaries, totaling approximately 900 $\text{km}^\text{2}$ of urban area. This collection is supplemented by four cities from prior literature across Sub-Saharan Africa and Latin America, with comprehensive data quality assessments provided for each city. We also propose a semi-supervised segmentation framework designed to mitigate the class imbalance and distribution mismatch inherent in standard semi-supervised learning pipelines. Our method integrates a Class-Aware Adaptive Thresholding mechanism that dynamically adjusts confidence thresholds to prevent minority class suppression, and a DINOv2-based unlabeled pool filter that removes out-of-distribution tiles prior to training to reduce covariate shift. Extensive experiments across seven cities spanning three continents, repeated over five random seeds, demonstrate gains of up to +5.9 pp mIoU over state-of-the-art semi-supervised baselines, with both components being architecture-agnostic and adding no inference overhead.
arXiv:2512.10903v2 Announce Type: replace Abstract: Circuit discovery aims to identify minimal subnetworks that are responsible for specific behaviors in large language models (LLMs). Existing approaches primarily rely on iterative edge pruning, which is computationally expensive and limited to coarse-grained units such as attention heads or MLP blocks, overlooking finer structures like individual neurons. We propose a node-level pruning framework for circuit discovery that addresses both scalability and granularity limitations. Our method introduces learnable masks across multiple levels of granularity, from entire blocks to individual neurons, within a unified optimization objective. Granularity-specific sparsity penalties guide the pruning process, allowing a comprehensive compression in a single fine-tuning run. Empirically, our approach identifies circuits that are smaller in nodes than those discovered by prior methods; moreover, we demonstrate that many neurons deemed important by coarse methods are actually irrelevant, while still maintaining task performance. Furthermore, our method has a significantly lower memory footprint, 5-10x, as it does not require keeping intermediate activations in the memory to work.
arXiv:2512.12313v3 Announce Type: replace Abstract: Software supply chain attacks on the npm ecosystem have grown increasingly sophisticated, exploiting obfuscation and complex logic to evade detection. Large Language Models (LLMs) offer strong semantic understanding of code but face practical constraints: limited context windows and high inference costs make full-package analysis infeasible, while naive token-based splitting fragments semantic context and degrades accuracy. This paper introduces an LLM-based framework for malicious npm package detection built on code-slicing techniques. We propose an adaptation of taint-based slicing for the npm ecosystem, guided by a curated inventory of JavaScript-specific sensitive APIs, to isolate security-relevant data flows from benign boilerplate. The approach reduces the mean input token count by 99.75% and the median by 93.7% while preserving critical malicious behaviors. Packages relying on dynamic code generation or obfuscation yield empty slices under static analysis and require deobfuscation preprocessing, a limitation we explicitly discuss. The framework is evaluated on a dataset of more than 7000 malicious and benign npm packages using DeepSeek-Coder6.7B. On the 2537 packages amenable to static taint analysis, taint-based slicing achieves 87.04% detection accuracy, outperforming both a naive token-splitting baseline at 75.41% and a CFG-only static slicing approach at 75.65%. These results demonstrate that semantically targeted input representations improve LLM-based detection performance beyond what is achievable through simple input-size reduction, providing an effective and computationally practical defense against evolving open-source supply-chain threats.
arXiv:2512.12737v2 Announce Type: replace Abstract: Decentralized federated learning (DFL) enables collaborative model training without a central server, but converges slowly under statistical heterogeneity. Recent work has shown that neural tangent kernel (NTK) methods achieve faster convergence than gradient-based updates in DFL, while momentum has proven effective for accelerating gradient-based FL. However, applying momentum to NTK updates can destabilize training under heterogeneous data. We propose SPARK, which addresses this instability with a stage-wise annealed soft-label regularizer evaluated on neighborhood-aggregated data, so that momentum can accelerate NTK updates stably. Under high heterogeneity, SPARK converges about 3$\times$ faster than baselines and lowers the total communication to a target accuracy by up to about 70\%, and it attains higher accuracy across heterogeneity levels. We further study random projection as an optional Jacobian-compression strategy for bandwidth-constrained settings. We validate the approach across multiple datasets, network topologies, and heterogeneity levels.
arXiv:2606.15367v1 Announce Type: new Abstract: Deep research agents aim to solve complex knowledge-intensive tasks through long-horizon planning, evidence gathering, reasoning, and report generation. While recent progress in search agents has demonstrated strong capabilities in information retrieval and answer verification, most existing training datasets remain search-centric, focusing primarily on closed-ended question answering and information localization. As a result, they mainly train information-seeking behavior while providing limited coverage of key deep research capabilities, including evidence integration, knowledge synthesis, planning, file understanding, and structured report generation. In this work, we propose a unified trajectory construction paradigm for deep research agents that combines closed-ended QA and open-ended exploration. The proposed framework consists of graph-grounded task formulation, agentic trajectory rollout, and multi-dimensional trajectory verification, enabling scalable synthesis of high-quality agentic trajectories spanning long-chain complex reasoning, deep research instruction following, report writing, file understanding and generation, and skills usage. Compared with existing search-oriented datasets, our synthesized trajectories place greater emphasis on knowledge synthesis, complex reasoning, and planning. S1-DeepResearch-32B achieves state-of-the-art performance among open-source models of comparable scale across 20 benchmarks spanning five capability dimensions, including complex reasoning, instruction following, report generation, file understanding, and skills usage. On several challenging deep research benchmarks, it approaches the performance of leading proprietary frontier models. These results highlight the importance of jointly modeling information acquisition, knowledge synthesis, and planning-oriented agent behaviors for building effective deep research agents.
arXiv:2606.16143v1 Announce Type: new Abstract: Probabilistic models (PMs) are essential in advancing machine learning capabilities, particularly in safety-critical applications involving reasoning and decision-making. Among the methods employed for inference in these models, sampling-based Markov Chain Monte Carlo (MCMC) techniques are widely used. However, MCMC methods come with significant computational costs and are inherently challenging to parallelize, resulting in inefficient execution on conventional CPU/GPU platforms. To overcome these challenges, this paper presents AIA, a multi-core RISC-V System-on-Chip (SoC) design fabricated using Intel's 16 nm process technology. Our Approximate Inference Accelerator (AIA) is specifically designed to empower edge devices with robust decision-making and reasoning abilities. The AIA architecture incorporates a RISC-V host processor to manage chip-to-chip data communication and a 2D mesh of 16 custom versatile RISC-V cores optimized for high-efficiency approximate inference. Each core features (i) custom instructions and datapath blocks for non-normalized Knuth-Yao (KY) sampling, as well as for the interpolation of non-linear functions (e.g., logarithmic, exponential), and (ii) direct data access to the register file of each neighboring core, to reduce the data movement costs of frequent data exchanges between nearby cores. To further capitalize on the parallelism potential in MCMC algorithms, we developed a specialized compile chain that enables efficient spatial mapping and scheduling across the cores.
arXiv:2606.12127v2 Announce Type: replace Abstract: We report the discovery of a previously unrecognized electrostatic hose instability in electron-beam-sustained plasmas, driven by the coupling between the electron beam centroid and the plasma generated via the beam-impact ionization. Unlike the conventional hose instability of relativistic beams propagating in underdense plasmas, this instability requires only ionization-capable electron beams readily produced by common emission processes and sheath acceleration, indicating broad relevance across various discharges. A linear theory is developed to predict the hosing frequency and growth rate, and particle-in-cell/Monte Carlo simulations confirm both the onset of instability and the theoretical predictions.
arXiv:2606.16148v1 Announce Type: new Abstract: Probabilistic graphical models (PMs) are popular to empower machine learning with the ability of reasoning and decision-making. To perform approximate inference in PMs, sampling-based Markov Chain Monte Carlo (MCMC) algorithms are commonly employed. Unfortunately, MCMC is compute-intensive and hard to run in parallel, resulting in inefficient execution on modern CPU/GPU platforms. This paper proposes \name{}, an Approximate Inference Accelerator designed to empower decision-making and reasoning at the edge. \name{} consists of a RISC-V host, and a 2D mesh of 16 customized RISC-V cores optimized to efficiently support PM inference, each featuring (i) a novel non-normalized Knuth-Yao sampler and interpolation unit; and (ii) core-to-core direct data access via the register file, which provides solutions for compute-intensive operations. To fully exploit the parallel potential of Markov Chain Monte Carlo (MCMC) algorithms, a customized compiler chain has been developed for effective spatial mapping and scheduling on the chip. \name{} can generate 1277 MSample/s at 0.9V and 20 GSamples/s/W at 0.7V which is up to 2$\times$ faster and 1.45x more energy efficient compared to the previous state-of-the-art Markov Random Field (MRF) accelerator. We further map Bayesian Networks benchmark onto \name{} to show the flexibility of our design.
arXiv:2606.15370v1 Announce Type: new Abstract: This work demonstrates a full reproduction and extension of MNet, a hybrid 2D/3D convolutional network designed for anisotropic medical image segmentation. The original architecture was re-implemented within the nnU-Net framework to verify its reported performance and robustness to variable voxel spacing, known as anisotropy. Experiments were conducted on PROMISE prostate MRI and a controlled subset of LiTS liver CT under matched preprocessing and compute constraints. The reproduced MNet achieved a Dice similarity coefficient (DSC) of 89.0 +/- 0.9% on PROMISE, within 0.8% of the published result, and 94.3 +/- 1.9% / 54.6 +/- 3.1% for liver and tumor segmentation on LiTS, respectively. Two lightweight extensions were further introduced: (1) a learned Fusion Gating mechanism enabling adaptive 2D-3D feature blending, and (2) a VMamba state-space module for efficient long-range depth modelling. The Spatial Gating variant improved DSC by +0.8% with less than 3% inference overhead, while VMamba improved performance consistency, reducing PROMISE Dice variation to +/- 0.7% and achieving the strongest LiTS liver performance at 95.8% Dice. Both extensions preserved MNet robustness to anisotropy, with delta Dice = 1.5% across 1-4 mm voxel spacing. Overall, the study confirms MNet reproducibility and demonstrates that adaptive fusion and state-space modelling have the potential to further strengthen segmentation reliability under anisotropic conditions. However, further tests are required to provide definitive conclusions.
arXiv:2606.14369v2 Announce Type: replace Abstract: Building upon the semi-implicit multilevel spectral deferred correction (SI-MLSDC) method introduced by Pfister and Stiller [38], this work presents a 1D space-time adaptive high-order method combining discontinuous Galerkin spectral element discretizations with multilevel spectral deferred corrections. The proposed approach enables genuine arbitrary-order accuracy in space and time while dynamically balancing spatial and temporal discretization errors to reduce computational cost. A key contribution is the development of a novel temporal error estimator that in combination with a spectral error estimator in space provides a reliable basis for adaptive refinement decisions. The error estimator is compared with two alternative refinement criteria to assess their impact on accuracy, computational efficiency and their suitability for complex problems. The performance of the adaptive method is demonstrated for nonlinear conservation laws ranging from Burgers' equation to the Euler equations. Numerical results show substantial runtime reductions while maintaining the desired accuracy. In particular, significant computational savings are achieved for Burgers' equation, and challenging benchmark problems such as the Shu-Osher shock-fluctuation benchmark. These results demonstrate the potential of adaptive SI-MLSDC methods for efficient high-order space-time adaptive simulations of complex flow problems.
arXiv:2606.14398v2 Announce Type: replace Abstract: Mixture-of-experts (MoE) layers enable the scaling of transformer models while keeping the inference compute fixed. While task-expert specialization has been observed in empirical studies of frontier MoE transformer models, existing theoretical work analyzes this using continuous mixture models that cannot be used to model natural language effectively. An important open question is to \textit{theoretically explain task-expert specialization in transformer MoE models using discrete models of language}. To address this, we represent structured knowledge via syntactic templates and finite key-value dictionaries, and prove formally that a single-layer MoE transformer can encode knowledge by using experts that specialize in the corresponding tasks. Our construction shows how queries are routed to unique, task-specific experts whose size depends solely on the intrinsic complexity of the given task (i.e. the combined size of its syntactic templates and factual dictionary). Our construction provides a theoretical support for empirical results on localized knowledge circuits in MoE models. We support our theoretical findings with experiments evaluating model performance under varying MoE loss functions.
arXiv:2606.13003v2 Announce Type: replace Abstract: Prevailing wisdom posits that Multi-Agent Systems (MAS) are superior to Single-Agent Systems (SAS), citing advantages like context protection, parallel processing and distributed decision-making. However, empirical support for this claim relies primarily on comparisons with SAS baselines using benchmarks that prioritize isolated reasoning tasks, which do not adequately assess these advantages. Focusing on automatically generated MAS that are designed for enhanced generalizability over manually-designed counterparts, we perform a rigorous, systematic evaluation against SAS, specifically Chain-of-Thought with Self-Consistency (CoT-SC). Across traditional reasoning datasets and tasks with interactive multi-step workflows (e.g., BrowseComp-Plus), we demonstrate that automatic MAS consistently underperform CoT-SC despite being up to 10x more expensive. To isolate these failures from limitations inherent to task structure, we introduce a diagnostic synthetic dataset tailored for MAS featuring explicit task decomposition, context separation and parallelization potential. We show that expert-architected MAS consistently outperforms automatically generated architectures in both raw performance and cost-efficiency on this dataset, demonstrating that existing evaluation frameworks mask critical architectural gaps and inefficiencies of complex MAS by failing to account for the marginal utility of increased computational cost. Critically, systematic deconstruction of the generated MAS architectures reveals that current automated design paradigms produce architectural bloat that prioritizes superficial complexity which does not translate into functional utility, exposing a fundamental misalignment with multi-agent principles.
arXiv:2603.26555v3 Announce Type: replace-cross Abstract: Mixtures of weakly and strongly hydrated anions induce non-additive changes in the LCST of thermoresponsive polymers such as PNIPAM and PEO. Large-scale atomistic simulations of PNIPAM--NaI--Na$_{2}$SO$_{4}$ mixtures have shown that these effects arise from the interplay between favorable PNIPAM--iodide interactions and depletion of strongly hydrated sulfate ions. Here, we investigate whether chemically specific polymer--anion interactions are necessary to reproduce such behavior. To this end, we study the coil--globule transition of a generic uncharged linear polymer with non-specific polymer--water and polymer--ion van der Waals interactions in atomistic aqueous solutions of single and mixed salts. Simulations are performed at fixed concentrations of the strongly hydrated salt, Na$_{2}$SO$_{4}$, and increasing concentrations of the weakly hydrated salts, NaSCN and NaI. The generic polymer qualitatively reproduces experimentally observed trends in pure NaSCN and Na$_{2}$SO$_{4}$ solutions, as well as the non-additive behavior in mixed salt solutions. In particular, the model captures the mutually reinforcing preferential accumulation of the weakly hydrated SCN$^{-}$ ions and depletion of the strongly hydrated SO$_{4}^{2-}$ ions near the polymer that underlies the non-additive behavior. This mutual enhancement correlates with partitioning of sodium ions from the counterion cloud of SCN$^{-}$ ions to that of SO$_{4}^{2-}$ ions and is consistent with atomistic simulations of PNIPAM solutions. The model also reproduces the effects of background salt concentration and weakly hydrated anion identity on the non-additive behavior. These results demonstrate that non-specific polymer--ion and polymer-water interactions are sufficient to reproduce non-additive salt effects, suggesting a dominant role of bulk ion--ion and ion--water interactions.
arXiv:2604.02901v2 Announce Type: replace-cross Abstract: This study focuses on energy conversion related to tearing-driven magnetic reconnection in the context of weakly collisional astrophysical plasmas. We present results from a two-dimensional hybrid particle-in-cell simulation employing novel periodic conditions with a topology akin to the M\"obius strip, which double the computation efficiency as compared to regular periodic conditions. Evaluation of the ion electric work rate ($\mathbf{j}_i \cdot \mathbf{E}$) and pressure strain interaction ($\mathbf{P}_i : \mathbf{\nabla u}_i)$ shows that most of the energy conversion occurs during the nonlinear phase of the instability, where magnetic energy is transferred towards ion kinetic energy (bulk outflows) and internal energy (heating). These energy conversion rates are of the same order but inhomogeneous. Heating predominantly occurs within the magnetic islands, while near the X-points, nearly the same amount of magnetic energy is transferred to bulk plasma flow and heating. The reconnected plasma moreover exhibits an ion temperature higher parallel than perpendicular to the local magnetic field $\mathbf{B}$. This temperature anisotropy is sustained by the islands' contraction, but eventually gets regulated by kinetic firehose instabilities (parallel and oblique), and/or firehose-like processes, whose main effect is to redistribute the internal energy from the parallel to the perpendicular direction.
arXiv:2605.22212v3 Announce Type: replace-cross Abstract: We prove that the three-dimensional incompressible Navier-Stokes equations with the deformation Laplacian on hyperbolic 3-space $\HH^3$ admit a unique global mild solution for sufficiently small initial data in $L^3(\HH^3)$, and that this solution decays exponentially to zero. The exponential decay rate is $\mu\lambda_\Def^{(3)}$, where $\mu$ is the kinematic viscosity and $\lambda_\Def^{(3)} = 26/9$ is the effective spectral gap of the deformation Laplacian in $L^3$. On flat $\R^3$, the corresponding Kato-type result gives only algebraic decay. The exponential stability is a macroscopic consequence of the spectral gap provided by negative curvature. We also show that the $L^2$ norm is supercritical on $\HH^3$ (as on $\R^3$), with the obstruction arising from the local ultraviolet scaling of the heat kernel, which is insensitive to global geometry. The boundary between what curvature can and cannot improve is located exactly: the Fujita-Kato integral has a scaling exponent $1/2 - 3/(2p)$ that depends only on the integrability of the initial data, not on the geometry of the manifold. For $p \geq 3$ (the Kato critical space), the integral is bounded and the spectral gap contributes exponential time decay. For $p < 3$, the integral diverges at $t = 0$ (and strictly diverges for all $t>0$ when $p \le 3/2$) regardless of the curvature.
arXiv:2606.07086v2 Announce Type: replace Abstract: Deep neural networks (DNNs) excel in computer vision tasks given large annotated datasets. In real-world applications, however, labels are often corrupted by ambiguity, human error, or dynamic environments. Over-parameterized DNNs easily memorize these noisy labels during training, degrading model accuracy and generalization. Existing data-cleaning and sample-selection strategies often rely on manually specified thresholds, prior knowledge of the noise ratio, or a single metric (either learning dynamics or geometric structure), making them unstable in complex data regimes. This paper proposes a self-adaptive data-cleaning framework that integrates local, global, and learning dynamics cues for robust noisy-label detection. Samples are mapped into a unified low-dimensional feature space through a modular feature concatenation paradigm. We provide two instantiations: a 2D metric integrating class-adaptive KNN-based local disagreement with k-means-based global centroid distance, and a 3D multi-metric that additionally incorporates a z-normalized score. Unlike conventional 1D Gaussian Mixture Models applied to a single scalar metric, our framework performs multi-metric clustering on the feature space to adaptively partition samples into clean-dominant and noise-dominant components without requiring manual thresholds or noise priors. Experiments on CIFAR-10, MNIST, and ImageNet-100 with 5% to 40% symmetric label noise show high recall across settings, including near-perfect recall (>=98%) on ImageNet-100 at 40% noise. Subsequent training yields accuracy gains across evaluated settings, especially under severe corruption on ImageNet-100. These findings suggest that multi-metric integration provides a threshold-free, practical, and low-tuning strategy for noisy label detection.
arXiv:2606.15385v1 Announce Type: new Abstract: Reward hacking, where AI systems exploit misspecified objectives to achieve high reward without satisfying intended goals, remains a central challenge in AI safety. Yet most known instances have been discovered post hoc in frontier systems where controlled study is impractical. We adapt the AI Safety Gridworlds framework into a text-based evaluation suite that reformulates classic reinforcement learning safety tasks for language-based agents. Across frontier and mid-scale models, we find that specification gaming emerges zero-shot: models systematically achieve high observed reward while underperforming on hidden safety objectives, and even apparently safe behaviors can reflect misunderstanding rather than principled safety. Reinforcement learning does not correct these failures: direct reward optimization widens the gap between observed and hidden reward, as the model's initial competence causes it to lock into locally rewarding strategies before discovering safer alternatives. This pattern persists across model scales (1.5B--14B) and is not resolved by finer credit assignment, exploration prompts, or entropy regularization. Our results show that reward hacking arises naturally when optimizing proxy objectives with capable language model agents and resists standard mitigations, suggesting that proxy-reward failures in agentic settings may require approaches beyond standard exploration and credit-assignment fixes. To facilitate reproducibility, the code for this work is available at \href{https://github.com/asparius/verl-agent-safety}{our public repository}.
arXiv:2606.15482v1 Announce Type: cross Abstract: Ricci flow is a curvature-guided diffusion process that deforms space by shrinking regions of high positive curvature and expanding those with negative curvature. Similarly, discrete Ricci flow on weighted graphs modifies edge weights by shrinking edges with positive Ricci curvature and stretching those with negative Ricci curvature, effectively increasing the separation between clusters. Inspired by these two cornerstone works, we propose a geometry-based RAG reranker enhancement procedure called Ricci-Filtration. By modeling the input query and initial retrieved chunks as a network, where the input query and chunks serve as nodes and embedding-based pairwise relations define an initial graph, Ricci-Filtration leverages discrete curvature and Ricci flow to evaluate the structural importance of each chunk with respect to the user query. The system first filters the initial chunks based on their geometric curvature relative to the query; then, a reranker processes the remaining chunks to enhance generative performance. We theoretically prove that normalized discrete Ricci flow can detect community structures by identifying distinct asymptotic behaviors in edge weights. This supports the removal of ``noisy'' document chunks characterized by large weights and negative Ricci curvature relative to the query node. Extensive experiments confirm that Ricci-Filtration outperforms several baseline reranking methods in accuracy, precision, recall, and F1 scores. Furthermore, ablation studies demonstrate that the Ricci-Filtration generally outperforms the baseline under various settings, highlighting the framework's robustness across different architectures.
arXiv:2606.16969v1 Announce Type: new Abstract: Low frame rates in neural audio codecs are attractive for autoregressive speech synthesis, where the generation cost scales linearly with the sequence length. Recent work has demonstrated that codecs can operate at 12.5 Hz and below, but the mechanisms underlying low frame rate degradation remain insufficiently understood. We investigate these mechanisms through a controlled frame rate ablation. We reproduce a quality cliff at 6.25 Hz reported in previous works and evaluate candidate explanations: phonemic collisions and codebook saturation, neither of which shows evidence of a fundamental barrier. The cliff is instead caused by suboptimal training configuration: fixed clip duration during training yields too few tokens at low frame rates, starving the decoder of inter-token context. Once corrected, WER degrades smoothly with phonemic load down to 3.1 Hz and 1.6 Hz, suggesting the inference-time efficiency gains of low frame rate codecs are more accessible than previously assumed.
arXiv:2606.12928v2 Announce Type: replace-cross Abstract: We design the first neural quantum state for continuum particles that, for any chosen allowed momentum $\mathbf{k}$, is by construction an exact eigenstate of total momentum with eigenvalue $\mathbf{k}$. Our architecture, EVE, enables off-the-shelf VMC to solve for momentum-sector ground states. We test EVE on 2D bosons with mutual $1/r$ interactions, finding that a single unified ansatz is capable of describing four qualitatively different states: superfluid, roton, crystal, and phonon. At different densities, we extract the underlying phase of matter from the dispersion's shape. At $r_s = 20.0$, we see the roton minimum at finite $k$ expected of a superfluid. At $r_s = 100.0$, we see striking zone folding indicative of crystalline order, with periodically spaced minima representing floating crystals connected by phonon arcs in between. Using density-density correlation functions, we confirm the phase diagnoses and probe the excitations' correlation structures. Finally, we analyze the roton's phase texture and find unexpected multi-particle phase strings, formed when several vortex dipoles merge, leaving two vortices connected by a phase slip.
arXiv:2606.02955v2 Announce Type: replace Abstract: Diffusion large language models promise parallel token generation, yet inference remains bottlenecked by deciding which masked tokens can be safely committed together. Fast-dLLM addressed this with KV caching and confidence-guided parallel decoding, but its decoding theory uses a homogeneous high-confidence assumption that effectively reduces each candidate set to its weakest selected token. We argue that this leaves speed on the table because real decoding steps exhibit heterogeneous confidence profiles. We propose \textbf{Fast-dLLM++}, a training-free extension that introduces \emph{Fr\'{e}chet profile decoding}: selecting parallel commit sets from the full sorted confidence profile rather than a single worst-case confidence. The resulting rule is a heterogeneous-confidence generalization of Fast-dLLM's factor selector and it recovers the previous rule exactly in the equal-confidence case and adds a provable \emph{heterogeneity bonus} when the selected tokens have uneven confidences. Fast-dLLM++ leaves the model, diffusion process, and cache implementation entirely unchanged, making it a drop-in replacement for existing Fast-dLLM decoding. Experiments on GSM8K, MATH, HumanEval, and MBPP with the LLaDA-8B model show that the theoretical improvement translates directly into empirical gains: profile-aware selection improves the accuracy--throughput frontier by exploiting safe parallelism that weakest-token rules miss, achieving up to 37\% higher throughput at comparable accuracy. Our code release is at https://github.com/Ringo-Star/FastdLLM_plusplus.