arXiv:2604.10842v3 Announce Type: replace Abstract: LLM-powered coding agents increasingly rely on tool-use protocols such as the Model Context Protocol (MCP) to read and write files on a developer's workstation. When a write fails - due to content filters, truncation, or an interrupted session - the agent typically receives no structured signal, loses the draft, and wastes tokens retrying blindly. We present Resilient Write, an MCP server that interposes a six-layer durable write surface between the agent and the filesystem. The layers - pre-flight risk scoring, transactional atomic writes, resume-safe chunking, structured typed errors, out-of-band scratchpad storage, and task-continuity handoff envelopes - are orthogonal and independently adoptable. Each layer maps to a concrete failure mode observed during a real agent session in April 2026, in which content-safety filters silently rejected a draft containing redacted API-key prefixes. Three additional tools - chunk preview, format-aware validation, and journal analytics - emerged from using the system to compose this paper. A 186-test suite validates correctness at each layer, and quantitative comparison against naive and defensive baselines shows a 5x reduction in recovery time and a 13x improvement in agent self-correction rate. Resilient Write is open-source under the MIT license.
Science Journals
arXiv:2604.10999v2 Announce Type: replace Abstract: Vision-language models (VLMs) perform strongly on multimodal benchmarks, but their ability to follow complex visual paths remains under-tested. We introduce TraversalBench, a controlled benchmark for exact visual path traversal. Each instance contains a continuous polyline with a unique start marker and labeled vertices; models must recover the ordered sequence encountered from start to finish. The benchmark balances self-intersection count, tortuosity, vertex count, and nearby confounding lines while limiting reliance on OCR, world knowledge, or open-ended planning. We find that self-intersections are the dominant source of difficulty. A first-crossing analysis localizes failures to crossing points: performance is stable before the first crossing, then drops sharply when the model must resolve the correct continuation. Nearby confounders have weaker but compounding effects, and an auxiliary reading-order benchmark reveals a consistent left-to-right bias. Together, these results characterize how VLMs perceive and fail on visual paths. Finally, we position TraversalBench as a new contribution to the growing line of sustained and precise visual grounding benchmarks for VLMs. Code, benchmark data, and rendered examples are available at https://github.com/clarapetrova/traversalbench.
arXiv:2606.09455v1 Announce Type: new Abstract: Many parcel networks dispatch vehicles on \emph{fixed routes} using a simple load-accumulation rule: a truck leaves the depot for a fixed route as soon as the volume (or item count) waiting for that route crosses a threshold. The threshold is usually parameterised as an affine function of route length, $\tau_r=\beta+\gamma\,d_r$, and the pair $(\beta,\gamma)$ is chosen once and frozen into production. This paper studies how good that frozen choice actually is, treating the question as a data-intensive, data-driven decision-making problem over a full month of real operational flow. Using a discrete-event simulator that replays the recorded arrival stream and reconstructs every trip, we sweep the $(\beta,\gamma)$ design space, evaluate the two competing objectives -- company operating cost and average parcel lead time -- and recover the Pareto frontier of efficient policies for two deployed variants (volume-triggered and item-count-triggered). The two policies turn out to be in strikingly different states of tune. The volume-threshold configuration lies on its own Pareto frontier: the simulator finds no $(\beta,\gamma)$ pair that strictly dominates it, so the deployed policy is \emph{already Pareto-efficient} -- an unusual positive audit result. The item-count configuration is the opposite: it is dominated by a concrete simulated configuration that is both faster and cheaper, and the available cost saving at equal lead time is about \num{5.0}\,\pct{}. We trace the item-count policy's inefficiency to a base that is too large and a length coefficient that is too small for the deployed truck capacity, and show that a \emph{steeper} threshold -- lower base, higher slope -- is preferable. Because the remedy is a two-scalar reconfiguration, the analysis converts directly into an actionable, zero-capital recurring saving.
arXiv:2606.09459v1 Announce Type: new Abstract: Retrieval-augmented generation often fails when the query, the document evidence, and the user's intent are expressed at different levels of abstraction. A query may ask about a class, a relation, or an event, while the document only states specific instances, indirect framings, or scoped formulations. We define this mismatch as an abstraction gap: the minimal set of typed assumptions required to align query intent with the available evidence. To close this gap, we introduce AbstRAG, which treats abstraction as an explicit retrieval object. AbstRAG decomposes the query--evidence gap into expression, conceptual, intent--evidence, and event-type components, and scores relevance by combining match quality, a query-independent utility prior, and the cost of the required bridges. Its central mechanism is reflective refinement: a critic diagnoses retrieval failures, localizes the failed abstraction operator, proposes a minimal stage-specific patch, and accepts the patch only under sufficiency and compression controls. Across three within-document retrieval benchmarks against seven baselines, AbstRAG outperforms on nDCG@10 in 18 of 21 paired-bootstrap contrasts and improves generation accuracy by 1.9%, 5.2%, and 4.0% across the three benchmarks; ablations confirm that reflective refinement drives most of the retrieval gain and the compression control alone reduces over-expansion false positives from 73.7% to 0% on a stress slice.
arXiv:2606.08885v1 Announce Type: new Abstract: This paper proposes a design-for-test (DFT) methodology and architecture for testing and validation of silicon photonic integrated circuits. We describe the design of silicon photonic circuits and components that comprise the proposed DFT architecture. The designs are extensively simulated and validated as test-access and fault-detection circuitry. We demonstrate how the DFT approach can be deployed on photonic integrated circuits and how they can be tested for correct operation, in terms of signal power and phase. The application is demonstrated on two distinct types of designs -- an optical neural network comprising optical devices in a feed-forward topology, and on an optical logic circuit with feedback loops.
arXiv:2606.09559v1 Announce Type: new Abstract: Offline safe reinforcement learning (Safe RL) enables policy learning without online interactions, making it suitable for safety-critical systems such as robotics systems. However, its reliance on static datasets exposes offline Safe RL to data poisoning attacks, where adversaries inject malicious samples that compromise safety and induce unsafe policy behavior. In this work, we propose a new learning paradigm, named safe reinforcement unlearning (Safe-RULE), used as a defense framework to remove the influence of poisoned data without retraining from scratch or requiring access to the original training environment. We further extend reinforcement unlearning to offline Safe RL by explicitly accounting for both task performance and safety constraints during the unlearning process. Experiments across benchmark Safe RL tasks demonstrate that our approach effectively enhances safety performance against data poisoning attacks.
arXiv:2606.08545v1 Announce Type: new Abstract: Building Information Modeling (BIM) projects require information requirements to be described as machine-checkable Information Delivery Specification (IDS) files in order to verify whether building models contain the required attributes. However, IDS authoring remains a practical bottleneck: practitioners must handle domain vocabulary, strict XML schema constraints, and external validator conformance while also checking whether the requirement itself is correctly expressed. We present Ishigaki-IDS, an open-weight LLM specialized for verifier-aware IDS draft generation. The model combines continued pretraining on BIM/IDS corpora, supervised fine-tuning on information-requirement-to-IDS pairs, and reinforcement learning with verifiable rewards from an external validator. The goal is not to replace expert review, but to move IDS authoring from low-level XML and schema repair toward validator-loadable drafts that practitioners can inspect and correct. On the 166-case expert-created Ishigaki-IDS-Bench, Ishigaki-IDS-8B achieves an IDSAuditPass score of 0.651, a validator-pass metric for generated IDS files, substantially outperforming Claude Opus 4.5, the strongest single-shot LLM baseline we evaluated, at 0.331. It also obtains an Audit-Gated FacetF1 of 0.282, which measures requirement-facet alignment among validator-passing drafts. The same recipe scales: 14B and 32B variants reach IDSAuditPass 0.753 / 0.693 and Audit-Gated FacetF1 0.392 / 0.369. In a workflow check with six BIM practitioners, Ishigaki-assisted authoring reduced aggregate work time by 54.7% under the same validation and alignment endpoint. These results suggest that verifier-aware IDS generation can reduce the practical burden of converting BIM information requirements into reviewable IDS drafts.
arXiv:2606.08932v1 Announce Type: new Abstract: Rule-following agents tasked with executing policies and regulations often fail via Silent Scope Omission (SSO): a model applies a general rule but silently drops nested exceptions or counter-exceptions, producing outputs that appear compliant yet break on important edge cases. Although such failures are often framed as an agentic-systems problem, the underlying bottleneck is statutory and policy understanding, a capability typically studied in legal NLP. However, most existing legal NLP benchmarks emphasize end-task outcomes, which can overlook the structural omissions that cause SSO. To diagnose and mitigate SSO, we introduce NormBench, a benchmark of 2,290 provisions spanning Chinese (laws and local policies), English (U.S. tax law, GDPR, and corporate policies), and cross-lingual settings, designed for defeasible scope parsing: identifying precisely which clause overrides which. NormBench uses Span-Grounded Deontic Trees (SG-DT), a compiler-style intermediate representation that anchors every logical branch to source spans and requires explicit exclusion guards, enabling deterministic compilation and audit. Evaluations of frontier LLMs reveal two recurring pathologies: (1) Recursion Decay, where performance drops sharply as defeater depth increases, and (2) an Auditability Trap, where models retrieve relevant spans but fail to assemble correct control flow. Using SG-DT as a constrained intermediate output improves whole-tree fidelity and defeater recovery, and downstream experiments show that its utility is mechanism-specific: gains concentrate on exception-active, SSO-prone cases, while aggregate accuracy can be mixed when the added structure is unnecessary or parser fidelity is low.
arXiv:2606.09463v1 Announce Type: new Abstract: With the rise of new industries, often new jobs emerge. Evolutionary Economic Geography and in particular Industry Life Cycle perspectives predict that these activities first emerge in a limited number of cities to then diffuse to other locations as job descriptions become more standardized. Here, we focus on a particularly important new industry: software development, an activity that is economically important, quickly changing, and has a pronounced spatial concentration in a small number of global IT hubs. We use an online database of over 60 million questions and answers about problems in software development that yields a longitudinal dataset of 237 software skills. By geo-locating 3 million posting users at regular intervals, we link these skills to cities worldwide. We find that, in spite of its digital nature, the software industry exhibits similar spatial regularities as previously observed in more traditional sectors. First, cities diversify into skills that are related to their existing ones. Second, new skills first emerge in cities with large and diversified software sectors, and later diffuse -- mostly unhindered by geographical distance -- to smaller cities specialized in closely related skills. We find suggestive but limited support for a windows of locational opportunity account: although even brand-new skills still emerge first in cities with strong prior specialization in related skills, concentrations of related activities impact less the emergence of new skills than the diffusion of existing ones.
arXiv:2606.08894v1 Announce Type: new Abstract: Reasoning Vision-Language Models (VLMs) achieve strong performance on complex multimodal tasks, but reliable real-world application requires handling visual inputs that are messier than clean, curated benchmarks. Existing works mainly evaluate such reliability of VLMs through input corruptions, such as noise, blur and weather effects, which make visual evidence harder to perceive. This leaves a critical reliability failure mode underexplored: a model may perceive the evidence correctly, yet reason from plausible but irrelevant and distracting evidence and propagate this mistake to its final answer. To address this gap, we introduce \textbf{Distract-Bench}, a benchmark for evaluating VLM robustness to \textbf{semantic visual distractions}, defined as meaningful but task-irrelevant visual cues added to inputs while preserving the ground-truth answer. We comprehensively evaluate eight leading open-source and two closed-source VLMs across conventional vision corruptions and Distract-Bench. Our results show that Distract-Bench exposes a robustness failure distinct from vision corruptions: reasoning VLMs largely track their non-reasoning base models under perceptual degradation, but show consistently lower robustness to semantic distractions. Further analysis shows that these distractions often enter the reasoning process of VLMs, are treated as evidence, and lead to incorrect answers. Together, these findings reframe robustness evaluation for reasoning VLMs, shifting the focus from degraded perception to distractions for reliable real-world visual reasoning. Our data and code are available at https://github.com/Yizheng-Sun/Distract-Bench.
arXiv:2606.08563v1 Announce Type: new Abstract: While global data-driven models excel at predicting continuous atmospheric variables, three-dimensional hydrometeor forecasting remains challenging due to the zero-inflated, long-tailed distributions of these variables. Standard deep learning optimization often yields overly smooth forecasts, attenuating extreme events and spatial textures. We propose PredHydro-Net, a physics-guided dual-decoding framework that mitigates this smoothing. To resolve multi-variable optimization conflicts, it employs a decoupled architecture where macroscopic thermodynamic and dynamic fields unidirectionally modulate hydrometeor generation. By integrating wavelet-based frequency decoupling, spectral amplitude matching, and adversarial training, the model achieves a favorable trade-off between quantitative accuracy and spatial fidelity. In a 72-h global evaluation, PredHydro-Net outperforms both spatiotemporal deep learning baselines (Earthformer and PredRNNv2) and the operational Global Forecast System (GFS) in extreme-event detection and spectral representation. Furthermore, it demonstrates strong climatological consistency with Global Precipitation Measurement (GPM) satellite retrievals. The model reasonably reproduces the three-dimensional cloud structures in extreme weather events, such as Hurricane Ian. Feature attribution confirms its dependence on physical precursors such as relative humidity and wind convergence, offering a robust, physics-informed approach to long-tailed atmospheric prediction.
arXiv:2606.08564v1 Announce Type: new Abstract: Robotic manipulation robustness often founders on the physics gap between simplified simulations and the resistance-laden real world. In this work, we emphasize that physical realism in articulated interaction is an important ingredient for robust policy learning. We present Real-IKEA, a dataset and simulation framework designed with physical accuracy as a first-class goal. Real-IKEA provides 1,079 articulated asset configurations, derived from 83 authentic IKEA handles and knobs processed through a meticulous six-step physical workflow. For contact-geometry accuracy, we introduce a bidirectional surface-deviation metric to quantify collision meshes. For dynamics realism, we establish resistance-calibrated configurations that vary damping and friction. Crucially, we demonstrate through a Reinforcement Learning (RL) policy that high-fidelity assets enable the discovery of robust "hooking" and "levering" strategies that prioritize mechanical advantage over fragile friction-pulling. Together, these results position Real-IKEA as a critical benchmark for developing manipulation policies capable of human-level robustness in articulated object tasks.
arXiv:2606.08408v1 Announce Type: new Abstract: We extend activation steering to diffusion language models (DLMs) and study a novel problem that arose due to the inference mechanism of DLMs: Modifying a text in-place to manifest a different concept. We propose TimpaTeks, an automatic in-place text modification mechanism using DLMs. Experiments on IMDB movie reviews (sentiment) and a synthetic Cats and Dogs Dataset (arbitrary, more unconventional concept steering) show that TimpaTeks provides a feasible novel mechanism to steer diffusion language model outputs in-place. TimpaTeks enables in-place modification while simultaneously lowers sentence perplexity and retaining the original sentence structre without the need of instruction tuned models. TimpaTeks is also computationally cheaper than prompt-based DLM steering, as it performs denoising in-place rather than constructing an additional prompt-conditioned output sequence.
arXiv:2606.07717v1 Announce Type: cross Abstract: This work proposes a lightweight 2D-U-Net-based framework for segmenting five abdominal organs in large field-of-view 3D CT scans. The method combines coarse-to-fine segmentation, predictions from multiple anatomical planes, and additional fuzzy 3D spatial maps that provide anatomical location cues to improve segmentation accuracy. We combine multi-planar 2D-U-Net models augmented by a spatial occurrence map. The approach involves two main stages. First, the abdominal volume of interest region is detected by traversing the whole scan axially with a 2D-U-Net and determining the x-y-z-minimum and -maximum extents of the 5 abdominal organs of interest. Second, we use spatial occurrence maps to enhance our multi-planar 2D-U-net architecture inside the bounds from the former stage. The method is evaluated on 80 CT scans from various public sources. The results show Dice improvements of about 4% at maximum compared to the same model trained without spatial occurrence maps.
arXiv:2606.08840v1 Announce Type: new Abstract: Code generation models are typically compared using compact execution benchmarks and aggregate pass rates, but such summaries obscure how performance varies across programming languages, problem families, and failure modes. We present a large-scale, execution-grounded evaluation of 9 openly accessible LLMs specialized for coding on 2,707 free LeetCode problems across 12 programming languages. Our corpus contains 325,343 problem-model-language jobs, each linked to prompt metadata, extracted code, LeetCode execution outcomes, and static-analysis signals. The results show that current open models remain far from the human acceptance reference: the best model, Yi-Coder-9B-Chat, reaches 23.64% mean correctness, compared with a 57.2% human acceptance baseline. Rankings are also slice-dependent: Qwen2.5-Coder-14B-Instruct is strongest on hard problems and distinct-problem coverage, while Gemma-2-27B-IT achieves the highest all-language lint pass rate. Failure analysis shows that compile errors account for 63.25% of non-accepted best submissions, indicating that many failures occur before semantic correctness can be tested. Static quality further diverges from functional correctness. Together, these findings show that multilingual, artifact-preserving evaluation reveals tradeoffs hidden by single-language or single-metric leaderboards.
arXiv:2606.08846v1 Announce Type: new Abstract: The ethanimine molecule, CH3CHNH, is one of the prebiotic molecules detected by astronomers in chemically-rich molecular clouds in the Galactic Center. The observations indicate a non-equilibrium distribution of rotational state populations in both the E- and Z-isomers of ethanimine, resulting from the competition between radiative processes and collisions with background gases such as He and H2. Accurate interpretation of these observations requires the use of radiative transfer models with collisional state-to-state transition processes included. Here, in order to compute cross sections for state-to-state transitions in both ethanimine isomers, accurate potential energy surfaces for their interaction with a He atom were constructed and three complementary methods for inelastic scattering were utilized: full-quantum coupled-channel and coupled-states methods, and the mixed quantum/classical theory. Strong propensities of transitions toward $\Delta j = 0$ and either $\Delta k_a = 0$ (with $\Delta k_c = \pm 1$) or $\Delta k_c = 0$ (with $\Delta k_a = \pm 1$) are reported and the origin of this effect is identified. Small but non-negligible differences between energy transfer in the two isomers, on the order of 10%, were found. The utility of the mixed quantum/classical approach to collisional energy transfer at higher collision energies is discussed.
arXiv:2606.08458v1 Announce Type: new Abstract: Proactive robot assistance in household environments requires accurate prediction of human activities and object usage under dynamic and noisy conditions. Existing approaches often rely on complex spatio-temporal models, which can be computationally expensive and sensitive to environmental variability. In this paper, we propose GLOBE, a lightweight framework that combines n-gram Markov models for capturing temporal behavioral patterns with uncertainty-guided large language model (LLM) reasoning. The framework performs sequential prediction efficiently while selectively invoking LLM reasoning only when the model confidence is low. To evaluate performance under realistic conditions, we introduce HOMER-Noise, a noisy extension of the HOMER+ dataset that simulates structured disturbances such as object movements caused by humans, pets, and toddlers. Experimental results show that GLOBE achieves competitive performance with state-of-the-art methods while improving robustness and computational efficiency across both clean and noisy settings. The framework is further validated through a proof-of-concept integration with a Stretch 3 mobile manipulator, demonstrating its potential application in real-world human-robot interaction scenarios.
arXiv:2605.21854v2 Announce Type: replace Abstract: Vision-Language-Action (VLA) models have rapidly converged on a small set of architectural patterns: discrete-token autoregression (e.g. OpenVLA) and continuous-action flow-matching (e.g. pi-0.5). Yet preference alignment via Direct Preference Optimisation (DPO) -- the de-facto post-training step in language models -- has been studied almost exclusively on autoregressive VLAs. We present CrossVLA, an empirical study of cross-paradigm VLA post-training. Three contributions: (i) a surrogate flow-matching log-probability estimator that lets DPO operate on continuous-action backbones without probability-flow ODE integration; (ii) a head-to-head comparison of LoRA and DoRA as the parameter-efficient layer for VLA DPO, finding DoRA improves over OpenVLA SFT by a mean +10.4 pp across LIBERO 4-suite (600 trials, 3 seeds) -- per-suite +20.0 Object, +11.0 Long-horizon, +8.0 Goal, +2.7 Spatial -- with zero seed variance on Object (38/50 on each of 3 seeds); (iii) an inference-time anatomy showing the denoise loop dominates 78.6% of sample_actions latency and prefix-K/V caching a la VLA-Cache caps at a 21% acceleration ceiling -- both chunk-level and token-level cache strategies degrade success rate to 0-80% in our benchmarks. We further pretrain a multi-view + temporal projection head on 6000 LIBERO frames, achieving 99.5% k-NN recall@1 for same-task retrieval (36x over random), available as a downstream initialisation. All code, ckpts, training logs, and reproduction scripts are open at https://github.com/lz-googlefycy/vla-lab.
arXiv:2605.22079v2 Announce Type: replace Abstract: Building Information Modeling (BIM) projects increasingly use Information Delivery Specification (IDS) to formalize information requirements in a machine-checkable XML format. Because IDS conditions are grounded in the Industry Foundation Classes (IFC) vocabulary, authoring them requires expertise in IFC concepts, validation tools, and property set conventions. Existing benchmarks for structured generation do not adequately capture the additional burden of vocabulary conformance and external-validator agreement that IDS imposes. We present Ishigaki-IDS-Bench, the first publicly released benchmark for IDS generation from BIM information requirements. The benchmark contains 166 examples spanning 83 practical scenarios authored in Japanese and English by six BIM/IDS experts, each paired with a gold IDS file and metadata covering input format, turn setting, target IFC versions, and construction domain. Evaluation proceeds in two stages: (i) formal validity scored by the buildingSMART IDSAuditTool along Processability, Structure, and Content, and (ii) content fidelity scored by facet-level macro-F1 against the gold IDS. Across 10 LLMs in zero-shot, the highest Facet F1 is 65.6%, achieved by GPT-5.5, while the highest Content pass rate is only 33.1%, achieved by Claude Opus 4.5. Ishigaki-IDS-Bench is released on Hugging Face (DOI 10.57967/hf/8873) under CC BY 4.0, and the evaluation code is released on Zenodo (DOI 10.5281/zenodo.20550510) under Apache-2.0.
arXiv:2605.22208v2 Announce Type: replace Abstract: Multimodal Large Language Model (MLLM)-driven image restoration agent demonstrates effectiveness in degradation coupling scenarios by flexibly selecting tools and determining removal orders. However, their zero-shot planning often fails without experience, necessitating severe trial-and-error overhead to achieve satisfactory outcomes. Currently, two paradigms are employed to address this issue, yet a dilemma persists: Training-based methods embed intrinsic experience into parameters, achieving high inference efficiency but lacking compatibility with new tools or degradation. In contrast, training-free methods utilize explicit experience storage for compatibility but still incur trial-and-error overhead due to naive experience. To resolve the dilemma, we propose EvoIR-Agent, which first systematically formulates the experience components of a training-free image restoration agent. Subsequently, a hierarchical experience pool is constructed, which enables coarse-to-fine guidance for diverse tools and removal orders. Furthermore, a self-evolving mechanism is introduced to update the pool from scratch using accumulated records, thereby greatly improving performance and efficiency. Extensive experiments reveal that EvoIR-Agent achieves a significant lead in the full reference metrics and yields a remarkable Pareto-optimal balance between performance and efficiency compared to the state-of-the-art methods.
arXiv:2605.22763v2 Announce Type: replace Abstract: Large language models (LLMs) increasingly excel at mathematical reasoning, but their unreliability limits their utility in mathematics research. A mitigation is using LLMs to generate formal proofs in languages like Lean. We perform the first large-scale evaluation of this method's ability to solve open problems. Our most capable agent autonomously resolved 9 of 353 open Erd\H{o}s problems at the per-problem cost of a few hundred dollars, proved 44/492 OEIS conjectures, and is being deployed in combinatorics, optimization, graph theory, algebraic geometry, and quantum optics research. A basic agent alternating LLM-based generation with Lean-based verification replicated the Erd\H{o}s successes but proved costlier on the hardest problems. These findings demonstrate the power of AI-aided formal proof search and shed light on the agent designs that enable it.
arXiv:2512.12225v3 Announce Type: replace Abstract: Developing artificial agents that unify representation, memory, adaptation, and prediction remains a fundamental challenge in artificial intelligence. Here we introduce a geometric framework in which cognitive computation emerges from Riemannian gradient flow on a learned latent manifold. The learned metric encodes representational constraints and computational preferences, while anisotropies in the geometry naturally generate multiple timescales of behaviour, yielding both rapid reactive responses and slower adaptive dynamics without explicit memory modules or recurrent mechanisms. We instantiate this framework through Riemannian representation and dynamics models and evaluate them in partially observable reinforcement-learning environments. Across observation masking, sensory blackouts, dynamics perturbations, and predictive latent-modelling tasks, the proposed approach consistently outperforms feedforward baselines, achieves robustness comparable to recurrent architectures, and produces highly predictable latent trajectories with low long-horizon rollout error. These results suggest that learned latent geometry can serve simultaneously as a substrate for representation, memory, adaptation, and prediction. More broadly, the framework provides a principled connection between dynamical systems, representation learning, and world-model-based intelligence.
arXiv:2512.20163v4 Announce Type: replace Abstract: For nearly two decades, population protocols have been extensively studied, yielding efficient solutions for central problems in distributed computing, including leader election, and majority computation, a predicate type in Presburger Arithmetic closely tied to population protocols. Surprisingly, no protocols have achieved both time- and space-efficiency for congruency predicates, such as parity computation, which are complementary in this arithmetic framework. This gap highlights a significant challenge in the field. To address this gap, we explore the parity problem, where agents are tasked with computing the parity of the given sub-population size. Then we extend the solution for parity to compute congruences modulo an arbitrary $m$. Previous research on efficient population protocols has focused on protocols that minimise both stabilisation time and state utilisation for specific problems. In contrast, this work slightly relaxes this expectation, permitting protocols to place less emphasis on full optimisation and more on universality, robustness, and probabilistic guarantees. This allows us to propose a novel computing paradigm that integrates population weights (or simply weights), a robust clocking mechanism, and efficient anomaly detection coupled with a switching mechanism (which ensures slow but always correct solutions). This paradigm facilitates universal design of efficient multistage stable population protocols. Specifically, the first efficient parity and congruence protocols introduced here use both $O(\log^3 n)$ states and achieve silent stabilisation in $O(\log^3 n)$ time. We conclude by discussing the impact of implicit conversion between unary and binary representations enabled by the weight system, with applications to other problems, including the computation and representation of (sub-)population sizes.
arXiv:2606.08496v1 Announce Type: new Abstract: Although Sparse Autoencoders (SAEs) have mitigated the opacity of large language models (LLMs) by decomposing dense representations into sparse features, explaining these features still remains a central challenge. Current explanation methods, however, typically operate within an open-loop paradigm, failing to leverage mechanistic feedback for further refinement. In this paper, we propose SAEExplainer, a training framework utilizes activation scores as an objective reward signal to train the model for self-correction and iterative bootstrapping. By iteratively verifying and correcting foundational explanations through a two-round optimization process, SAEExplainer achieves continuous improvement in its explanatory capabilities. This mechanism significantly reduces explanation hallucinations and reinforces causal triggering patterns. Extensive experiments demonstrate our approach improves upon established baselines across most metrics, especially in causal triggering and discriminative activation.
arXiv:2606.08508v1 Announce Type: new Abstract: Generative robot policies fail unpredictably at deployment: they hesitate at critical moments, drift off-task, or commit to unrecoverable actions. Existing online failure detectors either require white-box access to policy internals or add runtime overhead through resampling and observation-side signals. Our empirical analysis shows that emitted action chunks themselves already carry strong predictive signal for impending failures in generative robot policies. Motivated by this observation, we introduce ActProbe, a lightweight, pure action-space detector that uses two compact signals available from a single forward pass: Temporal Consistency Error (TCE) between consecutive action chunks and Action Chunk Magnitude (ACM) of the current chunk. ActProbe maps these signals to per-step failure probabilities with a task-conditioned LSTM-MLP architecture. Across a diverse suite of generative robot policies and benchmarks, ActProbe raises alerts before failures become visually recognizable, improving the accuracy (F1)-timeliness Pareto frontier of failure detection by an average hypervolume gain of +12.7% over both internal- and external-feature baselines, with a +9.0% early-detection ROC-AUC lead on unseen tasks. ActProbe further transfers to deployment, predicting failures on unseen real-robot pick tasks and accelerating RL fine-tuning (PPO) with 2.9x fewer environment interactions.