Search research

Fragmented safety evaluation undermines the governance of dangerous AI capabilities. We present a modular framework that evaluates each model through three orthogonal pipelines---Knowledge ($K$), Defense ($D$), and Harm ($H$)---under a unified protocol, aggregating results into a standardized dangerous-capability profile $φ$. Pluggable modules supply scenario seeds, knowledge banks, hazard queries, and judge rubrics, while the core evaluation engine remains unchanged across domains; the CB evaluation is complemented by a cyber pilot demonstrating protocol transfer. Instantiating the framework with a chemical-biological (CB) module, we evaluate 12 commercial LLMs from four families. Our first contribution is a horizontal comparison of dangerous capability across models and model families: the three dimensions expose sharply divergent profiles---models with comparable knowledge differ in refusal resilience, and strong defenders do not generate less harmful content when they do comply---while family-level patterns further separate Claude, DeepSeek, and GPT models. The second is a temporal analysis of capability evolution: tracking $K$, $D$, and $H$ against model release dates reveals that dangerous capability has not monotonically declined; newer models deepen knowledge while only partially improving defense, showing that scaling and alignment progress do not uniformly translate into safety. Reliability is established via cross-judge consistency (bootstrap $ρ> 0.79$, 4 of 5 judges) and pipeline orthogonality ($K$--$D$--$H$ inter-correlations $ρ\in [0.32, 0.52]$).

Federated Learning enables multiple clients to train a shared model while keeping their local datasets isolated. However, the exchanged model updates may still leak sensitive information, making private aggregation a central building block in practical deployments, especially in the cross-silo setting. Homomorphic Encryption naturally fits the client--aggregator communication pattern of Federated Learning, but conventional single-key deployments rely on strong non-collusion assumptions. Multiparty Homomorphic Encryption removes this limitation, although recent attacks under restricted decryption access require large-variance smudging noise during collaborative decryption, which significantly increases ciphertext size and implementation complexity. In this work, we propose lightweight multi-secret-key protocols for private average aggregation based on RLWE-based Homomorphic Encryption. Our construction departs from the usual multiparty blueprint by avoiding the generation of a collective public key. Instead, each client encrypts its update under its own secret key, while the resulting ciphertexts remain compatible with homomorphic aggregation and collaborative decryption. By explicitly tracking and cancelling the ciphertext noise during decryption, the protocol removes the need for large $λ$-dependent smudging noise. We instantiate the construction with both exact BFV-based and approximate CKKS-based variants, prove its security in the semi-honest model against an adversary corrupting the aggregator and up to $L-1$ clients, and compare its communication and runtime performance with state-of-the-art MHE-based aggregation. Our results show that the proposed approach substantially reduces ciphertext expansion and online cost, while preserving practical homomorphic aggregation performance.

A language model's prediction of its next token develops across layers, and lens methods track this process by decoding intermediate hidden states into tokens. But a lens reading reflects both the hidden state and the readout (the unembedding matrix) used to decode it. Many lenses are fit on a corpus, and we show that two lenses differing only in their fitting corpus can report different tokens for the same hidden states. We call this dependence corpus conditionality. To examine readout structure independently of the fitting corpus, we introduce Sparse Readout Prism (SRP), which decomposes the readout using only its weights and expresses any token logit or logit difference as a sum of contributions from sparse readout features. This reveals readout features as a new unit of analysis for lens readings, exposing structure that token identities can obscure and enabling comparisons across tokens, contexts, layers, and lenses. Replacing the original readout with SRP's sparse approximation reconstructs 8.9-17.3 percentage points more of the tested logit differences than the strongest of six baselines built on geometric relations among readout rows. Ablating features shifts logit differences in proportion to their SRP contributions. Although token readings vary with the fitting corpus, the dominant readout feature remains stable. Because SRP uses no corpus in its construction, it provides a control independent of the fitting corpus for lens analyses.

Recent work identifies a mid-depth band of verbalisable, causally potent representations in a standard feedforward transformer --- a functional analogue of a global workspace. Whether the same workspace functionality emerges when depth is implemented through recurrence rather than a stack of distinct layers remains unknown. Looped and depth-recurrent transformers provide a direct test of this question because they reuse the same weights across depth. We extend the Jacobian lens to iterated architectures using a virtual-unrolling adapter. We apply the full workspace suite --- lens fitting, readout, and eleven causal experiment families --- to Ouro-2.6B (48 layers looped 4 times, deeply supervised) and Huginn-0125 (a 4-layer core recurred 16 times, trained for latent reasoning), using Qwen3.6-27B (64 untied layers) as the standard baseline. We find that a workspace forms in the iterated part of each architecture, but that recurrence changes how it can be accessed. Ouro reconstructs workspace content in every loop, and linear transport cannot carry that content across loop boundaries; writes and ablations must therefore span every remaining loop. Huginn carries content forward across all sixteen recurrences, while reads, writes, and ablations act only within a sliding window of roughly two recurrences. Whether newly injected content can be verbalised tracks explicit per-iteration supervision; whether existing content can be steered does not.

Multi-agent AI systems improve inference by spawning agents and synthesizing reports. But another agent is not another observation: apparently independent reports may descend from the same evidence, and genuinely independent evidence can produce nearly identical reports. We formalize this as an epistemic Sybil problem. A report Z is an epistemic Sybil extension relative to reports R when I(Theta; Z | R) = 0. No report-only aggregator can generally distinguish replication from independent corroboration: identical reports can warrant different posteriors under unobserved ancestry. A Gaussian shared-root model shows common ancestry does not imply complete redundancy. Repeated extraction adds information toward a source-level ceiling, and correlated extraction errors, which a shared base model can induce among independent agents, lower that ceiling further. We test these predictions with more than 20,000 controlled LLM-agent report and extraction calls on synthetic evidentiary documents. Holding one evidence root fixed while report multiplicity rises from 1 to 32 collapses naive posterior coverage from 0.940 to 0.263. Holding report count fixed while evidence-root multiplicity rises from 1 to 16 closes the gap, and the aggregators are statistically indistinguishable at k = 16. The agent's replicate extraction errors are correlated (gamma_cal = 0.719, estimated out of sample), and a correlated-extraction aggregator restores calibration accordingly. A controlled manipulation isolates representation similarity from evidential ancestry. It changes a report-space deduplication mechanism's mean inferred cluster count by 1.425 (95% CI [1.363, 1.485]), whereas a fourfold change in true ancestry changes it by only 0.040 ([-0.045, 0.120]). Collective inference should therefore track evidential ancestry and dependence, not agent or report multiplicity or similarity.

We introduce a novel Divide and Conquer control design methodology leveraging differential games in single-agent, multi-objective dynamical systems. The proposed framework associates each control objective with a virtual input and establishes a non-cooperative, finite or infinite horizon differential game among representative players. Each player optimizes a distinct virtual cost function tailored to its specific goal, the full system state, and the other virtual inputs, while accounting for the remaining players' optimal policies. By establishing a Nash Equilibrium for this game, we synthesize a composite controller that achieves a stable balance across competing objectives, providing control engineers with an intuitive and modular framework for parameter re-tuning throughout the design cycle. We provide formal mathematical derivations for both continuous-time and discrete-time dynamical systems, targeting large-scale single-agent applications where complex, dynamically conflicting control objectives make global weighting intractable. To demonstrate the methodology, we developed an open-source Python package implementing a novel numerical algorithm for solving Coupled Algebraic Riccati Equations arising in infinite-horizon differential games. We evaluate the approach on two benchmark case studies: an inverted pendulum on a cart and a non-linear hierarchically controlled quadrotor. The resulting closed-loop performance is compared against the classical Linear Quadratic Regulator (LQR) across various transient and steady-state control metrics, demonstrating superior trajectory tracking and robust multi-objective regulation.

Long-running LLM agents rely on persistent memory to carry state across interactions, including permissions, restrictions, and revocations. When memory misrepresents this evolving authorization state, the agent's own records can grant authority that the underlying history never permitted, resulting in misaligned behavior without any external attacks. We term this failure endogenous authorization laundering, where spurious permissions written into memory lead to unauthorized actions as their provenance is washed away. We then introduce EAL-Bench, which measures how accurately persistent memory preserves evolving authorization state and whether errors propagate to downstream unauthorized actions. We evaluate five LLMs as memory writers and two as executors across procurement, cybersecurity, and finance. We find that under incremental memory updates, writers create false authority for up to 50.2% of unauthorized requests; once false authority is present, executors act on it in 98.6% of trials. Two safeguards, requiring stored permissions to be backed by valid source events, and tracking permission changes through bounded event sourcing, substantially reduce laundering, but both also reject more legitimate actions, exposing a safety-utility tradeoff. Persistent memory is therefore not merely a performance component, but a part of an LLM agent's effective authorization policy.

Real-world communication often requires pragmatic reasoning: interpreting meanings implied through context and cultural convention rather than stated literally. Existing pragmatic evaluation remains largely limited to English and high-resource languages, leaving Indic languages unexplored despite their linguistic and cultural diversity. We introduce VakyArth, the first pragmatic benchmark for Indic languages, designed as a diagnostic evaluation covering Hindi, Punjabi, Tamil, and Malayalam. VakyArth evaluates models across five phenomena: deixis, speech acts, implicature, social pragmatics, and coherence; through multiple-choice questions, natural language inference, and translation, with all items authored by native speakers. Across multilingual large language models (LLMs) of varying families and sizes, we find consistent failures on pragmatic meanings rooted in Indic linguistic and cultural conventions. Our analysis shows systematic differences across languages and tasks: MCQ accuracy exceeds NLI accuracy in all model-language combinations, translation performance does not reliably track pragmatic understanding, and Indo-Aryan languages show a translation advantage over Dravidian languages. We further show that automatic translation metrics can miss fluent but pragmatically unfaithful outputs, especially for implicature and deixis.

Kolmogorov--Arnold Networks (KANs) replace the fixed scalar weights of a standard network with learnable univariate functions on each edge, but existing variants still fix the \emph{basis} that those functions are built from: B-splines, Chebyshev polynomials, wavelets, or Jacobi polynomials, and learn only the combination weights over it. We introduce RecKAN, which instead defines the basis itself by a second order polynomial recurrence, $R_{n+1}(x) = (ax^2+bx+c)R_n(x) + (dx+e)R_{n-1}(x)$, whose five coefficients are learned jointly with the network. We show this recurrence recovers several classical polynomial families including both kinds of Chebyshev polynomials, Fibonacci, Pell, and Jacobsthal polynomials as special cases, and prove that its degree grows linearly in $n$ exactly on the sub-family containing all of them, giving a concrete sense in which the learned basis can move beyond any fixed classical choice. Across multiple benchmark datasets spanning image, text, biomedical time series classification, and time series forecasting, RecKAN outperforms three parameter-matched KAN baselines (Chebyshev, Jacobi, and spline based) on all classification tasks and achieves the lowest MSE on the ETTh1 forecasting benchmark. Additionally, when used as a classifier head with a convolutional backbone, RecKAN achieves higher accuracy than standard MLP heads on Fashion MNIST, CIFAR-10, and SVHN. On a synthetic function fitting benchmark it tracks a sharply oscillatory target that a parameter comparable MLP under fits. We further show that the learned recurrence coefficients are interpretable: on the task requiring the most local structure, training moves the basis away from the linear degree growth regime that contains every classical family we identify, consistent with our theoretical analysis of what that structural shift enables.

Stochastic transport describes physical systems in which an initially structured distribution spreads under unresolved forcing, scattering, or heterogeneous media. Useful surrogates for such systems should be probabilistic, time-resolved, and able to represent non-Gaussian distributional structure. Generative diffusion models, which corrupt data with Gaussian noise and learn a reverse flow back to structured states, have these properties. Their noise schedules, however, are usually chosen heuristically: image and audio generation---the canonical use cases---provide no physical clock. In transport, by contrast, the variance, or mean-square displacement, is often known from macroscopic theory or empirical scaling even when the full distribution is not. Here we prescribe the forward noising rate as the time derivative of this variance, turning generative time into a calibrated transport clock. The variance path is enforced by construction, while the learned score field represents how non-Gaussian structure inherited from entrance data is smoothed along that path, requiring no intermediate-time physical transport data. For ballistic-to-diffusive transport in turbulent plasmas, the surrogate matches test-particle distributions, reproduces the laboratory-measured variance scale, and tracks the simulated kurtosis evolution without schedule tuning, enabling calibrated emulation and likelihood-based inference.

AI tutors are most useful when they adapt to each student's strengths, weaknesses, and preferred guidance, but evidence about which guidance works for which student is sparse, slow, and costly to collect from real learners. Student simulators can provide this signal as a proxy, yet existing approaches are limited: state-tracking models fit student behavior but struggle to process explanations or corrections, while LLM role-play follows guidance fluently but does not reliably match the competence of the student being imitated. We present StudentSim, a training framework that turns sparse per-student data into individualized simulators through pooled training followed by per-student specialization. The resulting simulators both mirror a student's own responses and update them under tutor guidance. We also introduce StudentSimEval, a standardized protocol covering 60 students across chess, second-language English writing, and mathematics, using public learner datasets with de-identified records shared for research. StudentSimEval measures behavioral fidelity (F), or how well a simulator matches a student's responses, and guidance responsiveness (R), or how readily it updates under tutor guidance, with all methods fit and evaluated on the same records. Across all three domains, StudentSim outperforms GPT-5.4 on both metrics. In chess, StudentSim reaches F=0.51 and R=0.91, compared with 0.23 and 0.72 for GPT-5.4 and 0.45 and 0.27 for Maia2. As a proof of concept, using StudentSim as a reward model for tutor reinforcement learning produces a chess tutor that expert humans rate as more accurate, better-guided, and more personalized than a no-RL baseline and a tutor trained against a GPT-5.4 simulator reward. Code is available at https://github.com/microsoft/StudentSim.

Data-residency constraints force enterprises to self-host LLMs, but continuous adoption of newer models without decommissioning their predecessors expands the serving fleet, fragmenting a finite GPU pool. We consolidate traffic from over 200 internal applications onto a single model by closing quality gaps identified through production error analysis along three axes: instruction following, function-calling, and internal task distribution. Quality is tracked by offline benchmarks stratified to production traffic and scored by deterministic verifiers or calibrated LLM judges. Rather than optimising all objectives jointly, which introduces cross-domain reward interference, we train a separate GRPO expert per axis and merge them via two-stage SLERP. Each expert's reward exposes a distinct failure mode, namely semantic collapse, over-calling, and verbosity hacking, each requiring a domain-specific fix. In non-reasoning mode the recipe surpasses a ${\sim}7\times$ larger by total parameters baseline on the in-house Arena with 69.6 to 65.8, instruction following with 0.85 to 0.83, and function-calling with 0.79 to 0.77, while lifting general dialogue benchmarks. The model absorbs 50% of platform traffic, 116M requests per month, at a fraction of the serving cost.

Cooperation in multi-UAV systems requires reliable relative perception so that follower vehicles can maintain formation and continue their mission safely even when absolute positioning sensors degrade or fail. This paper presents a vision-based cooperative formation framework running on a follower UAV that uses a front-facing RGB-D camera to detect, track, and localize a leader UAV in real-time. A lightweight YOLO-based detector is trained on a dedicated drone dataset and deployed onboard to predict leader bounding boxes, which are then fused with depth information via a pinhole camera model to estimate the leader's relative pose. These estimates provide a leader-follower position controller and can also be used as a backup when GPS or external localization is unavailable. This framework is implemented as a set of ROS nodes and evaluated in a physics-based multi-UAV simulation built on XTDrone, with sensor noise and communication dropouts. We evaluate detection accuracy, runtime, and formation-keeping error under nominal conditions and under simulated failures of the positioning sensors. The results show that the proposed framework maintains stable leader-follower formations with reasonable computational cost and provides a practical basis for extending vision-based cooperative formation control to real-world multi-UAV systems.

Multilingual large language models (mLLMs) achieve strong performance in machine translation, yet our understanding of the mechanisms by which they transform representations from one language to another remains incomplete. Prior work suggests that translation decomposes into separable processes within an mLLM, where conceptual content is first represented independently, followed by a production into language-specific form. In this work, we show that translation is even more modular than previously assumed and that the output language production in translation processes is actually further separable into a syntax and a surface language process. We construct controlled multilingual datasets that isolate cross-linguistic differences in word-order and use causal interventions and probing to track how representations are transformed during translation. We find that models first construct target-side word-order before realizing the target language surface form. We identify individual attention heads that are selectively sensitive to syntactic transformations while remaining largely invariant to language identity. These results establish the commitment to a syntactic structure as an independent stage in translation, extending prior decompositions and showing how translation is implemented by functionally different components within mLLMs.

Sharing analog integrated circuit designs remains difficult: foundry non-disclosure agreements restrict the process details a design depends on, and the testbenches behind published results are rarely released. We present analog-db, an open-source, versioned database built on a shareable design representation. A domain-specific language captures each design as a process-neutral topology, reusable testbenches, and a machine-readable datasheet under one schema, so a design is shared in full and re-simulates on the process kits it is bound to. A parameterization scheme exposes functional sub-blocks and device sizes as named parameters that carry their matching constraints, making circuits composable and retargetable; a schema-governed contract and queryable catalog let AI design agents discover and reuse them directly. Across the regulator corpus, all 23 circuit-kit bindings on three open kits meet their own recorded specification bands (typical corner, matched devices, no layout) and 10 of 23 meet a common class band. Seventeen of the 23 imported sizings failed their testbenches and closed under a gm/ID sizing loop driven by the annotated sub-block roles, typically within one to three iterations. In a supervised case study, a coding agent working from the released artifacts sized the op-amp cores of a chopper instrumentation amplifier on an open 130nm kit, locating four hand-entry defects and a missing common-mode feedback loop that the sizing-only baseline did not repair. The database holds 68 circuits across sixteen classes, verifiable at schematic level under a tiered harness and tracked on a power/performance scoreboard, released at https://github.com/MacAnalog/spicexplorer-release.

Complete 3D perception from egocentric video requires recovering the surrounding scene and the wearer's full-body motion in a shared metric frame. Existing methods typically address scene reconstruction and motion estimation separately: scene reconstruction methods ignore the wearer, whereas motion estimation methods lack explicit scene geometry and often depend on external trajectories. Joint recovery is challenging because the two tasks exhibit asymmetric visibility and require different prediction paradigms. The largely visible scene supports deterministic geometric regression, whereas the severely occluded body requires generative motion inference. We therefore propose RESELF (REconstructing the Scene and the sELF), a unified framework that couples deterministic metric geometry reconstruction with geometry-conditioned motion generation. RESELF adapts a geometry foundation model pre-trained on large-scale exocentric data to egocentric video using frame-wise scale and relative-pose consistency objectives. The resulting camera trajectory and latent geometric features condition a diffusion model that recovers the wearer's motion. A subsequent closed-loop kinematic feedback stage further refines the camera head while preserving the reconstructed scene geometry. To support training and evaluation, we curate EE4D-JSM from EgoExo4D by aligning egocentric video, sparse metric scene geometry, camera trajectories, and full-body motion annotations. Experiments show that RESELF outperforms state-of-the-art methods designed for the individual tasks across depth estimation, camera tracking, and full-body motion estimation. Code, models, and datasets will be available at https://ka1guan.github.io/RESELF/.

Prospective memory means carrying out a deferred intention at the right future cue while other work continues. Benchmarks now isolate it as an agent skill, yet frontier LLMs still struggle: the best published PM-Bench scaffold reaches only 65.1% Set-F1. We argue that this loop is schema-constrained state tracking rather than open-ended reasoning, and that small models can execute it when the action space is typed. We propose the Prospective Intention Store (PIS) that puts lifecycle logic in code and scoped language work on the model. The scaffold is agentic and training-free: no selector fine-tuning and no trajectory distillation. On PM-Bench, DeepSeek-Chat with PIS reaches 82.9% Set-F1. On Gemma-E2B, Set-F1 is only 4.2% without a store and at most 6.6% under seven retrospective memories, while PIS reaches 66.2%. PIS further reaches 70.1% Set-F1, where retrospective memory methods stay at most 54.4%. PIS sets a new state of the art on this benchmark and enables small models to surpass the published large-model scaffold.

Large language models can produce locally plausible social turns, but fluent next-turn generation is not enough for social simulation. Human encounters such as restaurant lunches and hotel check-ins are bounded social episodes with roles, scripts, material state, obligations, commitments, timing, and closure conditions. We present EpisodeSim, a hybrid LLM-agent architecture that represents classic-AI structures as natural-language control state interpreted by LLM calls. A World Master maintains shared reality, constructs scenes, adjudicates proposed actions, tracks effects and obligations, and controls closure. Experiments with small qualitative ablations on two held-out settings support a design claim: LLM fluency supplies local texture, but coherent social simulation benefits from persistent classic-AI-style scaffolding that organizes behavior over time.

In the standard LM training pipeline, subword tokenisation is applied as a preprocessing step. Subword segmental language modelling is an alternative paradigm in which tokenisation is learned during training, allowing the model to discover subword units that optimise its training objective. In this paper, we present our submission to the 2026 BabyLM Challenge, for which we develop two new subword segmental LMs: SubSegGPT and SubSegDeBERTa. SubSegGPT is a decoder-only model that learns tokenisation during autoregressive pretraining. SubSegDeBERTa is an encoder-based model that jointly learns to generate and tokenise masked words. We train both for the Strict and Strict-small tracks. Our top submission to Strict is SubSegDeBERTa, which achieves notable gains in zero-shot evaluation. Our top submission to Strict-small is SubSegGPT, which outperforms tokenisation-based baselines. Our results show that learnable subword tokenisation can improve sample-efficiency for BabyLM pretraining. We analyse the subword learning dynamics of our models and find that tokenisation gradually converges on subword units that balance morphological alignment and fine-grained segmentation.

Solving problems through repeated attempts is a sequential modeling task: at each step, the solver receives feedback and decides how to revise their solutions. Predicting whether performance improves, plateaus, or regresses across attempts is central to understanding any iterative problem-solving process in both human learners and autonomous agents. Beyond outcomes, modeling what errors persist and how strategies shift across attempts provides deeper insight into the mechanics of sequential learning. Studying these dynamics requires observing many solvers as they attempt, receive feedback, and revise. Programming courses with automated grading provide this setting, as students iteratively submit code to test suites and receive feedback on every attempt. We therefore curate CodeInsight, a large-scale dataset of over 3 million submissions from 3,286 undergraduates across 2 introductory C++ courses in 2 academic years, with test-case-level outcomes, timestamps, and source code. On this dataset, we build a benchmark that evaluates models spanning parametric, sequential, and generative traditions under a shared calibration-and-scoring protocol, including a Recurrent State Space Model (RSSM) adapted to track solver characteristics through discrete latent variables and an LLM-based predictor that generates explicit solutions. The adapted RSSM achieves the strongest predictive accuracy on three of the four courses. The LLM predictor is less accurate but produces full submissions at each attempt, enabling direct analysis of failure modes. We find that the model's coding proficiency is inversely related to predictive performance in this setting, with the LLM better understood as a generative solver conditioned on context rather than a faithful predictor of solver behavior. We publicly release our code and the dataset on request to facilitate future research.

Recent advancements in Vision-Language Models (VLMs) have demonstrated impressive capabilities in static visual recognition and high-level semantic reasoning. However, current embodied exploration paradigms still heavily rely on imitation learning from human-annotated trajectories, which severely limits agents' generalization ability. The key bottleneck of realizing general autonomous embodied agents lies in Generalizable Visually Grounded Exploration: the ability to operate novel devices without manuals or specific training by actively grounding abstract world knowledge into fine-grained visual affordances. Yet, existing benchmarks fail to evaluate this capability: they generally rely on explicit documents and annotated trajectories, neglecting the dynamic Hypothesis-Interaction-Refinement process essential for functional device operation. To bridge this gap, we introduce VGEBench, a comprehensive benchmark designed to evaluate the generalizable visually grounded exploration capabilities of VLMs. Unlike static datasets, we construct a Logic-Driven State Machine framework. This framework simulates multi-turn interaction loops, compelling agents to achieve goals by active visual perception and feedback-driven correction. Experimental results demonstrate that existing VLMs face significant challenges in translating semantic knowledge into physical execution and maintaining long-horizon state tracking.

Self-play is an effective paradigm for language-model self-evolution, but without guidance, solver performance can plateau or decline across rounds. Unguided methods steer question generation with signals such as difficulty, learnability, or diversity. These signals keep questions challenging and varied but do not specify which unresolved reasoning weaknesses later rounds should target. Guided methods obtain direction from external task resources, including human examples, document corpora, or specified difficulty targets, and therefore rely on task information supplied outside the self-play loop. We show that the needed direction can instead be derived from the solver's own failure history. We introduce DiagEvo, whose diagnostician extracts recurring error causes from this history and stores them in a hierarchical error-cause memory. The memory groups related causes under skill nodes and tracks each as Active or Mastered according to self-consistency on targeted questions. The challenger uses these states and recurrence counts to balance cause-targeted generation with free exploration. Double-confidence filtering retains intermediate-difficulty questions only when the most common solver answer has a clear vote lead. DiagEvo derives its curriculum from information produced during self-play, without external task resources. With the default 4B diagnostician, DiagEvo outperforms every baseline in mean accuracy across all nine benchmarks for each of the three solvers: Qwen3-4B, Qwen3-8B, and OctoThinker-8B. On Qwen3-8B, it reaches 72.3% mean accuracy across five mathematical reasoning benchmarks, 4.5 percentage points above R-Zero. Its mean accuracy across all nine benchmarks is 57.4%, 1.1 percentage points above DARC. Ablations show that the hierarchical error-cause memory and double-confidence filtering both contribute to these gains.

Foundation models are increasingly being pursued in particle and nuclear physics, but existing approaches remain strongly tied to individual experiments through detector-specific architectures or pre-training objectives, limiting their reuse across sensing modalities. We show that a point cloud self-distillation framework yields a substantially more general sensor-level pre-training recipe. We show that the same refined architecture and objective can be independently pre-trained with minimal changes on three qualitatively different detector modalities: liquid argon time projection chamber (LArTPC), collider TPC, and water Cherenkov. Using 1,000 labeled images for downstream task adaptation, Panda V2 matches or exceeds specialized foundation-model baselines trained with orders of magnitude more supervision, matching state-of-the-art particle-clustering performance with 70x fewer labeled events on sPHENIX while substantially improving particle identification, and on LArTPC data matching Panda (arXiv:2512.01324) particle reconstruction with up to 1,000x fewer labels. Beyond reconstruction, simple linear probes reveal physically meaningful latent structure associated with particle causality and track curvature.

With the growing scale of multi-agent architectures and large language models, deployed AI assistants are increasingly tasked with reasoning over long, continuous, multi-session conversation histories. Current benchmarks evaluate this dialogue history as information retrieval over long horizon, temporal reasoning, or knowledge updates, while crucially ignoring the fundamental dynamics of human-agent interaction, i.e. how they said it. To address this gap, we present VoiceLongMemEval (VLME) benchmark, where every answer depends on paralinguistic metadata (emotion labels, prosody descriptors, and voice events) attached to conversational turns, which is otherwise unrecoverable from the words alone. Every item passes a three-stage adversarial gate, ensuring that a strong language model fails when given only the transcript. Evaluating leading frontier and open-weight models reveals a pervasive affect gap; providing text-track paralinguistic metadata yields a 0.09 to 0.38 accuracy boost (0.61 to 0.69 when prompted with evidence hints), while standard ASR pipelines systematically discard this signal. Additionally, audio-native models successfully extract these cues directly from speech (0.354 to 0.412 vs. 0.325 blind). Code and dataset will be made available upon acceptance.

Animals and robots encounter mud at the water-land interface. Like sand, mud can stay solid or flow like a fluid. Unlike sand, the yield strength of mud at which solid-fluid transitions occur depends on not only the amount of solid relative to fluid (water in mud, air in dry sand), but also how much coarse grains and fine clay are within the solid. Despite understanding of locomotion on/within dry sand dominated by coarse grains with repulsive normal forces and friction, little is known for mud dominated by fine clay with strong cohesion. Here, we developed methods to prepare uniform mud of controlled, variable yield strength and characterize and track its drift from water evaporation. Compared to other flowable substrates, mud strength measured by upward force during penetration is weaker and can vary more, and mud sticks more during extraction to pull downward, making it more challenging for locomotion.

Computer-use agents are emerging as a paradigm for agentic human-AI interaction, combining language reasoning with multi-modal interface grounding to operate GUIs. Yet their effectiveness for blind screen-reader users in real-world desktop workflows remains unclear. We present a three-week diary study with 8 blind users using OLLA, a screen-reader-accessible CUA prototype, collecting 1,258 commands across 12 applications with screenshots, UI trees, model responses, and action traces. We evaluate GPT-5 during deployment and re-execute the same commands with four additional models. GPT-5 achieved the highest success rate at 52.5%. Trace analysis reveals grounding, planning, constraint-tracking, and termination failures, while interviews reveal beyond-automation needs.

This paper introduces a structural taxonomy for constrained non-convex optimization based on the signature of Lagrange multipliers at KKT stationary points. Leveraging a unified game-theoretic interpretation of eight classical algorithm families--including block coordinate descent, ADMM, generalized Benders decomposition, successive convex approximation, interior-point methods, mirror descent, Frank-Wolfe, and Riemannian gradient descent--we show that the normalized multiplier vector carries an algorithm-independent structural fingerprint. Four scale-free shape features of this vector partition the dual space into five operational regimes: Unconstrained, Resource-Limited, Saturation, Strongly-Coupled, and Hybrid. We establish four structural theorems characterizing the partition: invariance under natural KKT symmetries, local stability under data perturbation with explicit Lipschitz margins from Robinson's strong regularity, codimension-one regime transitions, and the topological identification of the Hybrid regime as the Lebesgue-null boundary of the core regimes. A linear-time classifier is proposed with provable guarantees on correctness, iteration stabilization, sample complexity, and online tracking under data drift. Numerical experiments on 104 mixed-integer nonlinear programs and a downlink beamforming instance validate the theoretical predictions. The framework provides a foundational tool for regime-aware algorithm design and robustness analysis in non-convex optimization.

Extracting scenarios from unlabelled real-world sensor data streams is a critical but challenging task in the development process of automated driving systems (ADS). Automatically sifting through large datasets to spatially and temporally locate critical scenarios can enable scenario-based coverage analysis of ADS datasets. In this paper, we present a method for extracting scenarios from egocentric datasets using scene graphs and Linear Temporal Logic (LTL). We first process egocentric sensor data and HD maps to generate a sequence of scene graphs representing a driving scenario. Next, we use LTL to formally specify driving scenarios of interest, then extract all instances of the scenarios from the dataset using an off-the-shelf model checker, which evaluates the LTL formula against the sequence of scene graphs. Our approach can be used on both simulated and real world datasets. We evaluate the method on the training and validation datasets from Argoverse 2 consisting of 850 15-second real-world driving logs, and several videos of dashcam footage. We demonstrate the effectiveness of our approach for extracting and querying scenarios by evaluating against a rule-based benchmark based on track annotations and HD maps.

Emotional labor is the effortful management of emotional displays to meet social or professional expectations. Personality traits have been correlated with emotional labor strategies, yet research on this link relies almost exclusively on self-report scales administered only in occupational settings. We investigate whether large language models injected with psychometrically grounded personas reproduce these personality-driven selection patterns across everyday social scenarios. We construct the first emotional labor strategy dataset of 500 socially situated events, each offering three behavioral choices corresponding to surface acting, deep acting, and genuine expression. We source 50 fictional characters from a large-scale personality repository and profile each through two parallel tracks: observer-rated bipolar adjective composites and in-character self-report items. Five LLMs evaluate all scenarios under both persona conditions. We find that models align more towards deep acting, and that Conscientiousness and Emotional Stability consistently predict this preference. Entropy analysis confirms that persona reliably influences the output and varies across models and emotions.

Latent language identification is often used to argue that multilingual language models route computation through language-specific states, such as English pivots. However, existing probes infer latent language from different signals, such as the geometry of hidden states or what can be decoded from intermediate representations. Since such claims shape conclusions about how models share and route information across languages, we ask whether these probes measure the same phenomenon or expose distinct aspects of multilingual computation. We study this question across model families, training regimes, domains, tasks, checkpoints, and up to 27 languages. We find that identification probes systematically disagree: the GMM-based representation probe, which draws evidence from hidden state geometry, shows earlier cross-lingual mixing, whereas decoding-based probes, which rely on output-space decodability, retain sharper language-specific and more English-biased signals. These differences track model multilinguality and training progression, but are comparatively stable across domains. Our results suggest a more cautious interpretation of latent language identification, where current probes expose different aspects of multilingual processing, rather than directly revealing a single internal lingua franca.