Papers

World models are becoming core infrastructure for embodied intelligence, with action-conditioned video generation providing controllable predictions of how scenes evolve after agent interventions. Yet existing models are commonly trained with space-time-uniform mean squared error, allowing abundant background tokens to dominate the gradient while sparse interaction dynamics remain under-optimized; such uniform fitting rewards reconstructing appearance rather than learning how actions change the world. We introduce Causal Action Effect Reweighting (CAER), a general training paradigm that redistributes supervision toward the tokens whose predicted future is causally affected by the action. CAER contrasts the model's own predictions with and without action conditioning to localize these tokens online, then normalizes the resulting effect map into a weight that preserves the total coefficient mass and changes only where it is spent. This online signal requires no external annotations or offline preprocessing, avoids additional data-processing time, and scales naturally with model and dataset size. Experiments across heterogeneous action-conditioned world-model tasks show that CAER converges to better solutions than uniform MSE training, with consistent improvements in the physical consistency, controllability, and visual quality of generated videos.

Deploying a new control policy for voltage control in active distribution grids requires evidence that physical limits will be satisfied before the policy is tested on the physical grid. This assessment is difficult for two reasons. First, simulations cannot capture every disturbance, modeling error, and device interaction present in the real grid. Second, historical measurements reflect operation under existing control policies, whereas a new policy may drive the grid into different operating conditions. To address these challenges, we propose Distributionally Robust Conformal Safety Screening (DR-CSS), a policy-agnostic framework for pre-deployment, scenario-by-scenario screening of a new control policy using historical data and a nominal simulator. For each new scenario, the simulator predicts a future voltage trajectory for the whole grid; DR-CSS then constructs a conformal safety interval around this prediction using historical simulation-to-reality errors. The interval is further enlarged to account for closed-loop changes induced by the deployment of the new policy and its interactions with the remaining controllers. To the best of our knowledge, DR-CSS is the first framework in power systems to combine historical data from an existing control policy with an imperfect simulator for pre-deployment safety screening of a new policy. Experiments on the IEEE 33-bus and IEEE 141-bus systems evaluate the deployment of learning-based voltage control policies and show that DR-CSS identifies all unsafe test scenarios. To reduce unnecessary warnings on safe scenarios, we adapt the safety intervals to different operating conditions and gradually introduce new policies with recalibration after each stage. These extensions increase the informational value of the safety screening and support safer deployment decisions in active distribution grids.

While gender and racial biases in language models have been widely studied, anti-LGBTQ biases remain underexplored, particularly beyond English. Existing benchmarks often do not capture cultural and linguistic variation and rely on gender representations. This paper introduces a multilingual German-English benchmark dataset for the evaluation of anti-LGBTQ biases in language models. It combines community-sourced stereotypes from German-speaking queer individuals with a German translation of WinoQueer. The data is used to evaluate eight language models across sizes and architectures and explore mitigation through fine-tuning on community and progressive media content. Results show that language models reproduce anti-queer stereotypes, with variation across identities and models. Differences between the translated and community-based data highlight the importance of cultural adaptation for multilingual bias evaluation. Fine-tuning reduces bias on average, but not consistently across models and identities. Warning: This text contains examples of anti-queer hateful language and stereotypes.

LLMs are increasingly used for interpersonal advice and as tools for studying social behavior across languages and cultures. A common shortcut for eliciting language- or culture-related variation is to ask a model to answer as a native speaker. We test whether this native-speaker persona reproduces the outputs obtained when models instead generate advice in the target language and translate the response back into English. Using 600 interpersonal advice questions across 13 languages and eight LLMs, we compare native-language generation followed by translation (NL) with native-speaker persona prompting (NP), measuring linguistic style, behavioral scaffolding, and forced-choice action recommendations. We find that NP and NL are not interchangeable. Compared to NL, NP often increases lexical social cues, including affiliation and positive tone, while reducing qualities such as concreteness and social attunement; NP also provides less actionable scaffolding in open-ended advice. In forced-choice scenarios, NP changes which action the model selects, favoring confrontation over redirection, with effect sizes varying across languages, topics, and models. Our results show that cross-lingual elicitation strategy is a consequential methodological choice that can change both how advice is framed and which actions models recommend.

Let $r\geq3$ be fixed, and let $\mathbf{G}_n$ be the set of all simple graphs with vertex set $[n]=\{1,\ldots,n\}$. We consider an exponential random graph model which gives higher probability to $G \in \mathbf{G}_n$ than to $H \in \mathbf{G}_n$ if $G$ has fewer $r$-cliques than $H$. But all graphs in $\mathbf{G}_n$ have positive probability. The degree to which graphs with fewer $r$-cliques are given higher probability is determined by a positive weight $w$. We prove that, asymptotically almost surely as $n \to \infty$, a random graph from $\mathbf{G}_n$ has a vertex partition into $r-1$ parts of roughly equal size, the density of edges between the parts is close to $1/2$, and for every $\varepsilon > 0$ the density of edges within any part is less than $\varepsilon$. The asymptotic structural properties are independent of the weight $w$ as long as it is positive. We also extend the result to the context of several clique sizes, each one with its own weight.

Dakar's residential rental market remains poorly documented despite its economic and social importance: 54.4% of households are renters, compared to 23.3% nationally. This study develops a complete machine learning pipeline to predict residential rents in Dakar, from data collection to model interpretation. An original dataset of 1,507 rental listings was built through systematic web scraping and a documented cleaning pipeline, then enriched with four purpose-built features, including a luxury score and a keyword-based quality score. Five models were compared: linear regression, Random Forest (baseline), XGBoost, and LightGBM optimized through Bayesian optimization with Optuna, using leakage-free KFold target encoding for location. The optimized XGBoost model achieved the best performance with an $R^2$ of 0.847, an MAE of 210,902 XOF, and an RMSE of 324,195 XOF. Feature importance was assessed using native XGBoost gain and SHAP values, revealing a substantial difference in the ranking of location, which appears as a minor predictor by gain but as the second most influential variable by SHAP. This result carries methodological implications for hedonic studies using target-encoded categorical variables. This study provides an interpretable benchmark for Dakar's rental market and highlights several avenues for improvement, including the integration of geospatial features and conformal prediction.

Depression and anxiety in older adults with Mild Cognitive Impairment (MCI) are frequently underdiagnosed due to limited access to care. Multimodal analysis of remote clinical interviews is a scalable screening approach, but existing methods have three limitations. First, they do not correct temporal misalignment across multimodal features extracted at different resolutions, inducing spurious cross-modal associations. Second, remote recordings exhibit uneven modality dropout, but missing values are often zero-filled, making them indistinguishable from valid near-zero measurements. Finally, they do not jointly attribute predictions to modalities, questions, and interview moments, limiting fine-grained clinical interpretation. We propose a Temporally-Aligned, Missingness-Aware, Interpretable (TAMI) multimodal fusion framework. TAMI aligns speech, language, facial, and physiological features within question-answer segments on a shared timeline, encodes modality-level missingness over time, and conditions fusion on question context. In interviews with 49 older adults with MCI, TAMI achieved area under the receiver operating characteristic curve (AUROC) scores of 0.68 (depression) and 0.69 (anxiety). Fine-grained temporal alignment of multimodal features produced the largest performance gain ($\Delta{\geq}0.1$). Multi-level interpretability analysis revealed that depression classification relied on eyegaze and open-ended questions, while anxiety classification depended on eyegaze and head pose, with attribution uniformly distributed across questions. Using only responses to the open-ended questions (5.1min), the depression model achieved an AUROC score of 0.67, which was not significantly different from using the full interview (19min) ($p>0.05$). Our findings support designing interview protocols centered on open-ended questions for depression screening in older adults with MCI.

Fairness audits in clinical Artificial Intelligence convert continuous fairness metrics into binary pass-or-fail verdicts against operational thresholds, where hospital governance boards, payers, and regulators act on the resulting verdicts. Such audits are repeated over time and across hospital sites, thus the same verdict can flip between pass and fail across audits. Existing uncertainty methods such as Bayesian posteriors, bootstrap confidence intervals, and permutation tests address verdict instability only at the continuous-metric level. Converting metric-level uncertainty into a verdict-stability claim remains a manual step that scales poorly across the (model, metric, attribute) cells an audit covers. Existing uncertainty methods also leave open whether bias-mitigation steps, such as reweighing or per-group threshold shifts, yield a stable passing verdict at the cost of model discrimination measured as AUROC or AUPRC.To address this verdict-stability gap, we propose VFR-Audit, a framework built around the Verdict Flip Rate (VFR), a scalar bounded between 0 and 0.5 that measures the probability of verdict reversal under stratified bootstrap resampling. VFR-Audit reports VFR alongside three reliability axes, namely within-cohort resampling stability, audit-size sensitivity, and cross-hospital verdict agreement via Fleiss' kappa.

Interactive lesion segmentation in whole-body PET/CT requires a model to provide a strong initial prediction while also responding efficiently to sparse corrective scribbles during inference. This setting is particularly challenging because tracer distributions, physiological uptake patterns, lesion appearance, and acquisition characteristics differ substantially between FDG and PSMA studies. We present TRIAGE, Tracer-aware Refinement via Interactive Anatomy-Guided sEgmentation. The core backbone is a 3D STU-Net initialized through masked autoencoding pre-training with an asynchronous masking strategy, aiming to learn transferable anatomical and cross-modal representations before task-specific fine-tuning. In parallel, we train an auxiliary organ segmentation model whose predictions provide explicit anatomical context and help distinguish physiological uptake from malignant lesions. A dedicated tracer classifier first routes each study to an FDG- or PSMA-specific branch. Within each branch, a first-stage segmentation model consumes CT, PET, and organ context to generate an initial lesion mask. The initial prediction is then combined with cumulative foreground/background scribbles and refined by a second interactive segmentation network. The FDG and PSMA branches share the same overall processing pipeline but are trained independently to account for tracer-specific appearance and error modes. We additionally employ curriculum-style training and model ensembling to improve robustness across interaction steps and heterogeneous cohorts. Experiments are conducted using the official AutoPET V data and ten-fold split; quantitative results, ablations, and final test-set performance are left as placeholders to be completed after the challenge evaluation. Code: https://github.com/Liiiii2101/AUTOPET2026-MEDAI.

Large language models can often generate plausible mathematical reasoning traces, but reliably identifying the correct solution among multiple candidates remains a key challenge. Existing test-time reasoning pipelines typically rely on text-based verifiers that re-read each generated solution, making verification an expensive component of inference. Prior work has shown, however, that LLMs often encode correctness-related signals in their internal representations, including awareness of when their own answers are likely to be wrong. Building on this observation, we introduce HSRM, a lightweight hidden-state reward model that verifies candidate solutions by directly reading the generator's internal representations rather than re-processing its text. HSRM extracts hidden states from a frozen generator at reasoning-step boundaries and uses a small Transformer encoder to rank candidates. It is trained from self-generated trajectories with outcome labels, requiring neither human-written process supervision nor a large pretrained verifier. Across four mathematical reasoning benchmarks, HSRM matches or outperforms a 55M-parameter text-only energy verifier in 15 of 16 generator–dataset settings while using only about 2M parameters, providing an efficient alternative to text-only verification by reusing representations already computed during generation.

Background and Objective: Quality control is a prerequisite for whole-slide image analysis, yet the benchmarks on which quality-control methods are compared share four properties that make their reported differences hard to interpret: few independent slides, annotation concentrated in a minority of them, pooled ratio metrics with no closed-form standard error, and a single inherited train/test partition. We propose a reliability protocol for such benchmarks. Methods: The protocol quantifies four sources of variability - test-set sampling, training stochasticity, partition composition, and undocumented preprocessing - a claim is reportable only if it survives all four; three of the four cost minutes of compute. We apply it to an independent reconstruction of a published diffusion-based artifact detector, evaluated on the original 24-slide partition and against a supervised baseline. Results: The method's central mechanism reproduces: the auxiliary contrastive term improves pooled F1 from 0.673 to 0.688 and replicates under a second seed (+0.0156, p = 0.031; +0.0190, p = 0.005), although it acts on pen marking rather than the artifact types cited to motivate it. Its comparative claims do not: differences between design variants, and against the supervised baseline, fall inside the uncertainty of the evaluation. Four of 24 slides carry 70% of scored annotated pixels, giving an effective sample size of 6.2, and the inherited partition sits at the 7th percentile. An unreported tissue-restriction step excludes 41.4% of out-of-focus annotation against 2.6% of air bubble; such a gate is confounded with blur by construction. Conclusions: Small-cohort benchmarks support far weaker conclusions than current reporting implies. The four checks are cheap enough to accompany any evaluation on such a resource and separate reproducible effects from differences the evaluation cannot resolve.

Composable scene modeling aims to recover a real indoor scene as complete, editable object assets arranged as observed, giving robot simulation and embodied AI a simulation-ready replica of the real environment whose objects can be manipulated individually. Existing pipelines decompose the task into three steps—parse the observations into instances, generate an asset for each, and place each asset back—but every step presumes an input that a cluttered capture rarely provides: accurate instance geometry, unoccluded views, and assets that accurately match the observations. We propose Lucida, which keeps this order but redistributes the requirements, so each step consumes only what a real capture reliably provides and precision is reached at the end of the pipeline rather than demanded at its start. Lucida parses the video into a scene graph whose nodes carry per-instance multi-view evidence, generates a complete asset for each instance from its evidence, and places assets with GizmoAct, a VLM policy that casts placement as multi-turn GUI interaction, manipulating the object's gizmo in a closed loop and deciding itself when alignment is reached. Across scene-level 3D object detection, object pose estimation, and scene reconstruction, Lucida improves mAP over Boxer by 69% on R2S-Scene, raises ADD-SB@0.05 from 57.8% to 83.4% on CA-1M, and increases scene F-Score from 0.794 for SAM3D to 0.924.

Geospatial foundation models have emerged as state-of-the-art methods for downstream Earth observation tasks. However, existing pretraining methodologies process imagery through a single-concept lens, failing to capture the highly compositional nature of complex satellite scenes. We propose a composition-aware pretraining framework that explicitly encodes fractional land-cover mixtures. Each satellite image cell is mapped to a histogram representing its fractional land-cover distribution, which we term the "composition target". These targets serve as the primary prediction objective and are distilled into the backbone using Earth Mover's Distance. Experimental evaluation shows that composition-aware pretraining yields substantial gains on region-level understanding tasks requiring semantic similarity judgment, including zero-shot image retrieval and scene classification, while remaining competitive on tasks requiring fine-grained spatial precision, such as segmentation and object detection. With a 36.8M-parameter backbone, our framework outperforms SatMAE and Prithvi-EO-2.0, which contain 303M and 600M parameters, respectively, in most retrieval and scene classification settings. On the fine-grained ForestNet-12 dataset, a rigorous testbed for compositional discrimination, our method boosts baseline mAP@10 from 0.279 to 0.434, a 55.6% relative improvement, providing direct evidence for the effectiveness of explicit composition modeling. The code implementation can be found at https://github.com/05kashyap/GFM_Composition_Pretraining

Natural-language tasks can elicit different verdicts from protocol-following evaluators that receive the same declared information. We study aggregate disambiguation systems (ADSs). Given a task and a candidate solution, each evaluator casts a binary vote on whether the solution should be accepted, and the system aggregates the votes of a finite panel. The target is protocol reproducibility relative to an explicitly declared evaluator reference, not semantic truth. We separate fixed finite censuses, probabilistic evaluator populations, and growing-census limits, since their endpoint laws and guarantees are not interchangeable. In the population setting, we use finite samples to estimate how often a finite panel reaches the same decision as the declared evaluator population. We provide a lower confidence bound on the fraction of candidate solutions for which the disagreement probability is at most a chosen tolerance. The calculation accounts separately for sampling candidate solutions and sampling evaluators. The construction permits arbitrary dependence among columns induced by shared evaluator rows and uses exact binomial intervals at the evaluator layer and an exact one-sided binomial inversion at the generator layer. Simulations check the implementation against known population coverages and expose power limitations.

Obtaining data from neuromorphic sensors and processing it with Spiking Neural Networks is a promising solution to lower the energy cost of artificial intelligence. The current rarity of natively neuromorphic datasets promotes the development of software tools to translate input sensory data into spikes. However, highly bio-mimetic simulators can be challenging to implement on digital hardware. In this work, we evaluate the neuromorphic encoding and subsequent classification of audio into spikes using a non-learnable, high-level, programmable encoder targeting hardware implementation on FPGA. We quantify the pipeline's efficiency with hardware-agnostic metrics based on the quantitative spiking activity. Our study focuses on the simultaneous optimisation of encoder and classifier: the first provides efficient and informative data so that the latter achieves a better performance with an overall lower energy cost at learning and inference. This work introduces the first end-to-end neuromorphic spike-encoding and evaluation of the TIMIT dataset. Our simple feedforward network reaches a classification accuracy of 99.77% on a spike-encoded Heidelberg Digits, overcoming the neuromorphic state of the art on this benchmark dataset.

Production agent skills are directory bundles, not isolated prompts. The root is loaded at activation; references, schemas, scripts, assets, and nested subskills are loaded only when an execution path needs them. Compressing only the root misses most deployment cost and may move branch-specific details into the always-loaded context. Flattening instead destroys progressive-loading boundaries. We introduce \method, an evaluation-free compressor for complete, progressively loaded skill bundles. It leaves the agent harness unchanged and emits an ordinary directory. The method combines two safeguards. First, it compresses across files, removing content from a reference or subskill when the root or a declared environment contract already provides it. Second, it preserves routing, so every required file and directly callable entry remains reachable after rewriting. Users can configure \method along two independent axes. One-Shot mode rebuilds the full bundle; Continual mode reuses state and applies Zip-on-Write after each evolution patch. Persistent compression rewrites the shipped bundle to reduce storage and runtime context. Transient compression keeps that bundle byte-identical and builds a task-specific view, reducing only per-run context after build cost. Entry contracts mark private, public, and conditional resources; a multi-entry audit preserves standalone public subskills. On a production content-moderation skill evaluated by our industrial multi-round harness, \method removes \hl{38%} of skill bundle tokens and \hl{10.4%} of end-to-end per-run tokens with no quality loss, while an unprotected 71% configuration loses up to 26 accuracy points to one-sided false positives. On a multi-entry bundle, \method effeciently reduces token cost while near-perfectly preserving every route and public entry.

Before an intelligent system can decide whether an action is allowed, it must first know which rules the action has approached. A single proposed action can implicate several policy regimes at once. Their requirements may stack, overlap, or qualify one another, yet many remain written in natural language while the action itself arrives as an incomplete description of intent. The first problem is not judgment. It is attention. Policy-centroid routing creates a layer before adjudication. It compresses expressions within each policy regime into one or more representative centroids, places the proposed action in the same semantic space, applies a declared measure, and routes every regime crossing a declared threshold to authoritative review. Several regimes may trigger at once. The output is a review agenda, not permission, prohibition, legality, breach, compliance, certification, or enforcement. The paper develops six falsifiable propositions and seven follow-on studies comparing the hypothesis with structured workflows, lexical and semantic retrieval, hierarchical and direct classification, and selective prediction under matched review burden. The studies are designed to identify where policy geometry recovers applicable regimes, where compression loses rare or overlapping obligations, and where the mechanism should abstain. The paper includes a synthetic worked example and reports no empirical efficacy result.

Large language models (LLMs) have become an essential tool for assisting developers, yet we still lack knowledge on ways to effectively support their interactions during development activities. That is, the quality of interactions with a chat-based LLM still strongly depends on how developers phrase prompts and which information they include. Our goal is to evaluate whether interventions into these interactions with LLMs have an effect on software developers—be it harmful or beneficial. To this end, we conducted a four-month longitudinal study with third-semester computer science students working on a full-stack Web development project using chat-based LLMs under three conditions: (1) a context-aware group received intent-based conversation augmentation, (2) a proactive group received follow-up suggestions and tailored advice, and (3) a control group without intervention. Our augmentations are minimal: (i) to reduce confounding factors and (ii) to isolate treatment effects. Analyzing interaction logs and user surveys revealed no major differences in interaction patterns, indicating no detectable harmful effects in the measured outcomes when intervening in interactions. Moreover, we observed trends of increased satisfaction with the proactive treatment. The results indicate that even with minimal interventions, dynamic guidance mechanisms for developer-LLM interactions show observable effects, such that more severe augmentations may have the potential to substantially improve developer satisfaction.

In large-scale e-commerce retrieval, dual-encoder retrievers are op- timized for contrastive similarity, whereas downstream rerankers capture finer-grained relevance preferences; this objective mis- match limits end-to-end retrieval quality. Reinforcement Learning offers a way to use reward-model feedback for retriever adaptation, but we observe that standard policy-gradient updates can degrade embedding geometry, especially when the document index must remain frozen due to industrial constraints. To address this, we propose PAO (Positive-Advantage-Only), a selective RL optimization method. Our analysis reveals that in- discriminate penalization of negative samples (pushing away) in a frozen high-dimensional space disrupts pre-trained semantic man- ifolds. PAO selectively applies gradient updates only to retrieved items with positive advantages, effectively pulling query embed- dings toward high-reward regions while preserving global topo- logical stability. Experiments on both a massive industrial dataset and public benchmarks demonstrate that PAO significantly outper- forms standard RL and distillation baselines.

Large language models (LLMs) trained only on text and code can sometimes generate programs that draw recognizable images. However, it is unclear whether this reflects an internal representation of 2D spatial layout or simply the ability to translate spatial descriptions into code. We introduce Autoregressive Mosaics (AM-Bench), a benchmark that separates these factors: First, a translation task gives a model a fully specified geometry of a picture in words as a prompt and asks for the code that produces it. Second, a layout task requires the model to compose an image from an underspecified prompt. Across eight open-weight text-and-code-only models, all models reliably translate specified geometry into code, but their open-ended layout performance differs substantially, indicating that these differences are not explained by code-generation ability alone. An output-medium ablation further shows that the interface or medium of expression that the model uses matters: replacing procedural code with raw SVG improves layout scores across all models. Finally, probing model activations shows that a coarse layout plan is present before generation, but reflects only the layout implied by the prompt. During generation, models track the evolving geometric state instead of executing an initially fixed plan. Overall, these results show that 2D spatial performance in text-only LLMs depends on both the model and the output medium, and is not explained by code-generation ability alone.

Assessing claim check-worthiness is an essential first step in automated fact-checking pipelines. This work is motivated by a real deployment challenge at an early-stage startup: running large language models (LLMs) over every incoming claim is cost- and latency-prohibitive, yet smaller models sacrifice accuracy. We propose NN-PPI, a pointwise extension of Prediction-Powered Inference (PPI) that calibrates model predictions at inference time as a lightweight post-hoc layer, without re-training the underlying model. NN-PPI achieves weighted F1 gains ranging from 12% to 33.80% depending on the size and performance of the baseline model, bringing SLMs on par with larger LLMs. Beyond few-shot SLMs, NN-PPI further improves a production-deployed fine-tuned model, demonstrating that residual calibration is complementary to supervised fine-tuning. By recovering LLM-level accuracy from models that are an order of magnitude cheaper to serve, it makes accurate check-worthiness detection substantially cheaper to operate at scale. Our code and data can be found at https://anonymous.4open.science/r/arr-claim-worthiness-F237.

Small-object detection under long-tailed data distributions is a fundamental yet challenging problem in multimedia. Railway Foreign Object Detection (RFOD) epitomizes this challenge with easily confused small intrusions and scarce samples. To address these issues, we propose a generative-augmented detection paradigm that leverages multimodal image generation to enrich the feature space of rare and small objects. We first construct RailGen, a multimodal image generation agent based on large models. Under semantic constraints, RailGen automatically invokes tools to generate railway scenes, calibrate intrusion positions, extract foreign objects, and fuse them into realistic intrusion effects. This process produces high-quality synthetic samples that effectively densify the feature representations of tail classes and complete the small-object feature space. Within this paradigm, we further propose FocalDEIM, a detection framework designed to enhance training with generated data. FocalDEIM improves dense matching with Focal Modulation for better small-object discrimination and adopts Focal Loss to emphasize hard samples, thereby alleviating blurred inter-class boundaries in complex railway scenes. Experimental results demonstrate that RailGen can generate high-quality small-scale foreign objects, reducing the object pixel area by up to 58x and 13.85x on average. Equipped with these challenging samples, our paradigm surpasses the baseline DEIM by 5.6% and 7.5% in mAP@50 and mAP@(50-95), respectively, and outperforms existing state-of-the-art methods. Ablation studies verify RailGen's feature-space enrichment and FocalDEIM's boundary discrimination. The paradigm provides an effective multimodal generative solution for long-tailed small-object detection in safety-critical applications.

Multimodal Sentiment Analysis (MSA) is a fundamental component of affective computing that aims to decipher complex emotional states by integrating verbal content with non-verbal cues including vocal intonation and facial micro-expressions. While recent disentanglement-based approaches have advanced the field, their potential is hindered by two methodological challenges. First, static computation graphs process all samples indiscriminately regardless of semantic complexity, which leads to suboptimal representation for diverse emotional expressions and contextual scenarios. Second, generic contrastive objectives often neglect the intrinsic ordinal hierarchy of sentiment intensities. To systematically address these limitations, we introduce Multimodal Adaptive Expert Selection with Text Routing and Ordinal prototype optimization (MAESTRO), a novel framework designed to dynamically orchestrate and refine multimodal representations. Drawing inspiration from an orchestra conductor, we design a Text-Guided Hybrid Mixture-of-Experts (MoE) mechanism. Unlike static fusion, this module utilizes linguistic context as a routing signal to dynamically activate specific audio-visual experts, thereby resolving cross-modal ambiguity through adaptive feature enhancement. Furthermore, to capture fine-grained sentiment gradations, we propose an Ordinal-aware Prototype Contrastive Learning (O-PCL). By incorporating distance-based penalties into the prototype learning objective, O-PCL enforces a structured latent space that preserves the natural order of emotion. Extensive experiments on the CMU-MOSI and CMU-MOSEI benchmarks demonstrate that MAESTRO achieves state-of-the-art performance, and qualitative analysis further confirms the interpretability of our dynamic routing paradigm.

LLM agents are increasingly used to run autonomous ML experiments, iterating on target metrics with little human oversight. Prior work has documented reward hacking in these environments, bringing into question the validity of produced research and the broader safety case for AI R&D. Existing benchmarks do not measure exploits that live in the data or the modeling task itself. We introduce BAITBENCH, a suite of three synthetic tabular ML tasks that each contain a shortcut that allows agents to inflate the public test score but fail on a hidden test set. Since the shortcut is optional and using it breaks no stated rule, BAITBENCH measures how often models exploit the shortcut to achieve inflated scores. Across seven frontier agents scored by our two-stage judge pipeline, 57.1% of runs exhibit reward hacking, with five of seven above 50%. Agents cheat even under a second condition where they are prompted not to -the mean cheating rate remains above 50%. We release BAITBENCH, along with the judge implementation, and an annotated dataset of transcripts containing reward hacks as a testbed for evaluating reward-hacking mitigations head-to-head.

Railway foreign object detection (RFOD) is critical to safe railway operation, yet scarce real positive samples incompletely represent task-relevant variations in object scale, intrusion relation, railway scene, illumination, and adverse weather. Existing synthetic augmentation can improve RFOD detection, but its gains lack an explicit account of the task-relevant deficiencies complemented by the generated data. We therefore introduce RailSyn, a diagnosis-guided framework comprising a real-referenced Inspector and a requirement-aligned Generator. The Inspector constructs a variable-radius empirical cover from finite real observations to localize candidate completion regions and profile synthetic pools. The resulting audit identifies railway-context, intrusion-semantic, and visual-consistency requirements; the Generator addresses them through domain adaptation, agent-planned placement and physical contact relations, and plan-consistent conditional refinement. Using the Inspector, we further trace representation-space changes across generation variants; the complete system attains a local-shell occupation of $C_{gap}$ to 13.64%, which measures generated coverage of real-derived completion regions. Extensive experiments show AP50–95 gains of up to 4.9 points and consistent improvements across nine mainstream detectors, demonstrating broad cross-architecture utility.

We present SingProbe, an open intrinsic guardrail framework for generation-time monitoring of LLMs. Intrinsic guardrails reuse hidden states already produced by the base model during autoregressive decoding, rather than relying on an independent model to repeatedly process generated text. While this route has been explored in industrial systems, the community lacks a broadly reusable open stack that combines cross-model guard adaptations, unified training methods, serving integrations, and systematic evaluation resources. SingProbe is designed to provide this missing layer and uses a lightweight probe to continuously produce query-intent, response-safety, and hallucination-risk signals during decoding. This report describes the full intrinsic-guardrail stack: training methods, serving integrations with SGLang and vLLM, and adapted guard models for 29 open-source base models across diverse families and scales. We also introduce SingStreamBench, a benchmark that measures whether streaming guardrails remain inactive on benign prefixes while promptly detecting emerging unsafe content. Across evaluations of safety, streaming detection, hallucination detection, false-positive robustness, online monitoring, and runtime overhead, SingProbe provides performance competitive with, and in several settings stronger than, state-of-the-art standalone guardrails and specialized hallucination detectors, while adding less than 0.5% serving overhead in our implementation. Beyond passive monitoring, we show that intrinsic guard signals can guide constrained decoding and selectively activate medical-risk interventions in SingProbe-Med. By open-sourcing our infrastructure, training methods, and model adaptations, we aim to facilitate the broader adoption and deployment of intrinsic guardrails, as well as further research in this direction.

Large language models are increasingly used as agents for multistep retrosynthesis, raising the question of how much their search policy contributes independently of the underlying reaction model. We investigate this question in a biological setting through rule-based retrobiosynthesis: a deterministic biochemical engine generates the same validated transitions for every method, searching for routes that terminate in metabolites available to an Escherichia coli chassis, while the policy only selects which frontier molecule to expand next. Prompted and LoRA-tuned Qwen2.5-7B policies use a strict choice-only interface. The fine-tuned policy reaches $65\pm1$% solve rate at 10 expansions on LASER versus 59% for MCTS, and at 200 expansions reaches $78\pm1$% versus 75% on LASER, $88\pm3$% versus 80% on the RetroPath RL Golden benchmark, and $63\pm2$% versus 45% on the BioNavi-NP benchmark. Fine-tuning also consistently outperforms direct prompting. These results show that route-supervised frontier selection can improve budgeted search without altering biochemical generation, although performance remains dependent on frontier construction and reaction ranking.

Long-tailed distributions are prevalent in real-world semi-supervised learning (SSL), where pseudo-labels tend to favor majority classes, leading to degraded generalization. While many long-tailed semi-supervised learning (LTSSL) methods have been proposed, the mechanisms by which they implicitly debias logits remain poorly understood. In this work, we revisit LTSSL through the lens of learning dynamics and provide a theoretical characterization of logits debiasing. Specifically, we derive a step-wise decomposition of the logits updates, showing that predictions are dominated by class-imbalance bias that reliably reflects label priors. To expose this effect, we use the logits of a task-irrelevant baseline image as an indicator of accumulated bias and prove that they converge to the class prior. This provides a unified view where LTSSL remedies such as logit adjustment, reweighting, and resampling correspond to reshaping gradient dynamics. Based on this insight, we propose DyTrim, a principle-based dynamic pruning framework that reallocates gradient budget through class-aware pruning on labeled data and confidence-based soft pruning on unlabeled data. We provide theoretical guarantees that DyTrim reduces class bias and improves generalization. Extensive experiments on standard LTSSL benchmarks show consistent gains across architectures and methods. Code available at: https://jiajun0425.github.io/DyTrim

Large language model (LLM) agents are increasingly deployed in user-facing services that require iterative tool use under dynamic business conditions. Reliable evaluation is essential for sustained improvement: it must reveal capability deficiencies, inform priorities, and assess interventions. Yet industrial agent service unfolds both through the iterative trajectory of a current request and through continued user interaction. Final-outcome assessment can therefore obscure where deficiencies arise and whether later service remains aligned with context from earlier exchanges. We propose ATLAS, a dual-horizon diagnostic evaluation framework for industrial tool-use agents. At the request horizon, trajectory-wise diagnostic signals relate deficiencies to execution locations and capability concerns. At the interaction horizon, user-wise signals assess whether service remains responsive across continued interaction. Together, these views provide structured diagnostic evidence for analyzing execution deficiencies and sustained service behavior. ATLAS instantiates them as executable signals with explicit evidence scopes and decision boundaries. LLM judge interfaces are calibrated against high-confidence references from real business logs; when needed, their decision behavior is distilled into efficient diagnostic models for lower-latency, lower-cost evaluation. The resulting feedback supports policy optimization. We evaluate ATLAS on Meituan Xiaotuan production traffic. Offline experiments assess diagnostic-signal fidelity and replay-based policy improvement, while online A/B experiments show concurrent gains in user engagement, downstream business outcomes, and sampled human-audit quality.

Learning materials properties from scarce labels and unlabeled crystals is a central challenge for data-driven materials discovery. We present SemiMat, a controlled benchmark for semi-supervised materials property regression, and MatRank, a reliability-weighted objective for continuous pseudo-label uncertainty. SemiMat fixes labeled and unlabeled crystal inputs, graph-backbone interfaces, validation-only checkpoint selection, held-out test reporting, normalized MAE (NMAE), and method-rank summaries across six scarce-label tasks, four graph backbones, and five predefined split runs. MatRank builds pseudo-targets from labeled anchors, weights them by local reliability and weak-prediction agreement, trains weak and strong graph views consistently, and adds ranking signals so that unlabeled crystals shape both values and candidate order. Across the retained 24 backbone-task blocks, one fixed MatRank objective gives the lowest aggregate held-out test NMAE (0.896) and best average method rank (2.208). The component, OOD, and generated-pool diagnostics identify where the gain is reliable and where further screening evaluation remains necessary. Code is available at https://github.com/littlepeachs/SemiMat.