Latent reasoning performs multi-step inference in continuous hidden states, promising more compact and efficient reasoning. However, these opaque states raise a question of faithfulness: whether the latent reasoning steps drive the final answer. Prior work reports unfaithful behaviors at selected checkpoints, leaving their development during training unexamined. We track behavioral and activation-based evidence across training using verified counterfactual edits and interventions on the latent reasoning states. We find that accuracy can improve while models become less responsive to edits that change the required reasoning path and the correct answer. This counterfactual responsiveness follows distinct trajectories across latent reasoning approaches. On the graph reachability task ProsQA, output sensitivity to noise replacement declines alongside counterfactual responsiveness. Across separately trained binary-choice and open-ended versions of GSM8K math problems, interventions on all latent states show opposite sensitivity trajectories. Guided by these trajectories, we intervene on the training procedure and improve counterfactual responsiveness. These results show that evaluating only a final checkpoint can obscure both when counterfactual responsiveness changes and what the latent states contribute.
Uncertainty estimation is essential for the trustworthy deployment of LLMs: it underpins self-refinement and serves as a decision signal for downstream systems, from verifiers to collaborating agents. However, existing approaches operate at suboptimal granularities: token-level scores lack semantic coherence, while sequence-level scores fail to localize errors. We formalize Span-Level Uncertainty Estimation (SLUE), a new task that targets the natural granularity for uncertainty: semantically coherent text spans, each conveying a single assessable unit of meaning. To address this task, we introduce SPANUQ, a lightweight probe that distills the uncertainty knowledge from expensive multi-sample inference into a single forward pass over LLM hidden states. SPANUQ employs a DETR-style span decoder to simultaneously detect spans and estimate their uncertainty via a Mixture of Beta distribution, trained with a principled combination of Beta NLL regression and contrastive ranking objectives. We construct SPANUQ-BENCH, the first span-level uncertainty benchmark comprising 20K prompts, 293K annotated spans, and continuous soft labels derived from multi-sample claim verification. Experiments on five LLM backbones show that SPANUQ consistently achieves the best span-level uncertainty quality , outperforming the strongest probe baseline and all sampling-based methods while being 10~20x faster. Its DETR-based span detector attains 0.910 F1, surpassing the best heuristic by 39.4%, enabling precise error localization that sequence-level methods cannot provide. The same architecture ports to five LLMs spanning two model families, with one probe trained per backbone, and we additionally observe that sequence-level uncertainty is partially decomposable, suggesting that span-level estimation subsumes sequence-level as a special case. The project page is available damon-demon.github.io/SpanUQ.
The performance of deep learning models crucially depends on the settings of hyperparameters like learning rate, initialization scale, and weight decay. Hyperparameter transfer aims to make near-optimal hyperparameter settings consistent across model scale, so that large models can be optimized by proxy tuning their smaller, cheaper-to-optimize counterparts. While transfer principles are well-studied in the context of dense neural networks in language and vision tasks, they remain comparatively under-explored for graph neural networks (GNNs). We develop and validate a transfer parameterization for GNNs trained with SGD, Adam, and AdamW. Through theoretical scaling analyses and controlled experiments, we show that the proposed parameterization yields stable feature updates, learning rate transfer, and improved performance as width and depth increase. For SGD, we identify graph-dependent first-layer correction factors and show that their use can accelerate early training in graphs with sparse bag-of-words inputs. For Adam, we explore how different message passing normalizations affect early- and late-training transfer behavior, illustrating the importance of message passing normalization and advocating for an associated hyperparameter. For AdamW, we adapt a parameterization that allows for the joint transfer of weight decay and learning rate. Together, these results provide a practical recipe for scaling GNNs across a variety of learning tasks and training scenarios.
Can a simple labeled-pixel contrastive branch improve semi-supervised segmentation with a strong pretrained encoder? We study this question with PixCon, which adds a class-wise memory bank to a DINOv2-Base implementation of UniMatch V2. The bank stores ground-truth-labeled features only where the student prediction agrees with the label. A multi-positive contrastive loss uses these features to train the shared encoder-decoder; the head and bank are discarded at inference. This simplifies earlier labeled-pixel banks by removing confidence thresholds and learned ranking. On Pascal VOC with 183 labels, PixCon improves all three paired seeds by 0.20, 0.64 and 1.83 mIoU points, raising the mean from 87.01 to 87.90. With 92 labels the mean gain is 0.40 points; single-seed differences on Cityscapes and ADE20K range from -0.08 to +0.14 points. A predicted-label contrastive control reaches nearly the same Pascal accuracy, so these results support the added branch rather than an accuracy benefit from its correctness filter. An exact gradient decomposition distinguishes mislabeled-positive counts from their similarity-weighted influence. Separate teacher diagnostics find retained-pixel error rates of 1.8% on Pascal and 10.6% on ADE20K. These diagnostics do not measure the training bank or predict final accuracy. PixCon adds 15.3-19.7% iteration time in the measured runs and no inference computation. We report paired seeds, training cost and the limits of the evidence.
Restoring archival film remains a fundamentally challenging problem due to the absence of paired training data and the lack of standardized evaluation benchmarks. Pristine versions of deteriorated footage are physically unrecoverable, requiring supervised methods to rely on synthetic data that often fail to capture the complex, temporally coherent nature of real film degradation. At the same time, existing real-world datasets are limited in scale, quality, and accessibility, hindering reliable evaluation and fair comparison across methods. We address both limitations with AbsoluteDegradation, a physics-inspired, modular pipeline for synthesizing realistic film degradations, and a new large-scale archival benchmark. The proposed pipeline models the analog-to-digital process as a structured composition of artifact families, incorporating signal-dependent grain, parametric scratches, and temporally coherent camera motion, enabling controlled generation of diverse degradation regimes. In parallel, we introduce a curated dataset of 81,576 high-resolution frames sourced from real archival footage, designed for consistent evaluation under real-world conditions. Together, these contributions provide a unified framework for training and benchmarking restoration models. Experiments across six restoration architectures, evaluated with no-reference image quality metrics, clip-level temporal quality assessment, and a blinded human preference study (93 raters, 1,395 trials), show that models trained with AbsoluteDegradation generalize better to real-world footage: human raters preferred them in 64.7% of comparisons and for every evaluated architecture. The proposed benchmark also reveals systematic failure modes of current methods and of no-reference metrics. We hope this work establishes a foundation for reproducible and domain-authentic evaluation in archival film restoration.
Robot policies that remain broadly competent after deployment can degrade when dynamics, contact, geometry, actuation, or sensing differ from source conditions. We study post-hoc adaptation through structured output-space adapters: compact transformations of samples from a frozen base policy. We introduce Warp, a context-conditioned monotone bijective spline that enables nonlinear sample-dependent reshaping while preserving per-dimension ordering; Translation and Affine provide location-only and location--scale references. Across four ManiSkill tasks with controlled deployment shifts, Warp achieves the strongest recovery in a matched evolution-strategy (ES) adapter study, outperforming controls for spline parameterization, location--scale adaptation, and generic nonlinear capacity. Under PPO, the ordering becomes task dependent, showing that richer adapter structure does not automatically translate into better adaptation under a different optimizer. Finally, we test the practical benefit of this approach on physical SO101 peg insertion. Within an interaction-limited RLPD/SAC pipeline, Warp achieves comparable success to a strong stochastic residual while reducing median successful cycle time by approximately 30%. These results identify structured output-space adaptation as a useful design space for adapting frozen robot policies.
Benchmarks for detecting AI-generated video take their real clips from a few curated collections, so a clip's label is collinear with where it came from. We call a score's dependence on this origin corpus identity, and we measure it on four benchmarks and six real collections. Three probes read a file's declared frame types, the motion vectors its codec stores, and the frames' appearance. Frame-type counts alone separate real from generated at AUC 0.999 on GenVidBench, and re-encoding every clip with one setting at a fixed length still leaves 0.61 to 0.63 on three benchmarks. We introduce a fixed frame-type pattern that removes this leak from every clip of at least 2 s. The same re-encode lowers the ReStraV detector on GenVidBench from 0.85 to 0.60, just above the 0.59 its features reach between two real collections, because its 8 fps repeats frames in real clips. Motion vectors alone separate real collections from one another at AUC 0.51 to 0.98, and only one benchmark ships the two labeled real sources needed to measure this separation within it. In deployment, with each threshold set to flag 5% of its training benchmark's real clips, the motion-vector and appearance probes and linear readouts on ReStraV and WaveRep features each flag 15% or more of another real collection. Choosing the threshold on real videos from the target collection brings the rate back to about 5%, at the cost of catching fewer generated clips. Clearing the real-vs-real level does not guarantee replication: of the six feature sets whose GenVidBench score clears it by 0.15 or more, only WaveRep's two keep their score on AIGVDBench. A benchmark score is therefore best read together with that level, a score on a second benchmark, and false-positive rates on other real collections. The toolkit and the clip lists of every evaluation set are at https://github.com/KurbanIntelligenceLab/vidaudit.
We propose Online Latent prediction with Invariant Views and rEconstruction (OLIVE), a self-supervised speech representation learning framework that jointly optimizes analysis and synthesis objectives. OLIVE couples masked latent prediction and waveform reconstruction through a functional separation within a single encoder: reconstruction constrains early encoder features to retain signal-level information, while masked latent prediction shapes later contextual representations for robust downstream performance, with waveform view augmentations defining the invariances they learn. To the best of our knowledge, OLIVE is the first speech SSL model that directly reconstructs waveforms, retains its vocoder after pre-training, and remains competitive on discriminative tasks. Adding reconstruction largely preserves the speech recognition performance of masked latent prediction alone, while improving waveform reconstruction and speaker tasks. Adding waveform mixup and gain augmentations provides a controllable way to shift the representations further toward speaker, generation, and reconstruction tasks, while remaining competitive on recognition and semantic tasks.
Flicker-Banding (FB) arises from temporal aliasing between a camera's rolling shutter and a display's brightness modulation, degrading screen-captured image readability with color shifts and jagged patterns. Existing single-frame methods with simplified parametric stripe models cannot reliably distinguish these artifacts from genuine texture. To address this, we conduct a systematic analysis of complex FB morphologies and reveal their significant variation across exposure settings, motivating a multi-frame bracketed RAW restoration paradigm. We construct Bricker, a synthetic-real bracketed RAW dataset built via ray-tracing-based physical simulation and an automated multi-exposure capture tool. We further propose BRACE: Bracketed RAW Flicker-Banding Removal, a multi-frame restoration model that utilizes a frequency-aware banding prior and a multi-scale spatial cross-attention modulator (MSCAM) for cross-exposure spatial fusion. We also introduce the Stripe Frequency Consistency (SFC) metric to evaluate banding removal. Experiments demonstrate state-of-the-art performance on both synthetic and real benchmarks. Our dataset and code are available at: https://github.com/ZZH-qwq/BRACE
In this paper, two novel data-driven models based on Kriging and neural networks (NN) are proposed to predict pressure losses across perforated plates with circular perforations in turbulent flows. The models are developed using two sets of experimental data available in the literature. The predictive performance of the proposed models is assessed and compared against widely used empirical formulae. It is found that the proposed models consistently outperform existing empirical models for most perforated plate configurations contained in the experimental datasets. Besides, the predicted pressure losses generally show good agreement with experimental measurements, demonstrating that data-driven approaches based on Kriging and NN provide a feasible framework for modelling pressure losses across perforated plates. Overall, both approaches are promising, despite being trained on a relatively limited amount of experimental data, owing to the scarcity of measurements reported in the literature. To demonstrate the applicability of the proposed models in numerical simulations, two-dimensional channel flows are simulated using the Reynolds-averaged Navier-Stokes (RANS) equations, in which the new pressure-loss models are implemented as a source term in the momentum equations. The RANS predictions are insensitive to mesh resolution and the number of cells used to represent the perforated plate, and they agree well with the model predictions, confirming the applicability of the proposed approaches to practical computational fluid dynamics (CFD) simulations.
Semantic 3D Gaussian SLAM commonly fuses 2D semantic priors into a persistent 3D map using uniform optimization weights. However, such priors are not equally reliable in online mapping: occlusions, unsupported semantic boundaries, and ambiguous ray geometry introduce persistent semantic artifacts into the global Gaussian map. We propose VCS-SLAM, a visibility-consistent semantic fusion framework for RGB-D 3D Gaussian SLAM. Instead of treating all semantic priors as uniformly valid signals, VCS-SLAM evaluates their geometric reliability through visibility consistency, surface-supported boundary evidence, and ray-level conflict uncertainty. The resulting reliability-aware objective suppresses occluded semantic updates, reduces unsupported semantic bleeding, and delays premature label assignment in ambiguous regions. Experiments on Replica show improved semantic consistency, boundary preservation, and reconstruction quality, while ScanNet results confirm competitive tracking under real RGB-D input. Under structured corruption of the 2D priors, VCS-SLAM degrades at little more than half the rate of a comparable semantic Gaussian SLAM baseline and outperforms naive gating, smoothness, and entropy-weighting variants of its own modules, indicating that the gains follow from the specific formulation rather than from reweighting alone. Code is available at \textit{https://github.com/jhakrraman/VCS-SLAM}
Edible insects offer an efficient source of alternative protein, requiring less land and water and emitting less greenhouse gas than conventional livestock. However, their successful integration into the food supply chain demands reliable species authentication to control allergen exposure, prevent adulteration, and meet regulatory standards. Near-infrared spectroscopy provides a rapid analytical tool, but its performance drops when applied to production batches unseen during training due to batch-to-batch variation in spectral measurements. We introduce the Batch-Invariant Spectral Network (BISN), an end-to-end framework that combines a learnable preprocessing module, initialised with Savitzky-Golay filtering, with an unsupervised, entropy-regularised auxiliary branch connected through a gradient reversal layer. The branch uses no batch labels and supplies opposing entropy gradients to the auxiliary head and preprocessing module. Using 2,700 spectra from three species (Acheta domesticus, Hermetia illucens, and Tenebrio molitor) collected across three independent production batches, BISN achieves a mean leave-one-batch-out accuracy of 0.93 (standard deviation 0.04), outperforming the strongest baseline by four percentage points (p=0.002). Further insights gained by using explainable AI confirm that model decisions consistently rely on the lipid and protein absorption regions across all folds, connecting predictive performance to known insect biochemistry. BISN addresses both cross-batch robustness and biochemical interpretability for automated insect species authentication in a controlled benchtop setting. The source code and dataset are publicly available at https://github.com/majharB/bisn.
State-of-the-art medium-range AI weather models can outperform traditional Numerical Weather Prediction (NWP) but require massive training budgets. This restricts usage for under-resourced groups and severely limits fast model iteration. Here we develop Otter Weather, a highly efficient spatiotemporal forecasting model designed to democratise high-performance weather prediction with AI. Evaluated on ERA5 reanalysis data at 1.5{\deg} resolution using standard WeatherBench protocols, the Otter family significantly advances the skill-compute Pareto frontier. The deterministic version outperforms the best NWP baseline by 9.6% at a 24-hour lead time while requiring fewer than 3.5 A100-days for training. It provides a 2x efficiency gain over lightweight AI models and a 100-fold reduction in compute compared to resource-intensive frontier architectures. We extend these efficiency gains into probabilistic forecasting by training via the Continuous Ranked Probability Score (CRPS). Scaling to a larger architecture, Otter-XL achieves a 9.7% CRPS improvement over the IFS ENS baseline. This yields an almost two-fold increase in predictive skill over comparable lightweight models at similar compute budgets. Otter-XL also outperforms frontier architectures like GenCast by over 2%, while using an order of magnitude less compute. Finally, Otter is applied out-of-the-box to a complex acoustic scattering PDE task where it outperforms a state-of-the-art foundation modelling approach, suggesting that the advances made here might apply across a range of scientific domains.
Open domain subject-driven text-to-video (S2V) generation has drawn significant interest in academia and industry. Open domain S2V mainly involves two scenarios: in-domain, which requires retaining the reference subject features as much as possible, and cross-domain, which preserves the intrinsic features of the subject while allowing subject-irrelevant properties to vary flexibly according to the text prompt. Existing methods primarily focus on maximizing subject fidelity in in-domain scenarios, which limits their editability and adaptability in cross-domain scenarios, such as novel styles, semantic combinations, or domain attributes. In this study, we propose that an ideal S2V method should flexibly shuttle between different domains, achieving strong performance in both in-domain and cross-domain scenarios. To this end, we propose DomainShuttle, which could achieve high fidelity and generative flexibility for open domain video personalization. Specifically, we introduce Domain-MoT, which decouples videos and reference features and introduces the domain-aware AdaLN for domain-specific modeling of reference images. We then introduce the Video-Reference DualRoPE scheme, which places reference image tokens and video tokens in separate RoPE spaces to enable precise subject-level spatial modeling, and Cross-Pair Consistent Loss, which aims to extract intrinsic subject features unaffected by irrelevant features. Extensive experiments demonstrate that DomainShuttle achieves significant performance improvements over existing methods, exhibiting high subject fidelity and generative flexibility across diverse open domain application scenarios.
Scaling real-world dexterous-grasp data collection is limited by the sustained human effort required to operate the robot, verify grasp outcomes, and reset objects between trials. Simulation can generate grasp candidates at scale, but their success on hardware remains uncertain. We present AutoDex, a system that collects real-world dexterous-grasp data without continuous human operation. Given generated grasp candidates, AutoDex plans and executes grasps, labels their outcomes, and resets the object to continue collection across stable poses. Multi-view object tracking supports labeling under hand-object occlusion, while residual-torque collision monitoring and pose-directed reset enable the loop to operate without continuous human supervision. Using AutoDex, we collect over 3,500 grasp trials across 100 objects with Allegro and Inspire hands, recording both successes and failures together with synchronized visual observations and robot states. AutoDex reduces human effort by more than an order of magnitude, requiring an estimated 0.2 human-hours to collect 500 trajectories compared with 9.3 for teleoperation. As AutoDex collects trials, it builds a library of physically validated grasps for subsequent execution. In new object placements and obstacle configurations, retrieval from this library achieves 76% success, compared with 34% for candidates screened only in simulation. Code and data are available at https://snuvclab.github.io/AutoDex/.
Discrete text-trigger optimization -- searching for text sequences that, when ingested by a model, steer it toward a specified objective -- underpins model red-teaming (e.g., LLM jailbreaks), as well as auditing and interpretability. However, the current state of discrete optimizers hinders their adoption and progress. First, existing optimizers, when open-sourced at all, are scattered across research codebases tied to specific models, objectives, and problem domains. Second, optimizers proliferate, each requiring engineering overhead to use or extend, and remaining hard to compare head-to-head. Together, these raise the bar for adopting optimizers in existing or new domains, and for advancing them via new strategies. We address these gaps with TROPT, the first open-source framework that unifies discrete optimizers' execution and standardizes their development under a single interface. TROPT makes it easy to customize end-to-end optimization recipes by swapping any component -- models, objectives, and optimizers -- extending its reach across domains and new applications. TROPT currently ships with 40+ optimization recipes -- covering applications such as jailbreaking and probing model internals -- built from 15+ optimizers (spanning white-box to black-box access) and 15+ losses, from foundational to state-of-the-art methods. Demonstrating its utility, we leverage TROPT in several studies: (i) controlled, large-scale experiments comparing and enhancing optimization strategies for LLM jailbreaks, revealing potent-yet-underadopted techniques; and (ii) porting optimizers across applications and domains (e.g., from LLM jailbreaks to corpus poisoning). In all, TROPT significantly lowers the barrier to adopting and advancing discrete text optimization.
3D multimodal large language models (3D MLLMs) describe a 3D object as a whole but cannot address, name, or reason about its parts. Prior part-aware attempts add segmentation decoders, heavier 3D encoders, or bounding-box grammars at substantial parameter cost. We take a fundamentally different path: we reorganize the input token stream so that parts become directly addressable through the LLM's own vocabulary. Our model, 3D-PLOT-LLM, partitions the frozen point encoder's patches into K locally coherent regions and inserts, before each region's patch tokens, a learnable per-region marker and a reserved vocabulary token <part_k>; a Marker-Space Refinement (MSR) module then conditions each marker on its region's spatial statistics and adjacency neighbors. The model thus cites parts in its output and follows prompts that refer to parts by token, a capability absent from prior object-level 3D MLLMs. To probe this interface, we construct PartVerse-QA, a vocabulary-level part-QA benchmark adapted from PartVerse mesh annotations (77K training pairs and 588 held-out queries on disjoint object splits), on which 3D-PLOT-LLM reaches caption-to-slots Jaccard 0.459 and Exact-match 13.78%, with a slot-to-caption GPT-4o judge of 44.68. On the text-description track of the 3DCoMPaT-GrIn part-aware grounded description benchmark, 3D-PLOT-LLM outperforms PointLLM, Kestrel, PARIS3D, and SegPoint on every reported text metric, and ShapeLLM on 3 of 4, with up to +3.03 GPT-4o judge over PointLLM. On Objaverse whole-object captioning, adding PartVerse-QA at Stage 2 yields +0.65 SBERT and +1.85 GPT-4o over PointLLM, and tops PointLLM-PiSA on 4 of 5 traditional metrics despite its part-grounded objective. All with under 1M new trainable parameters on a frozen point encoder, an order of magnitude below prior part-aware 3D MLLMs, and no learned part detector, segmentation decoder, or bounding-box head.
LLM-as-a-Judge has become the dominant evaluation paradigm for language models, but judge validation in practice relies on exact-match agreement, a metric that does not correct for chance and systematically overstates discriminative ability. We present the largest systematic evaluation of LLM-as-a-Judge to date: 21 judges from nine providers across MT-Bench, JudgeBench, and RewardBench, evaluated under three protocols (agreement, consistency, bias audit) over 118 runs and approximately 541,000 individual judgments. Four findings emerge, consistent across the full cohort, including the April 2026 frontier: kappa deflation between exact match and Cohen's kappa is universal (33--41 pp on MT-Bench), judge rankings shift by up to 14 positions across benchmarks, high test--retest reliability (>0.95) coexists with severe position bias (>0.10) in two production-deployed judges (instantiating a consistency--bias paradox), and verbosity bias is small (<0.011) across our cohort under a single pairwise rubric. We distill these into a Minimum Viable Validation Protocol.
Context engineering has emerged as a primary lever for improving AI systems without parameter updates. Recent work showing that textual gradients do not function as real gradients motivates treating automatic prompt optimization (APO) as black-box search. We introduce SPO (Stochastic Prompt Optimization), a framework for stochastic search over prompt space, and compare three strategies of increasing sophistication: error-informed random search, a genetic algorithm with evolutionary operators, and SAGE (SPO via Agent-Guided Exploration), a multi-agent pipeline with diagnostic code execution. Across three benchmarks, no single strategy dominates; effectiveness depends on the interaction of landscape structure with error type. We further deploy SAGE on a mental-health chatbot under a continuous optimization paradigm, where it compounds eight cycles of individually-noisy A/B tests into a statistically robust gain in next-day retention. We argue that coupling qualitative diagnosis with quantitative validation is what makes agentic optimization effective for open-ended task-oriented dialogue.
Visual Geometry Grounded Transformer (VGGT) recovers dense 3D scene structure from multi-view images in one forward pass, but quadratic cross-frame attention limits its scalability. Existing training-free accelerators exploit token, frame, or attention redundancy; we investigate how layer-wise functional roles can guide compression. Our spectral, probing, and causal analyses reveal three regimes: shallow layers lack cross-view structure, middle layers drive cross-view alignment, and deep layers are redundant for dense geometry yet their cross-frame attention remains essential for pose. RegimeVGGT applies layer-wise U-shaped compression along two axes: Saliency-Guided Banded Merging protects geometry- and edge-salient tokens, while Selectively Protected K/V Downsampling preserves cross-frame spatial coverage and the pose-critical path through a phase-shifted spatial grid, a reference-frame anchor, and uncompressed camera/register tokens. Without any training, RegimeVGGT achieves a 6.7x speedup over the memory-optimized VGGT* at matched reconstruction quality. Project page: https://tracyou07.github.io/regimevggt/