Transformer
-
Let Agents Design How They Search: Automatic Interface Optimization for Agentic Search
2026-10-21 05:56:30Sangjune An, Jaeyoon Jung, Yumin Choi, Jinheon Baek, Junmo Cho, Sung Ju Hwang, Hae Beom LeeAbstract
Agentic search depends on the interface through which a language-model agent accesses and explores a corpus. Existing interface designs are fixed by hand and applied unchanged to every corpus, although effective designs may depend on the corpus. We introduce SearchWeaver, a framework for automatic search interface optimization guided by downstream search performance. A proposer uses execution trajectories and task rewards to jointly revise corpus artifacts, executable tools, and usage instructions. This pro cess optimizes what information is prepared in advance, which operations tools perform, and which decisions remain with the search agent. With GPT-5.4-nano, the optimized interfaces improve answer accuracy from 71.6% to 78.4% on BrowseComp-Plus and average nDCG@10 from 45.7 to 55.0 across four BRIGHT subsets over Direct Corpus Interaction. These findings highlight the value of jointly optimizing corpus artifacts, executable tools, and usage instructions to develop effective search interfaces.
Abstract (translated)
URL
https://arxiv.org/abs/2610.29250
PDF
-
PaceVLA: Adaptive Temporal Action Quantization in Vision-Language-Action Models
2026-10-21 05:55:50Hojin Jeon, Jimin Lee, Taekwan Kim, Nayoung Oh, Taeyoung Kim, Myungkyu Koo, Jinmo Koo, Joonwoo Ahn, Jinwoo ShinAbstract
Fast and reliable manipulation is crucial for practical robot deployment. Although Vision-Language-Action (VLA) models have recently shown remarkable progress as generalist robot policies, their execution often remains temporally inefficient. A key source of this inefficiency is that policies learn fine-grained temporal patterns from teleoperated demonstrations, requiring many low-level action steps. Temporal action quantization offers a simple way to accelerate execution by aggregating consecutive low-level actions into temporally coarser actions. However, applying action quantization without considering when fine-grained control is required introduces a speed-accuracy trade-off: accelerated actions may efficiently reduce redundant steps in some phases, while discarding critical motion details in precision-sensitive interactions. To address this, we propose PaceVLA, a unified framework for adaptive temporal execution in VLAs that integrates two coupled components: (1) mode-specific action decoders built on a shared action expert, supporting multiple temporal granularities and prediction horizons, and (2) a quantizability-guided mode selection scheme which decides whether coarse execution is appropriate, and a context-conditioned router that selects an execution mode at each inference step. Experiments across three multi-task simulation benchmarks and six real-world tasks show that PaceVLA improves execution efficiency while maintaining task success rate across multiple embodiments, including long-horizon and dexterous manipulation.
Abstract (translated)
URL
https://arxiv.org/abs/2610.25961
PDF
-
Failure Modes of Vision-Language Models for Human Activity Understanding in Thermal HRI
2026-10-21 05:49:05Han Hu, He ChenAbstract
Thermal cameras enable robots to perceive human activities in darkness while revealing less personally identifiable appearance information and fewer details of the surrounding environment than RGB cameras. As vision-language models (VLMs), originally developed for RGB imagery, are increasingly integrated into embodied systems for scene understanding and high-level planning, thermal sensing offers a promising way to extend activity perception to low-light and privacy-sensitive settings. However, the failure modes of VLMs in thermal human activity understanding for human-robot interaction (HRI) remain underexplored. We fill this gap by evaluating seven locally deployable VLMs, including models developed for physical AI and robotics, on two person-centric thermal datasets. Our diagnostic tasks examine fine-grained action recognition, activity timeline reconstruction, and the ability to withhold unsupported answers. Paired RGB recordings serve as a performance reference, while controlled variations in frame sampling, thermal input presentation, and reasoning are used to assess whether the observed errors can be mitigated. Our results show that models can recognize broad activity categories but struggle with fine-grained actions such as coughing, sneezing, and yawning. They also produce inaccurate event timelines and hallucinate actions involving objects. Simple thermal input preprocessing and additional reasoning offer only limited, model-dependent gains. Abstention behavior also varies across models even under explicit instructions, limiting their reliability for practical robotic use.
Abstract (translated)
URL
https://arxiv.org/abs/2610.29244
PDF
-
Encoded but Unused: Bounding the Sources of VLM Errors
2026-10-21 05:47:57Anat Veikherman, Mor Ventura, Nitay Calderon, Lihi Zelnik-Manor, Roi ReichartAbstract
In this paper, we investigate why vision-language models (VLMs) produce errors by decomposing them into two sources: visual-representation failures, in which the visual evidence required for the correct answer is not encoded, and visual-utilization failures, in which this evidence is encoded but not successfully used. We propose to estimate the visual-utilization failure rate using lightweight probes and derive theoretical bounds on the true rate that account for probe imperfections. We introduce VELens (Visual Evidence Lens), a rank-based probe that traces visual evidence for both a target object and its ground-truth attribute, such as color, material, or shape, across VLM layers. We evaluate VELens on over 6,000 images using LLaVA-1.6-7B and Qwen2.5-VL-7B. Our method achieves ROC-AUC scores of 0.95 and 0.91, respectively, and outperforms alternative probes in both low-resource and unseen-attribute settings. Using it to diagnose errors, we estimate that visual-utilization failures account for 84.4% of LLaVA errors and 70.0% of Qwen errors, with conservative lower bounds of 77.5% and 61.0%, respectively. Overall, our study identifies visual utilization as the dominant bottleneck in VLMs and suggests that a central opportunity for improvement lies in developing better mechanisms to locate, route, and use the visual information they already encode.
Abstract (translated)
URL
https://arxiv.org/abs/2610.29241
PDF
-
Think-Capture: Active 3D Geometry Acquisition for Spatial Reasoning in Vision-Language Models
2026-10-21 05:42:29Junjie Wang, Zhihao Yuan, Yunhan Tang, Sheng Xu, Jiaqi Mao, Xiaoyu Ye, Yueru Luo, Xu Yan, Chun-Mei Feng, Zhen LiAbstract
Vision-Language Models (VLMs) excel at 2D recognition but often fail to maintain consistent 3D relations across sparse viewpoints, leading to spatial hallucinations under occlusion and perspective changes. Existing methods commonly fuse all frames with dense geometry, incurring substantial computation and context overhead while encouraging blind reliance on imperfect estimators. We present Think-Capture, a query-conditioned framework that first identifies task-critical objects and then acquires their geometry on demand. A 3D toolkit lifts zero-shot masks into a common coordinate frame, and a Symbolic 3D State Abstraction compresses object and camera states into structured text. Hierarchical Group Relative Policy Optimization (GRPO) trains the VLM on cases in which geometry either helps or misleads, teaching it when to use tool evidence and when to fall back on visual cues. Think-Capture-4B reaches 54.4% on MindCube-Tiny (+21.5 points over its backbone), 39.1% on the MMSI-Bench positional subset (+7.9), and 42.8% on ViewSpatial (+5.2). It also remains robust under coordinate and mask perturbations. The full stack fits one 24 GB GPU and serves as an asynchronous high-level reasoner on an Agibot G1. Our code is available at this https URL.
Abstract (translated)
URL
https://arxiv.org/abs/2610.29232
PDF
-
DisasterRoute: Evidence-Guided Dual-System Routing for Visual Disaster Question Answering
2026-10-21 05:41:44Tan Zhang, Xing Deng, Lu Zhang, Xiaofeng Zhu, Ping HuAbstract
Visual question answering from low-altitude UAV imagery must deliver timely and accurate judgments to support disaster response under limited edge computing resources. Routine questions may be answered through direct decisions, whereas more challenging questions may benefit from step-by-step reasoning. However, existing research primarily focuses on improving models' visual understanding and generative reasoning capabilities, with limited exploration of efficient direct decision-making and its adaptive coordination with generative reasoning. We develop Disaster-S1 based on a vision-language model adapted to the disaster domain. Trained for option discrimination, it outputs a probability distribution over answer options in a single multimodal forward pass for fast decision-making. We further propose DisasterRoute, a dual-system routing framework that combines prediction confidence with disaster evidence to decide whether to accept the initial answer or invoke the disaster reasoning model DisasterVL for review. By allocating inference computation on demand, the framework retains the efficiency of direct decision-making and improves overall question-answering accuracy through selective review. Experiments on DisasterBench show that Disaster-S1 achieves superior accuracy with low latency, while DisasterRoute further improves accuracy through selective review at substantially lower latency than full reasoning.
Abstract (translated)
URL
https://arxiv.org/abs/2610.29231
PDF
-
When Multiple Choices Are Correct: Asymmetric Probability Allocation in Jev and Other Decision Models
Abstract
Decision models return normalized probability distributions over candidate choices, which downstream applications can use to select among choices. Although their providers emphasize the reliability of these probabilities, calibration on tasks with a single correct choice does not establish their reliability when multiple choices are correct. We evaluate three decision models (Jev, Liquid d1, and GPT-6 Luna Decisions) on three diverse datasets (MetaTool, HotpotQA, and ECtHR) to study probability allocation and thresholding. Even when models rank every correct choice above every distractor, in 28% to 50% of those lists, at least one correct choice receives a probability below the uniform probability of 1/n, where n is the number of choices. We also present each model with five identical copies of a correct choice and observe probability disparities, with mean gaps ranging from 17 to 79 percentage points. We further show how these allocation disparities affect the recovery of multiple correct choices under probability thresholding. Decision-model probabilities can rank correct choices, but a probability threshold does not identify the set of correct choices.
Abstract (translated)
URL
https://arxiv.org/abs/2610.29217
PDF
-
NavBind: Diagnosing Instruction-Action Binding at Route Forks in Navigation VLAs
2026-10-21 04:50:26Josh Qixuan Sun, Xiaoying Xing, Yurun Chen, Huaiyuan Weng, Wentao Zhu, Mark CrowleyAbstract
Vision-language-action (VLA) navigation models translate visual observations and instructions into actions. Navigation success assesses whether an agent reaches the goal. However, reaching the goal successfully does not establish whether the model genuinely follows the instructions or relies on visual cues or navigation history. We introduce NavBind to diagnose instruction-action binding: whether changing the instruction influences the probabilistic distribution of the actions, with the navigation context fixed. We construct 3,252 route-fork contexts by recombining route-aligned instruction segments from R2R-CE and RxR-CE. Each context pairs full instructions that share the description of the route already traveled but specify different branches. Both are evaluated under the same agent state and navigation history. We compare generated local action sequences, the likelihoods of fixed reference action sequences under both instructions, and changes in visual attention and attribution. Across the evaluated models, higher reported navigation success does not imply that the model more faithfully follows the instructions.
Abstract (translated)
URL
https://arxiv.org/abs/2610.29203
PDF
-
NeuralSRender: G-Buffer Guided Video Diffusion for Sim-to-Real Rendering
2026-10-21 04:45:19Yipeng Lu, Youqi Liao, Zhenyu Wang, Yuning Peng, Zhen DongAbstract
The scarcity of high-quality annotated visual data remains a critical bottleneck for autonomous-driving perception. Simulators provide training data at scale, but the limited realism of synthetic videos compromises perception performance in real-world applications. To address this limitation, we propose NeuralSRender, a sim-to-real video translation framework that enhances the photorealism of synthetic videos while maintaining geometric consistency with the source scenes. Built on a pretrained video diffusion transformer (DiT), NeuralSRender uses a lightweight projector to inject geometric and material priors from G-buffers, and LoRA enables parameter-efficient adaptation. The model is trained in two stages: it first learns G-buffer conditioning from synthetic videos with accurate G-buffers, then adapts to real-world appearance using real videos with estimated G-buffers. During the second stage, an additional LoRA module is trained while the projector and first-stage LoRA remain frozen to reduce interference from estimation noise. NeuralSRender introduces only 1.45\% parameter overhead and can be trained on consumer-grade GPUs with 24 GB of memory. To support training and evaluation, we introduce GenReal, a large-scale driving video dataset comprising synthetic and real-world videos with corresponding G-buffers. Experiments demonstrate improved photorealism with competitive geometric consistency. Training perception models on our translated videos instead of raw simulator renders improves detection $\mathrm{AP}_{50}$ on nuImages from 35.1 to 40.6 and segmentation mIoU on Cityscapes from 32.4 to 40.8. Our code, data, and model weights are available at: this https URL.
Abstract (translated)
URL
https://arxiv.org/abs/2610.29200
PDF
-
Cartender: Verifiable Language-Conditioned Driving by Learning the Drives Not Taken
2026-10-21 04:44:06Zhiwen Chen, Benjin Zhu, Yan Xie, Jun MaAbstract
Every driving scene is logged once, but driving is one-to-many: the scene determines which behaviors are feasible, while an instruction selects among them. A log records only the action taken, leaving no direct supervision for alternative feasible actions or instructions the scene cannot support. Consequently, a policy that ignores language and one that follows it without regard to scene constraints can fit the same data. We present Cartender, which learns from the drives not taken. Counterfactual Branching constructs matched and mismatched instruction-action pairs from real recordings without a simulator: a vision-language model proposes alternative futures and their selecting instructions, while unsupported instructions provide negative labels. A learned scene-language compatibility score regulates language conditioning in trajectory generation, and a reward-trained language residual re-ranks generated candidates over a scene-based score. The branches are used for training; instruction following is evaluated by swapping instructions under identical recorded inputs and measuring behavior rather than agreement with generated targets. On WOD-E2E, the best of four proposals achieves an RFS of 8.642 across 3,603 feasible instruction conditions. Across 8,974 admitted proposals, mean RFS is 8.203 versus 7.906 for matched logged trajectories, and 67% score higher than their logged counterpart. Compared with matched-only imitation using the same backbone, Cartender improves same-scene paired instruction success from 23.67% to 57.00% and RFS from 6.32 to 7.10; its gate reduces infeasible-instruction execution from 53.36% to 14.18%.
Abstract (translated)
URL
https://arxiv.org/abs/2610.26390
PDF
-
A Unified Evaluation Framework of Physical Adversarial Robustness in the Edge VLM Era
2026-10-21 04:38:02Eric Liang, Alan Luo, Kaiwen YuanAbstract
Recent advancements in Vision-Language Models (VLMs) enable end-to-end visual grounding via language prompts and generative decoding, demonstrating largely improved zero-shot and open-vocabulary detection capabilities. As a result, VLMs are rapidly being deployed at the edge in safety-critical real-world systems. However, the physical adversarial robustness of edge VLMs has not been systematically studied, and the fundamental differences in output with modern object detectors create methodological challenges for fair evaluation. To compare the two families directly, we evaluate every selected representative model on one outcome: whether an adversarial patch worn on a person evades detection of the model. We report this by introducing a unified metric, conditional attack success rate (cASR), counted only on images the model still localizes under a matched gray occluder, and partition every failure into omission (no box), substitution (a box on the patch), and displacement (a box on neither). We benchmark the physical adversarial vulnerability of compact grounding VLMs against a range of object detectors, and find that the VLMs can be as vulnerable as the detectors. We disclose that failure mode is largely associated with the output format of a model rather than its architecture family. Of the three, only omission is visible to a downstream system; displacement and substitution return confident, well-formed boxes, making them potentially more dangerous failures. Further, we find that, under distribution shift, VLMs that primarily failed by omission instead shift to failing more by displacement/substitution. Thus, a grounding VLM can inherit the detector's vulnerability while showing the possibility to adopt more dangerous failure modes. Our code and data are available at this https URL.
Abstract (translated)
URL
https://arxiv.org/abs/2610.29196
PDF
-
Resuming Vision-Language Inference across Onboard Interruptions
2026-10-21 04:14:43Xiaolong Huang, Yinyuan Zhang, Daliang Xu, Yun Ma, Qing Li, Mengwei Xu, Shangguang Wang, Gang Huang, Xuanzhe LiuAbstract
Executing vision-language models (VLMs) aboard low-Earth orbit satellites enables flexible processing of observation data without first downlinking it to the ground. On commercial off-the-shelf payloads, however, a request may span multiple execution windows: power and thermal constraints, fault-triggered reboots, and competing payload demands can interrupt inference before it completes. Each shutdown discards volatile activations and KV caches, forcing an unprotected request to repeat work after restart. We present OrbitWeaver, a framework that preserves VLM inference progress across onboard interruptions through coordinated checkpointing and recovery. Its phase-aware checkpointing selects recovery boundaries according to the state lifetimes and computational characteristics of vision encoding, language-model prefill, and decoding, balancing storage traffic against recomputation. Its dependency-aware dynamic recovery schedules restoration from partially completed states across the layer and token-chunk dimensions. By overlapping checkpoint and weight reads with recomputation, it reduces restart latency; the recovery planner computes an optimal execution order under its scheduling model within 100 ms. Experiments on satellite payload hardware using on-orbit telemetry traces show up to a 4.07-fold end-to-end speedup and a 3.56-fold reduction in SSD write volume over the evaluated baselines. A 72-hour deployment on two low-Earth orbit satellites demonstrates that OrbitWeaver retains inference progress and resumes execution across repeated bus-enforced shutdowns.
Abstract (translated)
URL
https://arxiv.org/abs/2610.29183
PDF
-
Correlated Failures of AI Approval Monitors
Abstract
Before an AI agent takes a consequential action, such as paying, messaging or changing code, several monitors check it. Their verdicts are combined by aggregating scores or by counting approvals, whatever inputs the monitors share. Monitors that share a base model or a poisoned input, however, may fail together. We borrow the fail-prone systems of distributed computing: lists of monitor sets that may fail together. A rule is safe if it executes no bad action and live if it blocks no good one. We measured twelve small open-weights monitors on payment approval under two pre-registered protocols. Without pre-registration, we also measured three closed frontier models (Claude Opus 5.5, GPT-6 Astra, Gemini 4 Argon) and the open DeepSeek-V3.2. Every reliable fooling of a monitor into approving a fraud came through an evidence source and reached only that source's readers. Fooling it into blocking a correct payment depended on injection resistance, a monitor's refusal to follow instructions planted in its data. Small models and DeepSeek obeyed such notes. The three closed models ignored them in all but one case, and Opus also resisted persuasion tailored to it. On new data, a rule that counts sources instead of votes blocked 0.9% of correct payments, against 7.4% for the best counting rule, at equal safety. The pre-registered worst-case test failed. One attack left the approvers on a single source, so no rule over the twelve language-model monitors survives every measured attack. Adding a rule engine, which cannot be persuaded, restored the condition for a safe and live rule on the measured attacks. Diversify the inputs, harden the monitors, count sources, not votes.
Abstract (translated)
URL
https://arxiv.org/abs/2610.29178
PDF
-
TrainBridge: Agentic NPU Training Optimization Guided by GPU Compilation Evidence
2026-10-21 04:00:48Yupeng Chen, Kunxiang Zhao, Xingjian Wang, Yue Zhao, Yijing Wu, Ye Liu, Kai Li, Congliang Chen, Ruoyu SunAbstract
We study how to improve training efficiency on Ascend NPUs. In the NPU training programs we study, framework-level compilation optimizations such as operator fusion are only partially applied, leaving fragmented computation and avoidable execution overheads. Profiling-guided tuning can address these inefficiencies, but it often requires costly trial and error. To make training on NPUs more efficient, we propose TrainBridge, an agentic approach for transferring GPU optimization guidance to NPU training. TrainBridge uses the highly optimized execution produced by mature GPU compilation stacks as a task-specific reference for NPU tuning. It extracts optimization intents from the GPU training execution to guide NPU program rewrites and further refines the rewritten program through iterative profiling. Our GPU-to-NPU migration experiments show that TrainBridge achieves an average 1.7x speedup across four workloads covering LLM pretraining, image classification, and vision-language training, and an average speedup of nearly 2.0x across 27 Modded-NanoGPT speedrun records with highly optimized GPU implementations.
Abstract (translated)
URL
https://arxiv.org/abs/2610.29171
PDF
-
SMR: Learning streaming motion retargeting for humanoids from human semantic
2026-10-21 03:59:54Chun Gu, Juntong Zhang, Jingyu Li, Li ZhangAbstract
Retargeting human motion to humanoids requires preserving motion semantics while accounting for the robot's physical constraints. Kinematic pipelines prescribe a robot trajectory through geometric matching, without establishing whether it can be executed under the robot's dynamics. Prior physics-supervised neural methods train bidirectional full-sequence models on physics-refined kinematic trajectories, remaining tied to prescribed kinematic solutions and requiring noncausal context. We propose SMR, a streaming neural motion retargeter learned from semantic-conditioned physics experts. We convert human motion into robot-aware task-space targets for body landmark positions, orientations, and velocities. Physics experts track these targets directly in simulation under contact, dynamics, and actuator constraints. We train a single strictly causal Transformer on their executed state trajectories that complete the motion and pass semantic and physical validity checks. At each incoming frame, the student uses only current and past human observations to generate a robot motion reference for a downstream whole-body controller. Experiments demonstrate that semantic-conditioned physics experts generate motion with higher semantic fidelity and physical validity than direct optimization-based retargeting, while the resulting streaming retargeter achieves improved closed-loop performance over kinematic and physics-refined baselines under a common whole-body controller.
Abstract (translated)
URL
https://arxiv.org/abs/2610.29169
PDF
-
SimPT: An Automated Simulation Framework for Intervention-Based VLA Post-Training
2026-10-21 03:53:00Caicheng Wang, Zipeng Ye, Zhexuan Zhou, Zilong Huang, Xi Zhang, Yuchen XieAbstract
Human-in-the-loop (HIL) post-training improves vision-language-action (VLA) policies through expert corrections, but costly human supervision and repeated real-robot evaluation limit algorithm exploration and development. We introduce SimPT (Simulation-based Post-Training), a simulation framework that automates intervention-based VLA post-training by integrating policy rollouts, corrective data collection, policy training, and standalone evaluation. To automatically determine intervention timing and recovery strategies, we develop an Agent-Compiled Progress Graph that tracks task progress using privileged simulator state and retains stable snapshots during policy execution. After a rollout fails, a CuRobo-based expert attempts recovery from these snapshots, starting at more advanced task stages and falling back to earlier ones as needed to generate corrective trajectories. We implement and compare four post-training methods, RECAP, HG-DAgger, DPO+SFT, and FlowPRO, across seven RoboTwin manipulation tasks. Relative to a fixed-interval snapshot baseline, the Progress Graph reduces recovery attempts by 75.0% and increases recovery success from 93.4% to 98.9%. Using SimPT, we develop RECAP-CF-LPF, a refinement of RECAP that combines cross-fitting and low-pass filtering of value predictions. This refinement increases the overall success rate from 76.50% to 79.36% in held-out simulation trials. In real-robot HIL post-training, it increases the success rate from 72% to 96% across 25 scenarios of an Ethernet cable-insertion task. These results demonstrate the utility of SimPT for exploring and developing VLA post-training algorithms.
Abstract (translated)
URL
https://arxiv.org/abs/2610.29161
PDF
-
RiO-DETRv2: Geometry-Aware Uncertainty Modeling Meets Real-Time Oriented Object Detection Transformer
2026-10-21 03:48:13Zhangchi Hu, Zilong He, Yifan Zhao, Xiangyu Yu, Jie Chen, Peixi Wu, Yansong Peng, Hebei Li, Xiaoyan SunAbstract
RiO-DETR is the first real-time end-to-end transformer for oriented object detection, yet further localization gains require a more expressive box representation. Although distributional regression has succeeded in horizontal DETRs, extending it to oriented detection introduces geometry-misaligned boundary modeling and topology-inconsistent orientation refinement. We present RiO-DETRv2, a geometry-aware uncertainty modeling framework for real-time oriented object detection. To resolve the geometry mismatch, Object-Aligned Boundary Distribution Refinement represents box boundaries as probability distributions in an object-aligned coordinate system and progressively refines them across decoder layers. For better orientation topology consistency, we introduce Circular-to-Local Angular Residual Refinement that transitions from global periodic search to local fine-grained estimation. Unified OBB Quality-Aligned Learning further aligns classification, boundary and angular refinement with complete-OBB localization quality. Together, these designs improve localization without compromising efficiency, achieving a better balance between speed and accuracy. Specifically, RiO-DETRv2-l/x achieves 53.7/54.1 mAP50-95 on DOTA, outperforming RiO-DETR-l/x by 1.2/1.5 points with nearly unchanged computational cost. Code is available at this https URL.
Abstract (translated)
URL
https://arxiv.org/abs/2610.29159
PDF
-
DGCM-SSFA: Dual-Granularity Cross-Modulation and Sparse Strip Frequency Attention for Efficient 3T-to-7T MRI Synthesis
2026-10-21 03:21:45Zhichun Fan, Zhiqiang Chen, Lanling Liu, Lingjun Liu, Zhonghua XieAbstract
Synthesizing high-fidelity 7T MRI from routinely acquired 3T scans offers a practical pathway to enhanced neuroimaging without the prohibitive cost of ultra-high-field scanners. While existing deep learning approaches have achieved promising results, state-of-the-art methods typically rely on either large-scale architectures exceeding 45 million parameters or computationally intensive ensemble strategies, limiting their practicality in resource-constrained clinical environments. In this work, we propose a compact multi-domain attention framework, termed DGCM-SSFA, that achieves competitive synthesis quality with only 11M parameters. The proposed architecture integrates two complementary modules into a U-Net Transformer backbone. First, a Dual-Granularity Cross-Modulated Feature Modulator (DGCM) is embedded at multiple hierarchical levels to enhance local edge and texture details through Sobel-guided spatial attention and depthwise frequency-aware channel attention with cross-branch gating. Second, a Sparse Strip Frequency Attention (SSFA) module at the bottleneck captures global spatial context via strip-pooled attention with near-linear complexity, while simultaneously performing FFT-based spectral modulation in the frequency domain. These two modules operate synergistically with the backbone's native channel attention, forming a triple-granularity feature refinement pipeline. Extensive experiments on the UNC and BNU 3T-to-7T MRI datasets across T1w and T2w modalities demonstrate that DGCM-SSFA achieves strong structural fidelity, attaining an SSIM of 0.8431 on BNU T2w---approaching the best benchmark value of 0.8624---and 0.8100 on BNU T1w, while using substantially fewer parameters than Transformer-based alternatives such as ResViT (123.4M) and FS-RWKV (45.4M).
Abstract (translated)
URL
https://arxiv.org/abs/2610.29141
PDF
-
GRAFT: Restoring the Missing Sampling Interface in Continuous Diffusion Language Models
2026-10-21 03:15:30Haoning Yang, Lili PanAbstract
Continuous diffusion language models generate a whole sequence in parallel, never drawing a single word at any intermediate step; only the final step turns embeddings into text. Models such as ELF now report autoregressive-level quality at a fraction of the sampling steps of discrete-token diffusion models, and this design principle -- no step-wise discretization -- is what their own ablations credit for the gain. We show that this same principle carries a hidden cost: a collapse of the lexical long tail. Measured against held-out human text, ELF's output matches the human rate on common words almost exactly, yet its rarest words appear at only 0.30x the human rate, a gap that does not close from 105M to 652M parameters and that no sampler setting these models expose can reach. A same-size GPT-2, in contrast, spans zero to above the human rate by its decoding rule alone. We trace the difference to what continuous diffusion models lack: with no step that ever draws a discrete word, every step is instead driven by an average over the candidates still in play, and averaging a rare candidate together with the common word competing for its slot lands on the common word. An autoregressive model avoids this because its decoding rule samples one candidate rather than averaging over all of them. We propose GRAFT, a training-free method that grafts this missing choice back onto a frozen model: while the model is still undecided, GRAFT commits the rare words the average would otherwise discard, then regenerates the sequence from scratch with those commitments held in place. Across three ELF scales and on LangFlow, an independent continuous dLLM, GRAFT restores the rare-word share of generated text from 0.42-0.77x the human rate to 0.82-1.10x, a relative gain of 34-109%, and moves the entire word-frequency profile closer to human on every model tested, by 2.7x on ELF-B and 12x on LangFlow.
Abstract (translated)
URL
https://arxiv.org/abs/2610.29136
PDF
-
Recursive Self-Improvement on a Budget: Scoring Code Edits Before They Run
2026-10-21 03:04:25Erfan Darzi, Sree Bhargavi Balija, Sheng HeAbstract
A coding agent that rewrites its own scaffold has to choose which proposed edit to expand before that edit has been evaluated, and the search then has to return one agent. Each run fits one linear readout on the parent's steering-split solve rate and eight coordinates of the proposed diff, from 96 scaffold edits labeled by four evaluations of a 16-instance fit split, and then freezes the coefficients. HumanEval+ reserves 48 instances as the steering split and Polyglot reserves 64. That split scores parents and keeps an archive of four nodes. Huxley-Gödel Machine selection draws a parent by Thompson sampling on the steering counts of nodes that have already been evaluated, and it expands the child that parent proposes. Every method then returns the earliest evaluated node whose steering-split solve count is maximal. The tables report the held-out score of that node on the other 100 HumanEval+ instances and the other 145 Polyglot instances, next to the root scaffold and the calibration edit with the highest fit-split label. Twelve runs of GPT-4o return a node at 87.0% on held-out HumanEval+, against 84.2% for the parent-rate ablation. On that split Huxley-Gödel Machine selection returns 85.1%, the root is at 81.2%, and the calibration-best edit is at 82.8%. On the same split Claude Sonnet 4.5 returns 86.6%, beside 86.5% for Huxley-Gödel Machine selection, 85.3% for the parent-rate ablation, and 85.1% for the calibration-best edit.
Abstract (translated)
URL
https://arxiv.org/abs/2610.29121
PDF
