# Research Papers — 2025-10 · bundle page 171 of 217 > Full-text markdown of papers #8501–8550 of 10843 published in 2025-10, newest first. AI-readable mirrors of the full 10843-paper 2025-10 archive (50 papers per file). Canonical page: https://aipod.dev/papers/archive/2025-10/page/171 [← Previous bundle](https://aipod.dev/papers/archive/2025-10/page/170.md) [Next bundle →](https://aipod.dev/papers/archive/2025-10/page/172.md) ### BlockGPT: Spatio-Temporal Modelling of Rainfall via Frame-Level Autoregression Authors: Cristian Meo, Varun Sarathchandran, Avijit Majhi, Shao Hung, Carlo Saccardi, Ruben Imhoff, Roberto Deidda, Remko Uijlenhoet, Justin Dauwels Published: 2025-10-07 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2510.06293v2 Predicting precipitation maps is a highly complex spatiotemporal modeling task, critical for mitigating the impacts of extreme weather events. Short-term precipitation forecasting, or nowcasting, requires models that are not only accurate but also computationally efficient for real-time applications. Current methods, such as token-based autoregressive models, often suffer from flawed inductive biases and slow inference, while diffusion models can be computationally intensive. To address these limitations, we introduce BlockGPT, a generative autoregressive transformer using batched tokenization (Block) method that predicts full two-dimensional fields (frames) at each time step. Conceived as a model-agnostic paradigm for video prediction, BlockGPT factorizes space-time by using self-attention within each frame and causal attention across frames; in this work, we instantiate it for precipitation nowcasting. We evaluate BlockGPT on two precipitation datasets, viz. KNMI (Netherlands) and SEVIR (U.S.), comparing it to state-of-the-art baselines including token-based (NowcastingGPT) and diffusion-based (DiffCast+Phydnet) models. The results show that BlockGPT achieves superior accuracy, event localization as measured by categorical metrics, and inference speeds up to 31x faster than comparable baselines. --- ### ChainMPQ: Interleaved Text-Image Reasoning Chains for Mitigating Relation Hallucinations Authors: Yike Wu, Yiwei Wang, Yujun Cai Published: 2025-10-07 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2510.06292v2 While Large Vision-Language Models (LVLMs) achieve strong performance in multimodal tasks, hallucinations continue to hinder their reliability. Among the three categories of hallucinations, which include object, attribute, and relation, relation hallucinations account for the largest proportion but have received the least attention. To address this issue, we propose ChainMPQ (Multi-Perspective Questions guided Interleaved Chain of Image and Text), a training-free method that improves relational inference in LVLMs by utilizing accumulated textual and visual memories. ChainMPQ first extracts subject and object keywords from the question to enhance the corresponding image regions. It then constructs multi-perspective questions that focus on the three core components of a relationship: the subject, the object, and the relation that links them. These questions are sequentially input to the model, with textual and visual memories from earlier steps providing supporting context for subsequent ones, thereby forming an interleaved chain of images and text that guides progressive relational reasoning. Experiments on multiple LVLMs and benchmarks show that ChainMPQ substantially reduces relation hallucinations, while ablation studies further validate the effectiveness of its three core modules. --- ### Traj-Transformer: Diffusion Models with Transformer for GPS Trajectory Generation Authors: Zhiyang Zhang, Ningcong Chen, Xin Zhang, Yanhua Li, Shen Su, Hui Lu, Jun Luo Published: 2025-10-07 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2510.06291v1 The widespread use of GPS devices has driven advances in spatiotemporal data mining, enabling machine learning models to simulate human decision making and generate realistic trajectories, addressing both data collection costs and privacy concerns. Recent studies have shown the promise of diffusion models for high-quality trajectory generation. However, most existing methods rely on convolution based architectures (e.g. UNet) to predict noise during the diffusion process, which often results in notable deviations and the loss of fine-grained street-level details due to limited model capacity. In this paper, we propose Trajectory Transformer, a novel model that employs a transformer backbone for both conditional information embedding and noise prediction. We explore two GPS coordinate embedding strategies, location embedding and longitude-latitude embedding, and analyze model performance at different scales. Experiments on two real-world datasets demonstrate that Trajectory Transformer significantly enhances generation quality and effectively alleviates the deviation issues observed in prior approaches. --- ### Soft-Evidence Fused Graph Neural Network for Cancer Driver Gene Identification across Multi-View Biological Graphs Authors: Bang Chen, Lijun Guo, Houli Fan, Wentao He, Rong Zhang Published: 2025-10-07 Categories: q-bio.GN, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2510.06290v1 Identifying cancer driver genes (CDGs) is essential for understanding cancer mechanisms and developing targeted therapies. Graph neural networks (GNNs) have recently been employed to identify CDGs by capturing patterns in biological interaction networks. However, most GNN-based approaches rely on a single protein-protein interaction (PPI) network, ignoring complementary information from other biological networks. Some studies integrate multiple networks by aligning features with consistency constraints to learn unified gene representations for CDG identification. However, such representation-level fusion often assumes congruent gene relationships across networks, which may overlook network heterogeneity and introduce conflicting information. To address this, we propose Soft-Evidence Fusion Graph Neural Network (SEFGNN), a novel framework for CDG identification across multiple networks at the decision level. Instead of enforcing feature-level consistency, SEFGNN treats each biological network as an independent evidence source and performs uncertainty-aware fusion at the decision level using Dempster-Shafer Theory (DST). To alleviate the risk of overconfidence from DST, we further introduce a Soft Evidence Smoothing (SES) module that improves ranking stability while preserving discriminative performance. Experiments on three cancer datasets show that SEFGNN consistently outperforms state-of-the-art baselines and exhibits strong potential in discovering novel CDGs. --- ### BuilderBench: The Building Blocks of Intelligent Agents Authors: Raj Ghugare, Roger Creus Castanyer, Catherine Ji, Kathryn Wantlin, Jin Schofield, Karthik Narasimhan, Benjamin Eysenbach Published: 2025-10-07 Categories: cs.AI, cs.LG arXiv: https://arxiv.org/abs/2510.06288v4 Today's AI models learn primarily through mimicry and refining, so it is not surprising that they struggle to solve problems beyond the limits set by existing data. To solve novel problems, agents should acquire skills by exploring and learning through experience. Finding a scalable learning mechanism for developing agents that learn through interaction remains a major open problem. In this work, we introduce BuilderBench, a benchmark to accelerate research into agent training that centers open-ended exploration. BuilderBench requires agents to learn how to build any structure using blocks. BuilderBench is equipped with (1) a simulator of a robot interacting with various physical blocks, and (2) a task-suite with over 50 diverse target structures that are carefully curated to test an understanding of physics, mathematics, and long-horizon planning. Agents are provided with a target structure at the start, and can interact with the environment for multiple episodes to experiment and learn various skills for building the structure. Solving these tasks requires \emph{embodied reasoning} in a way that is not reflected in words but rather in actions, experimenting with different strategies and piecing them together. Our experiments with multiple state-of-the-art frontier language model based agents and tabula rasa reinforcement learning algorithms show that these agents cannot solve any of the non-trivial tasks in the BuilderBench. Our analysis throws light on the lack of exploration abilities in these models. --- ### Mass Conservation on Rails -- Rethinking Physics-Informed Learning of Ice Flow Vector Fields Authors: Kim Bente, Roman Marchant, Fabio Ramos Published: 2025-10-07 Categories: physics.ao-ph, cs.LG, physics.comp-ph, physics.flu-dyn, physics.geo-ph, stat.ML arXiv: https://arxiv.org/abs/2510.06286v1 To reliably project future sea level rise, ice sheet models require inputs that respect physics. Embedding physical principles like mass conservation into models that interpolate Antarctic ice flow vector fields from sparse & noisy measurements not only promotes physical adherence but can also improve accuracy and robustness. While physics-informed neural networks (PINNs) impose physics as soft penalties, offering flexibility but no physical guarantees, we instead propose divergence-free neural networks (dfNNs), which enforce local mass conservation exactly via a vector calculus trick. Our comparison of dfNNs, PINNs, and unconstrained NNs on ice flux interpolation over Byrd Glacier suggests that "mass conservation on rails" yields more reliable estimates, and that directional guidance, a learning strategy leveraging continent-wide satellite velocity data, boosts performance across models. --- ### Human3R: Everyone Everywhere All at Once Authors: Yue Chen, Xingyu Chen, Yuxuan Xue, Anpei Chen, Yuliang Xiu, Gerard Pons-Moll Published: 2025-10-07 Categories: cs.CV arXiv: https://arxiv.org/abs/2510.06219v2 We present Human3R, a unified, feed-forward framework for online 4D human-scene reconstruction, in the world frame, from casually captured monocular videos. Unlike previous approaches that rely on multi-stage pipelines, iterative contact-aware refinement between humans and scenes, and heavy dependencies, e.g., human detection, depth estimation, and SLAM pre-processing, Human3R jointly recovers global multi-person SMPL-X bodies ("everyone"), dense 3D scene ("everywhere"), and camera trajectories in a single forward pass ("all-at-once"). Our method builds upon the 4D online reconstruction model CUT3R, and uses parameter-efficient visual prompt tuning, to strive to preserve CUT3R's rich spatiotemporal priors, while enabling direct readout of multiple SMPL-X bodies. Human3R is a unified model that eliminates heavy dependencies and iterative refinement. After being trained on the relatively small-scale synthetic dataset BEDLAM for just one day on one GPU, it achieves superior performance with remarkable efficiency: it reconstructs multiple humans in a one-shot manner, along with 3D scenes, in one stage, in real-time (15 FPS) with a low memory footprint (8 GB). Extensive experiments demonstrate that Human3R delivers state-of-the-art or competitive performance across tasks, including global human motion estimation, local human mesh recovery, video depth estimation, and camera pose estimation, with a single unified model. We hope that Human3R will serve as a simple yet strong baseline, which can be easily adapted for downstream applications. Code, models and 4D interactive demos are available at https://fanegg.github.io/Human3R/. --- ### EgoNight: Towards Egocentric Vision Understanding at Night with a Challenging Benchmark Authors: Deheng Zhang, Yuqian Fu, Runyi Yang, Yang Miao, Tianwen Qian, Xu Zheng, Guolei Sun, Ajad Chhatkuli, Xuanjing Huang, Yu-Gang Jiang, Luc Van Gool, Danda Pani Paudel Published: 2025-10-07 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2510.06218v2 Most existing benchmarks for understanding egocentric vision focus primarily on daytime scenarios, overlooking the low-light conditions that are inevitable in real-world applications. To investigate this gap, we present EgoNight, the first comprehensive benchmark for nighttime egocentric vision, with visual question answering (VQA) as the core task. A key feature of EgoNight is the introduction of day-night aligned videos, which enhance night annotation quality using the daytime data and reveal clear performance gaps between lighting conditions. To achieve this, we collect both synthetic videos rendered by Blender and real-world recordings, ensuring that scenes and actions are visually and temporally aligned. Leveraging these paired videos, we construct EgoNight-VQA, supported by a novel day-augmented night auto-labeling engine and refinement through extensive human verification. Each QA pair is double-checked by annotators for reliability. In total, EgoNight-VQA contains 3658 QA pairs across 90 videos, spanning 12 diverse QA types, with more than 300 hours of human work. Evaluations of state-of-the-art multimodal large language models (MLLMs) reveal substantial performance drops when transferring from day to night, underscoring the challenges of reasoning under low-light conditions. Beyond VQA, EgoNight also introduces two auxiliary tasks, day-night correspondence retrieval and egocentric depth estimation at night, that further explore the boundaries of existing models. We believe EgoNight-VQA provides a strong foundation for advancing application-driven egocentric vision research and for developing models that generalize across illumination domains. The code and data can be found at https://github.com/dehezhang2/EgoNight. --- ### TaTToo: Tool-Grounded Thinking PRM for Test-Time Scaling in Tabular Reasoning Authors: Jiaru Zou, Soumya Roy, Vinay Kumar Verma, Ziyi Wang, David Wipf, Pan Lu, Sumit Negi, James Zou, Jingrui He Published: 2025-10-07 Categories: cs.AI, cs.CL, cs.LG arXiv: https://arxiv.org/abs/2510.06217v1 Process Reward Models (PRMs) have recently emerged as a powerful framework for enhancing the reasoning capabilities of large reasoning models (LRMs), particularly in the context of test-time scaling (TTS). However, their potential for supervising LRMs on tabular reasoning domains remains underexplored. Through detailed empirical analyses, we identify that existing PRMs, though widely adopted for supervising text-only reasoning steps, struggle with table-specific operations such as sub-table retrieval and schema interaction, leading to critical performance bottlenecks. To address this limitation, we propose TaTToo, a novel table-grounded PRM framework that (i) reasons explicitly over tabular reasoning steps and (ii) integrates tool-based verification to provide precise reward supervision. Concretely, we first design a scalable data curation pipeline that constructs over 60k high-quality step-level annotations by integrating table verification rationales with tool-based executions. Building on the collected data, we train TaTToo with a dual-stage paradigm: cold-start supervised fine-tuning to capture tool-use reasoning patterns, followed by reinforcement learning with tool-grounded reward shaping to align our model with table-based verification. We provide a comprehensive evaluation of the policy improvement induced by our newly designed PRM. Across 5 challenging tabular reasoning benchmarks covering numerical reasoning, fact-checking, and data analysis, TaTToo improves downstream policy LRMs by 30.9% at inference, surpasses strong PRM baselines such as Qwen-2.5-Math-PRM-72B with only 8B parameters, and demonstrates strong generalizability across diverse TTS strategies. --- ### Dropping the D: RGB-D SLAM Without the Depth Sensor Authors: Mert Kiray, Alican Karaomer, Benjamin Busam Published: 2025-10-07 Categories: cs.CV, cs.RO arXiv: https://arxiv.org/abs/2510.06216v2 We present DropD-SLAM, a real-time monocular SLAM system that achieves RGB-D-level accuracy without relying on depth sensors. The system replaces active depth input with three pretrained vision modules: a monocular metric depth estimator, a learned keypoint detector, and an instance segmentation network. Dynamic objects are suppressed using dilated instance masks, while static keypoints are assigned predicted depth values and backprojected into 3D to form metrically scaled features. These are processed by an unmodified RGB-D SLAM back end for tracking and mapping. On the TUM RGB-D benchmark, DropD-SLAM attains 7.4 cm mean ATE on static sequences and 1.8 cm on dynamic sequences, matching or surpassing state-of-the-art RGB-D methods while operating at 22 FPS on a single GPU. These results suggest that modern pretrained vision models can replace active depth sensors as reliable, real-time sources of metric scale, marking a step toward simpler and more cost-effective SLAM systems. --- ### Fine-grained Defocus Blur Control for Generative Image Models Authors: Ayush Shrivastava, Connelly Barnes, Xuaner Zhang, Lingzhi Zhang, Andrew Owens, Sohrab Amirghodsi, Eli Shechtman Published: 2025-10-07 Categories: cs.CV arXiv: https://arxiv.org/abs/2510.06215v2 Current text-to-image diffusion models excel at generating diverse, high-quality images, yet they struggle to incorporate fine-grained camera metadata such as precise aperture settings. In this work, we introduce a novel text-to-image diffusion framework that leverages camera metadata, or EXIF data, which is often embedded in image files, with an emphasis on generating controllable lens blur. Our method mimics the physical image formation process by first generating an all-in-focus image, estimating its monocular depth, predicting a plausible focus distance with a novel focus distance transformer, and then forming a defocused image with an existing differentiable lens blur model. Gradients flow backwards through this whole process, allowing us to learn without explicit supervision to generate defocus effects based on content elements and the provided EXIF data. At inference time, this enables precise interactive user control over defocus effects while preserving scene contents, which is not achievable with existing diffusion models. Experimental results demonstrate that our model enables superior fine-grained control without altering the depicted scene. --- ### Stratified GRPO: Handling Structural Heterogeneity in Reinforcement Learning of LLM Search Agents Authors: Mingkang Zhu, Xi Chen, Bei Yu, Hengshuang Zhao, Jiaya Jia Published: 2025-10-07 Categories: cs.LG, cs.AI, cs.CL arXiv: https://arxiv.org/abs/2510.06214v1 Large language model (LLM) agents increasingly rely on external tools such as search engines to solve complex, multi-step problems, and reinforcement learning (RL) has become a key paradigm for training them. However, the trajectories of search agents are structurally heterogeneous, where variations in the number, placement, and outcomes of search calls lead to fundamentally different answer directions and reward distributions. Standard policy gradient methods, which use a single global baseline, suffer from what we identify and formalize as cross-stratum bias-an "apples-to-oranges" comparison of heterogeneous trajectories. This cross-stratum bias distorts credit assignment and hinders exploration of complex, multi-step search strategies. To address this, we propose Stratified GRPO, whose central component, Stratified Advantage Normalization (SAN), partitions trajectories into homogeneous strata based on their structural properties and computes advantages locally within each stratum. This ensures that trajectories are evaluated only against their true peers. Our analysis proves that SAN eliminates cross-stratum bias, yields conditionally unbiased unit-variance estimates inside each stratum, and retains the global unbiasedness and unit-variance properties enjoyed by standard normalization, resulting in a more pure and scale-stable learning signal. To improve practical stability under finite-sample regimes, we further linearly blend SAN with the global estimator. Extensive experiments on diverse single-hop and multi-hop question-answering benchmarks demonstrate that Stratified GRPO consistently and substantially outperforms GRPO by up to 11.3 points, achieving higher training rewards, greater training stability, and more effective search policies. These results establish stratification as a principled remedy for structural heterogeneity in RL for LLM search agents. --- ### Training Dynamics Impact Post-Training Quantization Robustness Authors: Albert Catalan-Tatjer, Niccolò Ajroldi, Jonas Geiping Published: 2025-10-07 Categories: cs.LG arXiv: https://arxiv.org/abs/2510.06213v2 While post-training quantization is widely adopted for efficient deployment of large language models, the mechanisms underlying quantization robustness remain unclear. We conduct a comprehensive analysis of quantization degradation across open-source language model training trajectories up to 32B parameters and 15T training tokens to accurately assess the relationship between training dynamics and quantization performance. Our key finding is that quantization errors in large-scale training runs are driven by a complex interplay between learning rate and other training hyperparameters. Specifically, once learning rates decay, validation loss and quantization error diverge, largely independent of training data scale. To investigate interventions on the training dynamics and identify specific configurations that can modulate quantization robustness favorably, we train our own models in controlled experiments up to 100B tokens. Our results challenge the assumption that increasing dataset scale inherently compromises quantization effectiveness, demonstrating instead that strategic training hyperparameter interventions can improve quantization quality at scale. --- ### Drive&Gen: Co-Evaluating End-to-End Driving and Video Generation Models Authors: Jiahao Wang, Zhenpei Yang, Yijing Bai, Yingwei Li, Yuliang Zou, Bo Sun, Abhijit Kundu, Jose Lezama, Luna Yue Huang, Zehao Zhu, Jyh-Jing Hwang, Dragomir Anguelov et al. Published: 2025-10-07 Categories: cs.CV arXiv: https://arxiv.org/abs/2510.06209v1 Recent advances in generative models have sparked exciting new possibilities in the field of autonomous vehicles. Specifically, video generation models are now being explored as controllable virtual testing environments. Simultaneously, end-to-end (E2E) driving models have emerged as a streamlined alternative to conventional modular autonomous driving systems, gaining popularity for their simplicity and scalability. However, the application of these techniques to simulation and planning raises important questions. First, while video generation models can generate increasingly realistic videos, can these videos faithfully adhere to the specified conditions and be realistic enough for E2E autonomous planner evaluation? Second, given that data is crucial for understanding and controlling E2E planners, how can we gain deeper insights into their biases and improve their ability to generalize to out-of-distribution scenarios? In this work, we bridge the gap between the driving models and generative world models (Drive&Gen) to address these questions. We propose novel statistical measures leveraging E2E drivers to evaluate the realism of generated videos. By exploiting the controllability of the video generation model, we conduct targeted experiments to investigate distribution gaps affecting E2E planner performance. Finally, we show that synthetic data produced by the video generation model offers a cost-effective alternative to real-world data collection. This synthetic data effectively improves E2E model generalization beyond existing Operational Design Domains, facilitating the expansion of autonomous vehicle services into new operational contexts. --- ### ShapeGen4D: Towards High Quality 4D Shape Generation from Videos Authors: Jiraphon Yenphraphai, Ashkan Mirzaei, Jianqi Chen, Jiaxu Zou, Sergey Tulyakov, Raymond A. Yeh, Peter Wonka, Chaoyang Wang Published: 2025-10-07 Categories: cs.CV arXiv: https://arxiv.org/abs/2510.06208v1 Video-conditioned 4D shape generation aims to recover time-varying 3D geometry and view-consistent appearance directly from an input video. In this work, we introduce a native video-to-4D shape generation framework that synthesizes a single dynamic 3D representation end-to-end from the video. Our framework introduces three key components based on large-scale pre-trained 3D models: (i) a temporal attention that conditions generation on all frames while producing a time-indexed dynamic representation; (ii) a time-aware point sampling and 4D latent anchoring that promote temporally consistent geometry and texture; and (iii) noise sharing across frames to enhance temporal stability. Our method accurately captures non-rigid motion, volume changes, and even topological transitions without per-frame optimization. Across diverse in-the-wild videos, our method improves robustness and perceptual fidelity and reduces failure modes compared with the baselines. --- ### Modulation Discovery with Differentiable Digital Signal Processing Authors: Christopher Mitcheltree, Hao Hao Tan, Joshua D. Reiss Published: 2025-10-07 Categories: cs.SD, cs.LG, eess.AS arXiv: https://arxiv.org/abs/2510.06204v1 Modulations are a critical part of sound design and music production, enabling the creation of complex and evolving audio. Modern synthesizers provide envelopes, low frequency oscillators (LFOs), and more parameter automation tools that allow users to modulate the output with ease. However, determining the modulation signals used to create a sound is difficult, and existing sound-matching / parameter estimation systems are often uninterpretable black boxes or predict high-dimensional framewise parameter values without considering the shape, structure, and routing of the underlying modulation curves. We propose a neural sound-matching approach that leverages modulation extraction, constrained control signal parameterizations, and differentiable digital signal processing (DDSP) to discover the modulations present in a sound. We demonstrate the effectiveness of our approach on highly modulated synthetic and real audio samples, its applicability to different DDSP synth architectures, and investigate the trade-off it incurs between interpretability and sound-matching accuracy. We make our code and audio samples available and provide the trained DDSP synths in a VST plugin. --- ### Reference Grounded Skill Discovery Authors: Seungeun Rho, Aaron Trinh, Danfei Xu, Sehoon Ha Published: 2025-10-07 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2510.06203v3 Scaling unsupervised skill discovery algorithms to high-DoF agents remains challenging. As dimensionality increases, the exploration space grows exponentially, while the manifold of meaningful skills remains limited. Therefore, semantic meaningfulness becomes essential to effectively guide exploration in high-dimensional spaces. In this work, we present Reference-Grounded Skill Discovery (RGSD), a novel algorithm that grounds skill discovery in a semantically meaningful latent space using reference data. RGSD first performs contrastive pretraining to embed motions on a unit hypersphere, clustering each reference trajectory into a distinct direction. This grounding enables skill discovery to simultaneously involve both imitation of reference behaviors and the discovery of semantically related diverse behaviors. On a simulated SMPL humanoid with $359$-D observations and $69$-D actions, RGSD successfully imitates skills such as walking, running, punching, and sidestepping, while also discover variations of these behaviors. In downstream locomotion tasks, RGSD leverages the discovered skills to faithfully satisfy user-specified style commands and outperforms imitation-learning baselines, which often fail to maintain the commanded style. --- ### TokenChain: A Discrete Speech Chain via Semantic Token Modeling Authors: Mingxuan Wang, Satoshi Nakamura Published: 2025-10-07 Categories: eess.AS, cs.AI, cs.CL, cs.LG, cs.SD arXiv: https://arxiv.org/abs/2510.06201v2 Machine Speech Chain, simulating the human perception-production loop, proves effective in jointly improving ASR and TTS. We propose TokenChain, a fully discrete speech chain coupling semantic-token ASR with a two-stage TTS: an autoregressive text-to-semantic model co-trained with ASR and a masked-generative semantic-to-acoustic model for synthesis only. End-to-end feedback across the text interface is enabled with straight-through argmax/Gumbel-Softmax and balanced with supervised ASR via dynamic weight averaging. Ablations examine optimal temperature schedules for in- and cross-domain transfer. Evaluation reveals TokenChain surpasses baseline accuracy 2-6 epochs earlier and yields 5-13% lower equal-epoch error with stable T2S on LibriSpeech, and reduces relative ASR WER by 56% and T2S WER by 31% on TED-LIUM with minimal forgetting, showing that chain learning remains effective with token interfaces and models. --- ### StarEmbed: Benchmarking Time Series Foundation Models on Astronomical Observations of Variable Stars Authors: Weijian Li, Hong-Yu Chen, Nabeel Rehemtulla, Ved G. Shah, Dongho Kim, Dennis Wu, Qinjie Lin, Adam A. Miller, Han Liu Published: 2025-10-07 Categories: astro-ph.SR, astro-ph.IM, cs.AI arXiv: https://arxiv.org/abs/2510.06200v4 Current time series foundation model (TSFM) training corpora largely omit data with certain complexities like irregular temporal sampling. Astronomical time series of stellar fluxes (light curves) are available in immense quantities and exhibit irregular sampling, multiple variates, and heteroskedasticity. We introduce StarEmbed, the first public benchmark for light curves comprised of real observations of ~40,000 stars expert-labeled across seven classes and evaluations in clustering, classification, and out-of-distribution (OOD) source detection. We benchmark TSFMs with differing architecture and training strategies as well as domain-specific transformers. Our results demonstrate that the Chronos family, despite being pre-trained on regularly sampled non-astronomical data, yields state-of-the-art (SOTA) performance in light curve clustering and OOD detection. While no TSFM strictly surpasses the classification performance of the long-established domain baseline, they do demonstrate excellent generalization abilities. StarEmbed marks a step toward universal light curve embeddings and improved TSFM performance on challenging data. --- ### The Answer Lies Within: Self-Derived Rewards Enable Explainable Relation Extraction Authors: Xinyu Guo, Zhengliang Shi, Minglai Yang, Mihai Surdeanu Published: 2025-10-07 Categories: cs.CL, cs.IR arXiv: https://arxiv.org/abs/2510.06198v3 Despite the remarkable reasoning capabilities of large language models, they still struggle with one-shot relation extraction without predefined relation labels. We identify two pitfalls: models are often misled by irrelevant tokens instead of relation-conveying semantics, and they often fail to align with the abstraction level human annotators expect. We introduce a novel framework that closes this gap with two components: (1) COGRE, a cognitively-inspired reasoning framework that structures RE into a series of processes mimicking human text-processing; and (2) HIT@DICT, a reinforcement learning intermediate reward strategy that encourages reasoning to align with relational labels by rewarding relation-relevant phrases in reasoning. The reward is derived on a credit dictionary automatically extracted from correct predictions. Our experiments show that our framework improves both accuracy and explanation quality by addressing these two pitfalls. For example, COGRE with Qwen2.5-14B-Instruct on One-shot NYT29 achieves 24.65% F1, surpassing prior reasoning-based designs. Optimizing this approach with RL using HIT@DICT further improves performance by +23.46% points. Finally, human evaluation shows that our best model generates relational phrases closely aligned with gold labels, increasing human explanation quality ratings by 54% (relative). --- ### Latent Speech-Text Transformer Authors: Yen-Ju Lu, Yashesh Gaur, Wei Zhou, Benjamin Muller, Jesus Villalba, Najim Dehak, Luke Zettlemoyer, Gargi Ghosh, Mike Lewis, Srinivasan Iyer, Duc Le Published: 2025-10-07 Categories: cs.CL, cs.AI, cs.LG, eess.AS arXiv: https://arxiv.org/abs/2510.06195v2 Auto-regressive speech-text models pre-trained on interleaved text tokens and discretized speech tokens demonstrate strong speech understanding and generation, yet remain substantially less compute-efficient than text LLMs, partly due to the much longer sequences of speech tokens relative to text. This modality imbalance disproportionately allocates pre-training and inference compute to speech, potentially hindering effective cross-modal alignment and slowing performance scaling by orders of magnitude. We introduce the Latent Speech-Text Transformer (LST), which aggregates speech tokens into latent speech patches that serve as higher-level autoregressive units. This design aligns the sequence-modeling granularity between speech and text while improving computational efficiency. The resulting patches can align with textual units to facilitate cross-modal knowledge transfer and compactly capture recurring acoustic patterns such as silence. Across story-completion benchmarks under both compute-controlled and data-controlled settings, LST consistently improves speech accuracy while also improving text performance, achieving up to +6.5% absolute gain on speech HellaSwag in compute-controlled training (+5.3% in data-controlled training). Under compute-controlled scaling from 420M to 1.8B parameters in a near compute-optimal regime, gains grow with scale, and improvements persist up to 7B parameters under fixed-token budgets. These benefits extend to downstream tasks: LST stabilizes ASR adaptation and reduces the effective autoregressive sequence length during ASR and TTS inference, lowering computational cost without degrading reconstruction quality. The code is available at https://github.com/facebookresearch/lst. --- ### Overlap-aware segmentation for topological reconstruction of obscured objects Authors: J. Schueler, H. M. Araújo, S. N. Balashov, J. E. Borg, C. Brew, F. M. Brunbauer, C. Cazzaniga, A. Cottle, D. Edgeman, C. D. Frost, F. Garcia, D. Hunt et al. Published: 2025-10-07 Categories: hep-ex, astro-ph.IM, cs.CV arXiv: https://arxiv.org/abs/2510.06194v2 The separation of overlapping objects presents a significant challenge in scientific imaging. While deep learning segmentation-regression algorithms can predict pixel-wise intensities, they typically treat all regions equally rather than prioritizing overlap regions where attribution is most ambiguous. Recent advances in instance segmentation show that weighting regions of pixel overlap in training can improve segmentation boundary predictions in regions of overlap, but this idea has not yet been extended to segmentation regression. We address this with Overlap-Aware Segmentation of ImageS (OASIS): a new segmentation-regression framework with a weighted loss function designed to prioritize regions of object-overlap during training, enabling extraction of pixel intensities and topological features from heavily obscured objects. We demonstrate OASIS in the context of the MIGDAL experiment, which aims to directly image the Migdal effect--a rare process where electron emission is induced by nuclear scattering--in a low-pressure optical time projection chamber. This setting poses an extreme test case, as the target for reconstruction is a faint electron recoil track which is often heavily-buried within the order(s)-of-magnitude brighter nuclear recoil track. Compared to unweighted segmentation regression, we demonstrate OASIS's novel overlap region-targeted loss function weight to be the single most important training weight for improving intensity and topological reconstructions of the low-energy electron tracks that tend to be most dominated by pixel overlap. Averaging over eight training campaigns, we further show the addition of overlap-targeted weights to improve median intensity reconstruction errors from -41.1% to -13.3% for these low-energy electrons. These performance gains demonstrate OASIS as a generalizable methodology for recovering obscured signals in overlap-dominated regions. --- ### On Powerful Ways to Generate: Autoregression, Diffusion, and Beyond Authors: Chenxiao Yang, Cai Zhou, David Wipf, Zhiyuan Li Published: 2025-10-07 Categories: cs.LG arXiv: https://arxiv.org/abs/2510.06190v2 Diffusion language models have recently emerged as a competitive alternative to autoregressive language models. Beyond next-token generation, they are more efficient and flexible by enabling parallel and any-order token generation. However, despite empirical successes, their computational power and fundamental limitations remain poorly understood. In this paper, we formally study whether non-autoregressive generation in Masked Diffusion Models (MDM) enables solving problems beyond the reach of Auto-Regressive Models (ARM). Our results show that MDM with sufficiently large context length is computationally universal with decoding steps matching the optimal parallel time complexity in PRAM. However, when controlling for other factors, MDM's flexibility to generate in any-order does not expand what ARM can already solve. To address this, we propose a new form of generation called any-process generation, which extends MDM with capabilities to remask, insert and delete tokens, allowing self-correction, length-variable editing, and adaptive parallelism. Theoretically and empirically, we demonstrate these capabilities enable scalability to significantly harder reasoning problems that are otherwise intractable for ARM and vanilla MDM. Additionally, they prove essential for generation tasks where objects naturally evolve through non-sequential processes, crucial for extending current LLMs beyond natural language to domains such as coding and science. --- ### Barbarians at the Gate: How AI is Upending Systems Research Authors: Audrey Cheng, Shu Liu, Melissa Pan, Zhifei Li, Bowen Wang, Alex Krentsel, Tian Xia, Mert Cemri, Jongseok Park, Shuo Yang, Jeff Chen, Lakshya Agrawal et al. Published: 2025-10-07 Categories: cs.AI arXiv: https://arxiv.org/abs/2510.06189v3 Artificial Intelligence (AI) is starting to transform the research process as we know it by automating the discovery of new solutions. Given a task, the typical AI-driven approach is (i) to generate a set of diverse solutions, and then (ii) to verify these solutions and select one that solves the problem. Crucially, this approach assumes the existence of a reliable verifier, i.e., one that can accurately determine whether a solution solves the given problem. We argue that systems research, long focused on designing and evaluating new performance-oriented algorithms, is particularly well-suited for AI-driven solution discovery. This is because system performance problems naturally admit reliable verifiers: solutions are typically implemented in real systems or simulators, and verification reduces to running these software artifacts against predefined workloads and measuring performance. We term this approach as AI-Driven Research for Systems (ADRS), which iteratively generates, evaluates, and refines solutions. Using penEvolve, an existing open-source ADRS instance, we present case studies across diverse domains, including load balancing for multi-region cloud scheduling, Mixture-of-Experts inference, LLM-based SQL queries, and transaction scheduling. In multiple instances, ADRS discovers algorithms that outperform state-of-the-art human designs (e.g., achieving up to 5.0x runtime improvements or 50% cost reductions). We distill best practices for guiding algorithm evolution, from prompt design to evaluator construction, for existing frameworks. We then discuss the broader implications for the systems community: as AI assumes a central role in algorithm design, we argue that human researchers will increasingly focus on problem formulation and strategic guidance. Our results highlight both the disruptive potential and the urgent need to adapt systems research practices in the age of AI. --- ### BanglaTalk: Towards Real-Time Speech Assistance for Bengali Regional Dialects Authors: Jakir Hasan, Shubhashis Roy Dipta Published: 2025-10-07 Categories: cs.CL, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2510.06188v2 Real-time speech assistants are becoming increasingly popular for ensuring improved accessibility to information. Bengali, being a low-resource language with a high regional dialectal diversity, has seen limited progress in developing such systems. Existing systems are not optimized for real-time use and focus only on standard Bengali. In this work, we present BanglaTalk, the first real-time speech assistance system for Bengali regional dialects. BanglaTalk follows the client-server architecture and uses the Real-time Transport Protocol (RTP) to ensure low-latency communication. To address dialectal variation, we introduce a dialect-aware ASR system, BRDialect, developed by fine-tuning the IndicWav2Vec model in ten Bengali regional dialects. It outperforms the baseline ASR models by 12.41-33.98% on the RegSpeech12 dataset. Furthermore, BanglaTalk can operate at a low bandwidth of 24 kbps while maintaining an average end-to-end delay of 4.9 seconds. Low bandwidth usage and minimal end-to-end delay make the system both cost-effective and interactive for real-time use cases, enabling inclusive and accessible speech technology for the diverse community of Bengali speakers. Code is available in https://github.com/Jak57/BanglaTalk --- ### Automated Program Repair of Uncompilable Student Code Authors: Griffin Pitts, Aum Pandya, Darsh Rank, Tirth Bhatt, Muntasir Hoq, Bita Akram Published: 2025-10-07 Categories: cs.SE, cs.AI, cs.CY arXiv: https://arxiv.org/abs/2510.06187v3 A significant portion of student programming submissions in CS1 learning environments are uncompilable, limiting their use in student modeling and downstream knowledge tracing. Traditional modeling pipelines often exclude these cases, discarding observations of student learning. This study investigates automated program repair as a strategy to recover uncompilable code while preserving students' structural intent for use in student modeling. Within this framework, we assess large language models (LLMs) as repair agents under high- and low-context prompting conditions. Repairs were evaluated for compilability, edit distance, and preservation of students' original structure and logic. While all models produced compilable repairs, they differed in how well they preserve students' control flow and code structure, affecting their pedagogical utility. By recovering uncompilable submissions, this work enables richer and more comprehensive analyses of learners' coding processes and development over time. --- ### RECODE-H: A Benchmark for Research Code Development with Interactive Human Feedback Authors: Chunyu Miao, Henry Peng Zou, Yangning Li, Yankai Chen, Yibo Wang, Fangxin Wang, Yifan Li, Wooseong Yang, Bowei He, Xinni Zhang, Dianzhi Yu, Hanchen Yang et al. Published: 2025-10-07 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2510.06186v2 Large language models (LLMs) show the promise in supporting scientific research implementation, yet their ability to generate correct and executable code remains limited. Existing works largely adopt one-shot settings, ignoring the iterative and feedback-driven nature of realistic workflows of scientific research development. To address this gap, we present RECODE-H, a benchmark of 102 tasks from research papers and repositories that evaluates LLM agents through multi-turn interactions with LLM-simulated human feedback. It includes structured instructions,unit tests, and a five-level feedback hierarchy to reflect realistic researcher-agent collaboration. We further present ReCodeAgent, a framework that integrates feedback into iterative code generation. Experiments with leading LLMs, including GPT-5, Claude-Sonnet-4, DeepSeek-V3.1, and Gemini 2.5, show substantial performance gains with richer feedback, while also highlighting ongoing challenges in the generation of complex research code. RECODE-H establishes a foundation for developing adaptive, feedback-driven LLM agents in scientific research implementation --- ### Mixing Mechanisms: How Language Models Retrieve Bound Entities In-Context Authors: Yoav Gur-Arieh, Mor Geva, Atticus Geiger Published: 2025-10-07 Categories: cs.CL arXiv: https://arxiv.org/abs/2510.06182v2 A key component of in-context reasoning is the ability of language models (LMs) to bind entities for later retrieval. For example, an LM might represent "Ann loves pie" by binding "Ann" to "pie", allowing it to later retrieve "Ann" when asked "Who loves pie?" Prior research on short lists of bound entities found strong evidence that LMs implement such retrieval via a positional mechanism, where "Ann" is retrieved based on its position in context. In this work, we find that this mechanism generalizes poorly to more complex settings; as the number of bound entities in context increases, the positional mechanism becomes noisy and unreliable in middle positions. To compensate for this, we find that LMs supplement the positional mechanism with a lexical mechanism (retrieving "Ann" using its bound counterpart "pie") and a reflexive mechanism (retrieving "Ann" through a direct pointer). Through extensive experiments on nine models and ten binding tasks, we uncover a consistent pattern in how LMs mix these mechanisms to drive model behavior. We leverage these insights to develop a causal model combining all three mechanisms that estimates next token distributions with 95% agreement. Finally, we show that our model generalizes to substantially longer inputs of open-ended text interleaved with entity groups, further demonstrating the robustness of our findings in more natural settings. Overall, our study establishes a more complete picture of how LMs bind and retrieve entities in-context. --- ### Conformalized Gaussian processes for online uncertainty quantification over graphs Authors: Jinwen Xu, Qin Lu, Georgios B. Giannakis Published: 2025-10-07 Categories: cs.LG, eess.SP, stat.ML arXiv: https://arxiv.org/abs/2510.06181v1 Uncertainty quantification (UQ) over graphs arises in a number of safety-critical applications in network science. The Gaussian process (GP), as a classical Bayesian framework for UQ, has been developed to handle graph-structured data by devising topology-aware kernel functions. However, such GP-based approaches are limited not only by the prohibitive computational complexity, but also the strict modeling assumptions that might yield poor coverage, especially with labels arriving on the fly. To effect scalability, we devise a novel graph-aware parametric GP model by leveraging the random feature (RF)-based kernel approximation, which is amenable to efficient recursive Bayesian model updates. To further allow for adaptivity, an ensemble of graph-aware RF-based scalable GPs have been leveraged, with per-GP weight adapted to data arriving incrementally. To ensure valid coverage with robustness to model mis-specification, we wed the GP-based set predictors with the online conformal prediction framework, which post-processes the prediction sets using adaptive thresholds. Experimental results the proposed method yields improved coverage and efficient prediction sets over existing baselines by adaptively ensembling the GP models and setting the key threshold parameters in CP. --- ### Climate Model Tuning with Online Synchronization-Based Parameter Estimation Authors: Jordan Seneca, Suzanne Bintanja, Frank M. Selten Published: 2025-10-07 Categories: nlin.CD, cs.LG, physics.ao-ph arXiv: https://arxiv.org/abs/2510.06180v2 In climate science, the tuning of climate models is a computationally intensive problem due to the combination of the high-dimensionality of the system state and long integration times. Supermodelling is a technique which has shown the potential for reducing climate model biases by dynamically coupling multiple models together, and training their coupling on a short timescale. Here, we introduce a new approach called \emph{adaptive supermodeling}, where the internal model parameters of the member of a supermodel are tuned. We perform three experiments. We first directly optimize the internal parameters of a climate model. We then optimize the weights between two members of a supermodel in a classical supermodel approach. For a case designed to challenge the two previous methods, we implement adaptive supermodeling, which achieves a performance similar to a perfect model. --- ### Differentiable Model Predictive Control on the GPU Authors: Emre Adabag, Marcus Greiff, John Subosits, Thomas Lew Published: 2025-10-07 Categories: math.OC, cs.LG, eess.SY arXiv: https://arxiv.org/abs/2510.06179v1 Differentiable model predictive control (MPC) offers a powerful framework for combining learning and control. However, its adoption has been limited by the inherently sequential nature of traditional optimization algorithms, which are challenging to parallelize on modern computing hardware like GPUs. In this work, we tackle this bottleneck by introducing a GPU-accelerated differentiable optimization tool for MPC. This solver leverages sequential quadratic programming and a custom preconditioned conjugate gradient (PCG) routine with tridiagonal preconditioning to exploit the problem's structure and enable efficient parallelization. We demonstrate substantial speedups over CPU- and GPU-based baselines, significantly improving upon state-of-the-art training times on benchmark reinforcement learning and imitation learning tasks. Finally, we showcase the method on the challenging task of reinforcement learning for driving at the limits of handling, where it enables robust drifting of a Toyota Supra through water puddles. --- ### VecInfer: Efficient LLM Inference with Low-Bit KV Cache via Outlier-Suppressed Vector Quantization Authors: Dingyu Yao, Chenxu Yang, Zhengyang Tong, Zheng Lin, Wei Liu, Jian Luan, Weiping Wang Published: 2025-10-07 Categories: cs.CL arXiv: https://arxiv.org/abs/2510.06175v1 The Key-Value (KV) cache introduces substantial memory overhead during large language model (LLM) inference. Although existing vector quantization (VQ) methods reduce KV cache usage and provide flexible representational capacity across bit-widths, they suffer severe performance degradation at ultra-low bit-widths due to key cache outliers that hinder effective codebook utilization. To address this challenge, we propose VecInfer, a novel VQ method for aggressive KV cache compression while enabling efficient inference. By applying smooth and Hadamard transformations, VecInfer suppresses outliers in the key cache, enabling the codebook to comprehensively cover the original data distribution and thereby reducing quantization difficulty. To facilitate efficient deployment, we design an optimized CUDA kernel that fuses computation with dequantization to minimize memory access overhead. Extensive evaluations demonstrate that VecInfer consistently outperforms existing quantization baselines across both long-context understanding and mathematical reasoning tasks. With only 2-bit quantization, VecInfer achieves performance comparable to full precision, while delivering up to $\mathbf{2.7\times}$ speedup in large-batch self-attention computation and $\mathbf{8.3\times}$ reduction in single-batch end-to-end latency on Llama-3.1-8B with a 196k sequence length. --- ### Thermodynamic Performance Limits for Score-Based Diffusion Models Authors: Nathan X. Kodama, Michael Hinczewski Published: 2025-10-07 Categories: cs.LG, cond-mat.stat-mech arXiv: https://arxiv.org/abs/2510.06174v1 We establish a fundamental connection between score-based diffusion models and non-equilibrium thermodynamics by deriving performance limits based on entropy rates. Our main theoretical contribution is a lower bound on the negative log-likelihood of the data that relates model performance to entropy rates of diffusion processes. We numerically validate this bound on a synthetic dataset and investigate its tightness. By building a bridge to entropy rates - system, intrinsic, and exchange entropy - we provide new insights into the thermodynamic operation of these models, drawing parallels to Maxwell's demon and implications for thermodynamic computing hardware. Our framework connects generative modeling performance to fundamental physical principles through stochastic thermodynamics. --- ### Smartphone-based iris recognition through high-quality visible-spectrum iris image capture.V2 Authors: Naveenkumar G Venkataswamy, Yu Liu, Soumyabrata Dey, Stephanie Schuckers, Masudul H Imtiaz Published: 2025-10-07 Categories: eess.IV, cs.AI, cs.CV arXiv: https://arxiv.org/abs/2510.06170v3 Smartphone-based iris recognition in the visible spectrum (VIS) remains difficult due to illumination variability, pigmentation differences, and the absence of standardized capture controls. This work presents a compact end-to-end pipeline that enforces ISO/IEC 29794-6 quality compliance at acquisition and demonstrates that accurate VIS iris recognition is feasible on commodity devices. Using a custom Android application performing real-time framing, sharpness evaluation, and feedback, we introduce the CUVIRIS dataset of 752 compliant images from 47 subjects. A lightweight MobileNetV3-based multi-task segmentation network (LightIrisNet) is developed for efficient on-device processing, and a transformer matcher (IrisFormer) is adapted to the VIS domain. Under a standardized protocol and comparative benchmarking against prior CNN baselines, OSIRIS attains a TAR of 97.9% at FAR=0.01 (EER=0.76%), while IrisFormer, trained only on UBIRIS.v2, achieves an EER of 0.057% on CUVIRIS. The acquisition app, trained models, and a public subset of the dataset are released to support reproducibility. These results confirm that standardized capture and VIS-adapted lightweight models enable accurate and practical iris recognition on smartphones. --- ### Higher-Order Feature Attribution: Bridging Statistics, Explainable AI, and Topological Signal Processing Authors: Kurt Butler, Guanchao Feng, Petar Djuric Published: 2025-10-07 Categories: cs.LG, eess.SP, math.ST, stat.ML arXiv: https://arxiv.org/abs/2510.06165v2 Feature attributions are post-training analysis methods that assess how various input features of a machine learning model contribute to an output prediction. Their interpretation is straightforward when features act independently, but it becomes less clear when the predictive model involves interactions, such as multiplicative relationships or joint feature contributions. In this work, we propose a general theory of higher-order feature attribution, which we develop on the foundation of Integrated Gradients (IG). This work extends existing frameworks in the literature on explainable AI. When using IG as the method of feature attribution, we discover natural connections to statistics and topological signal processing. We provide several theoretical results that establish the theory, and we validate our theory on a few examples. --- ### TabPFN-Wide: Continued Pre-Training for Extreme Feature Counts Authors: Christopher Kolberg, Jules Kreuer, Jonas Huurdeman, Sofiane Ouaari, Katharina Eggensperger, Nico Pfeifer Published: 2025-10-07 Categories: cs.LG arXiv: https://arxiv.org/abs/2510.06162v2 Revealing novel insights from the relationship between molecular measurements and pathology remains a very impactful application of machine learning in biomedicine. Data in this domain typically contain only a few observations but thousands of potentially noisy features, posing challenges for conventional tabular machine learning approaches. While prior-data fitted networks emerge as foundation models for predictive tabular data tasks, they are currently not suited to handle large feature counts (>500). Although feature reduction enables their application, it hinders feature importance analysis. We propose a strategy that extends existing models through continued pre-training on synthetic data sampled from a customized prior. The resulting model, TabPFN-Wide, matches or exceeds its base model's performance, while exhibiting improved robustness to noise. It seamlessly scales beyond 30,000 categorical and continuous features, regardless of noise levels, while maintaining inherent interpretability, which is critical for biomedical applications. Our results demonstrate that prior-informed adaptation is suitable to enhance the capability of foundation models for high-dimensional data. On real-world omics datasets, we show that many of the most relevant features identified by the model overlap with previous biological findings, while others propose potential starting points for future studies. --- ### LLMs as Policy-Agnostic Teammates: A Case Study in Human Proxy Design for Heterogeneous Agent Teams Authors: Aju Ani Justus, Chris Baber Published: 2025-10-07 Categories: cs.LG, cs.AI, cs.HC arXiv: https://arxiv.org/abs/2510.06151v1 A critical challenge in modelling Heterogeneous-Agent Teams is training agents to collaborate with teammates whose policies are inaccessible or non-stationary, such as humans. Traditional approaches rely on expensive human-in-the-loop data, which limits scalability. We propose using Large Language Models (LLMs) as policy-agnostic human proxies to generate synthetic data that mimics human decision-making. To evaluate this, we conduct three experiments in a grid-world capture game inspired by Stag Hunt, a game theory paradigm that balances risk and reward. In Experiment 1, we compare decisions from 30 human participants and 2 expert judges with outputs from LLaMA 3.1 and Mixtral 8x22B models. LLMs, prompted with game-state observations and reward structures, align more closely with experts than participants, demonstrating consistency in applying underlying decision criteria. Experiment 2 modifies prompts to induce risk-sensitive strategies (e.g. "be risk averse"). LLM outputs mirror human participants' variability, shifting between risk-averse and risk-seeking behaviours. Finally, Experiment 3 tests LLMs in a dynamic grid-world where the LLM agents generate movement actions. LLMs produce trajectories resembling human participants' paths. While LLMs cannot yet fully replicate human adaptability, their prompt-guided diversity offers a scalable foundation for simulating policy-agnostic teammates. --- ### Implicit Updates for Average-Reward Temporal Difference Learning Authors: Hwanwoo Kim, Dongkyu Derek Cho, Eric Laber Published: 2025-10-07 Categories: stat.ML, cs.LG arXiv: https://arxiv.org/abs/2510.06149v1 Temporal difference (TD) learning is a cornerstone of reinforcement learning. In the average-reward setting, standard TD($λ$) is highly sensitive to the choice of step-size and thus requires careful tuning to maintain numerical stability. We introduce average-reward implicit TD($λ$), which employs an implicit fixed point update to provide data-adaptive stabilization while preserving the per iteration computational complexity of standard average-reward TD($λ$). In contrast to prior finite-time analyses of average-reward TD($λ$), which impose restrictive step-size conditions, we establish finite-time error bounds for the implicit variant under substantially weaker step-size requirements. Empirically, average-reward implicit TD($λ$) operates reliably over a much broader range of step-sizes and exhibits markedly improved numerical stability. This enables more efficient policy evaluation and policy learning, highlighting its effectiveness as a robust alternative to average-reward TD($λ$). --- ### Non-iid hypothesis testing: from classical to quantum Authors: Giacomo De Palma, Marco Fanizza, Connor Mowry, Ryan O'Donnell Published: 2025-10-07 Categories: quant-ph, cs.DS, cs.LG arXiv: https://arxiv.org/abs/2510.06147v1 We study hypothesis testing (aka state certification) in the non-identically distributed setting. A recent work (Garg et al. 2023) considered the classical case, in which one is given (independent) samples from $T$ unknown probability distributions $p_1, \dots, p_T$ on $(d) = \{1, 2, \dots, d\}$, and one wishes to accept/reject the hypothesis that their average $p_{\mathrm{avg}}$ equals a known hypothesis distribution $q$. Garg et al. showed that if one has just $c = 2$ samples from each $p_i$, and provided $T \gg \frac{\sqrt{d}}{ε^2} + \frac{1}{ε^4}$, one can (whp) distinguish $p_{\mathrm{avg}} = q$ from $d_{\mathrm{TV}}(p_{\mathrm{avg}},q) > ε$. This nearly matches the optimal result for the classical iid setting (namely, $T \gg \frac{\sqrt{d}}{ε^2}$). Besides optimally improving this result (and generalizing to tolerant testing with more stringent distance measures), we study the analogous problem of hypothesis testing for non-identical quantum states. Here we uncover an unexpected phenomenon: for any $d$-dimensional hypothesis state $σ$, and given just a single copy ($c = 1$) of each state $ρ_1, \dots, ρ_T$, one can distinguish $ρ_{\mathrm{avg}} = σ$ from $D_{\mathrm{tr}}(ρ_{\mathrm{avg}},σ) > ε$ provided $T \gg d/ε^2$. (Again, we generalize to tolerant testing with more stringent distance measures.) This matches the optimal result for the iid case, which is surprising because doing this with $c = 1$ is provably impossible in the classical case. We also show that the analogous phenomenon happens for the non-iid extension of identity testing between unknown states. A technical tool we introduce may be of independent interest: an Efron-Stein inequality, and more generally an Efron-Stein decomposition, in the quantum setting. --- ### Bimanual 3D Hand Motion and Articulation Forecasting in Everyday Images Authors: Aditya Prakash, Richard Li, David Forsyth, Saurabh Gupta Published: 2025-10-07 Categories: cs.CV, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2510.06145v2 We tackle the problem of forecasting bimanual 3D hand motion and articulation from a single image in everyday settings. To address the lack of 3D hand annotations in diverse settings, we design an annotation pipeline consisting of a diffusion model to lift 2D hand keypoint sequences to 4D hand motion. For the forecasting model, we adopt a diffusion loss to account for the multimodality in hand motion distribution. Extensive experiments on 6 datasets show the benefits of training with our imputed labels (14% improvement) and the effectiveness of our lifting (45% better) & forecasting (16.4% gain) models, over the best baselines, especially in zero-shot generalization to everyday images. --- ### RoSE: Round-robin Synthetic Data Evaluation for Selecting LLM Generators without Human Test Sets Authors: Jan Cegin, Branislav Pecher, Ivan Srba, Jakub Simko Published: 2025-10-07 Categories: cs.CL arXiv: https://arxiv.org/abs/2510.06143v2 LLMs are powerful generators of synthetic data, which are used for training smaller, specific models. This is especially valuable for low-resource languages, where human-labelled data is scarce but LLMs can still produce high-quality text. However, LLMs differ in how useful their outputs are for training. Selecting the best LLM as a generator is challenging because extrinsic evaluation requires costly human annotations (which are often unavailable for low-resource languages), while intrinsic metrics correlate poorly with downstream performance. We introduce Round robin Synthetic data Evaluation (RoSE), a proxy metric for selecting the best LLM generator without human test sets. RoSE trains a small model on the outputs of a candidate generator (LLM) and then evaluates it on generated synthetic examples from all other candidate LLMs. The final RoSE score is the mean performance of this small model. Across six LLMs, eleven languages, and three tasks (sentiment, topic, intent), RoSE identifies the optimal generator more often than any other intrinsic heuristics. RoSE outperforms intrinsic heuristics and comes within 0.76 percentage points of the optimal generator baseline. This result is measured in terms of downstream performance, obtained by training a small model on the chosen generator's outputs (optimal vs. proxy metric selected) and evaluating it on human-labelled test data. Additionally, RoSE is the only metric to achieve a positive correlation with performance on human test data. --- ### High-Probability Convergence Guarantees of Decentralized SGD Authors: Aleksandar Armacki, Ali H. Sayed Published: 2025-10-07 Categories: cs.LG, cs.MA, math.OC arXiv: https://arxiv.org/abs/2510.06141v6 Convergence in high-probability (HP) has attracted increasing interest, due to implying exponentially decaying tail bounds and strong guarantees for individual runs of an algorithm. While many works study HP guarantees in centralized settings, much less is understood in the decentralized setup, where existing works require strong assumptions, like uniformly bounded gradients, or asymptotically vanishing noise. This results in a significant gap between the assumptions used to establish convergence in the HP and the mean-squared error (MSE) sense, and is also contrary to centralized settings, where it is known that $\mathtt{SGD}$ converges in HP under the same conditions on the cost function as needed for MSE convergence. Motivated by these observations, we study the HP convergence of Decentralized $\mathtt{SGD}$ ($\mathtt{DSGD}$) in the presence of light-tailed noise, providing several strong results. First, we show that $\mathtt{DSGD}$ converges in HP under the same conditions on the cost as in the MSE sense, removing the restrictive assumptions used in prior works. Second, our sharp analysis yields order-optimal rates for both non-convex and strongly convex costs. Third, we establish a linear speed-up in the number of users, leading to matching or strictly better transient times than those obtained from MSE results, further underlining the tightness of our analysis. To the best of our knowledge, this is the first work that shows $\mathtt{DSGD}$ achieves a linear speed-up in the HP sense. Our relaxed assumptions and sharp rates stem from several technical results of independent interest, including a result on the variance-reduction effect of decentralized methods in the HP sense, as well as a novel bound on the moment-generating function of strongly convex costs, of interest even in centralized settings. Numerical experiments validate our theory. --- ### Deforming Videos to Masks: Flow Matching for Referring Video Segmentation Authors: Zanyi Wang, Dengyang Jiang, Liuzhuozheng Li, Sizhe Dang, Chengzu Li, Harry Yang, Guang Dai, Mengmeng Wang, Jingdong Wang Published: 2025-10-07 Categories: cs.CV arXiv: https://arxiv.org/abs/2510.06139v2 Referring Video Object Segmentation (RVOS) requires segmenting specific objects in a video guided by a natural language description. The core challenge of RVOS is to anchor abstract linguistic concepts onto a specific set of pixels and continuously segment them through the complex dynamics of a video. Faced with this difficulty, prior work has often decomposed the task into a pragmatic `locate-then-segment' pipeline. However, this cascaded design creates an information bottleneck by simplifying semantics into coarse geometric prompts (e.g, point), and struggles to maintain temporal consistency as the segmenting process is often decoupled from the initial language grounding. To overcome these fundamental limitations, we propose FlowRVS, a novel framework that reconceptualizes RVOS as a conditional continuous flow problem. This allows us to harness the inherent strengths of pretrained T2V models, fine-grained pixel control, text-video semantic alignment, and temporal coherence. Instead of conventional generating from noise to mask or directly predicting mask, we reformulate the task by learning a direct, language-guided deformation from a video's holistic representation to its target mask. Our one-stage, generative approach achieves new state-of-the-art results across all major RVOS benchmarks. Specifically, achieving a J&F of 51.1 in MeViS (+1.6 over prior SOTA) and 73.3 in the zero shot Ref-DAVIS17 (+2.7), demonstrating the significant potential of modeling video understanding tasks as continuous deformation processes. --- ### Multi-Task Reinforcement Learning with Language-Encoded Gated Policy Networks Authors: Rushiv Arora Published: 2025-10-07 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2510.06138v1 Multi-task reinforcement learning often relies on task metadata -- such as brief natural-language descriptions -- to guide behavior across diverse objectives. We present Lexical Policy Networks (LEXPOL), a language-conditioned mixture-of-policies architecture for multi-task RL. LEXPOL encodes task metadata with a text encoder and uses a learned gating module to select or blend among multiple sub-policies, enabling end-to-end training across tasks. On MetaWorld benchmarks, LEXPOL matches or exceeds strong multi-task baselines in success rate and sample efficiency, without task-specific retraining. To analyze the mechanism, we further study settings with fixed expert policies obtained independently of the gate and show that the learned language gate composes these experts to produce behaviors appropriate to novel task descriptions and unseen task combinations. These results indicate that natural-language metadata can effectively index and recombine reusable skills within a single policy. --- ### Pushing Test-Time Scaling Limits of Deep Search with Asymmetric Verification Authors: Weihao Zeng, Keqing He, Chuqiao Kuang, Xiaoguang Li, Junxian He Published: 2025-10-07 Categories: cs.AI arXiv: https://arxiv.org/abs/2510.06135v1 Test-time compute can be scaled both sequentially and in parallel. Sequential scaling involves lengthening the generation process, while parallel scaling involves verifying and selecting among multiple candidate outputs. Combining these two strategies has led to the most powerful AI systems, such as Grok 4 Heavy and GPT-5 Pro. In certain contexts (e.g., solving Sudoku puzzles), verifying responses can be substantially easier than generating them. This property, referred to as \emph{asymmetric verification}, highlights the strong potential of test-time scaling (TTS). In this work, we study both sequential and parallel TTS of deep search agents, motivated by the intuition that verification in this setting is often much easier than generation. In experiments, we first show that sequential scaling methods, such as budget forcing, can be effective initially but soon degrade performance. Leveraging asymmetric verification, however, we are able to achieve substantial improvements by allocating only a modest amount of compute to the verifier. We conduct experiments with flagship open-source models and extend them to their ``Heavy'' variants through TTS. These deep research agents achieve gains of up to 27 absolute points on benchmarks such as BrowseComp. Remarkably, as an open-source alternative, GLM-4.5 Heavy reaches accuracy of {\bf 54.0\%} on BrowseComp and {\bf 66.0\%} on GAIA, placing it comparable to the best proprietary choices such as OpenAI Deep Research. Tongyi-DeepResearch Heavy further achieves {\bf 69.0\%} accuracy on BrowseComp, greatly surpassing the best proprietary results. --- ### CreditDecoding: Accelerating Parallel Decoding in Diffusion Large Language Models with Trace Credit Authors: Kangyu Wang, Zhiyun Jiang, Haibo Feng, Weijia Zhao, Lin Liu, Jianguo Li, Zhenzhong Lan, Weiyao Lin Published: 2025-10-07 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2510.06133v3 Diffusion large language models (dLLMs) generate text through iterative denoising. In commonly adopted parallel decoding schemes, each step confirms only high-confidence positions while remasking the others. By analyzing dLLM denoising traces, we uncover a key inefficiency: models often predict the correct target token several steps before its confidence becomes high enough to be decoded. This gap between early prediction and late decoding forces repeated remasking of already-correct tokens, causing redundant iterations and limiting acceleration. To exploit this temporal redundancy, we introduce Trace Credit to quantify a token's decoding potential by accumulating historical evidence. Building on this, we propose CreditDecoding, a training-free parallel decoding method that fuses Trace Credit with current logits to boost the confidence of correct but underconfident tokens, thereby accelerating denoising and improving robustness. On eight benchmarks, CreditDecoding achieves up to 5.48 times speedup with +0.48 accuracy on LLaDA-8B and consistently improves performance across diverse dLLM architectures and parameter scales. It further scales to long contexts and remains orthogonal to mainstream inference optimizations, making it a practical and widely applicable solution. --- ### Discrete Diffusion Models with MLLMs for Unified Medical Multimodal Generation Authors: Jiawei Mao, Yuhan Wang, Lifeng Chen, Can Zhao, Yucheng Tang, Dong Yang, Liangqiong Qu, Daguang Xu, Yuyin Zhou Published: 2025-10-07 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2510.06131v1 Recent advances in generative medical models are constrained by modality-specific scenarios that hinder the integration of complementary evidence from imaging, pathology, and clinical notes. This fragmentation limits their evolution into foundation models that can learn and reason across the full spectrum of biomedical data. We propose MeDiM, the first medical discrete diffusion model that learns shared distributions across modalities without modality-specific components. MeDiM unifies multiple generative tasks: translating between images and text, and jointly producing image-report pairs across domains in response to prompts. Built on a discrete diffusion framework, MeDiM bridges vision and language representations through a shared probabilistic space. To enable unified and flexible medical generation, we employ a multimodal large language model (MLLM) as the diffusion backbone, leveraging its prior knowledge and cross-modal reasoning. Two key designs are introduced: (1) removing the causal attention mask for bidirectional context, and (2) injecting continuous timestep embeddings for diffusion awareness. Experiments demonstrate high-fidelity medical generation (FID 16.60 on MIMIC-CXR and FID 24.19 on PathGen) and accurate report generation (METEOR 0.2650 and 0.2580). Jointly generated image-report pairs further enhance downstream performance (plus6.43 percent BLEU-1, plus18.57 percent BLEU-2, plus31.58 percent BLEU-3, plus4.80 percent METEOR), showing that MeDiM supports coherent and clinically grounded multimodal outputs. --- ### Parallel Tokenizers: Rethinking Vocabulary Design in Cross-Lingual Transfer of Low-Resource Languages Authors: Muhammad Dehan Al Kautsar, Fajri Koto Published: 2025-10-07 Categories: cs.CL arXiv: https://arxiv.org/abs/2510.06128v3 Tokenization forms the basis of multilingual language models, yet existing methods often limit cross-lingual transfer by mapping semantically equivalent words to different embeddings. For example, 'I eat rice' in English and 'Ina cin shinkafa' in Hausa are typically mapped to different vocabulary indices, preventing shared representations and limiting cross-lingual generalization. This problem is even more pronounced in low-resource languages, where shared representations could offer the greatest benefit. We introduce parallel tokenizers, a new framework that first trains tokenizers monolingually and then aligns their vocabularies exhaustively using bilingual dictionaries or word-to-word translation. This alignment enforces a shared semantic space across languages while naturally improving fertility balance. To assess their effectiveness, we pretrain a transformer encoder from scratch on thirteen low-resource languages and evaluate it on sentiment analysis, hate speech detection, emotion classification, and sentence embedding similarity. Across all tasks, models trained with parallel tokenizers outperform conventional multilingual baselines, confirming that rethinking tokenization is essential for advancing multilingual representation learning--especially in low-resource settings. --- ### lm-Meter: Unveiling Runtime Inference Latency for On-Device Language Models Authors: Haoxin Wang, Xiaolong Tu, Hongyu Ke, Huirong Chai, Dawei Chen, Kyungtae Han Published: 2025-10-07 Categories: cs.LG, cs.PF arXiv: https://arxiv.org/abs/2510.06126v1 Large Language Models (LLMs) are increasingly integrated into everyday applications, but their prevalent cloud-based deployment raises growing concerns around data privacy and long-term sustainability. Running LLMs locally on mobile and edge devices (on-device LLMs) offers the promise of enhanced privacy, reliability, and reduced communication costs. However, realizing this vision remains challenging due to substantial memory and compute demands, as well as limited visibility into performance-efficiency trade-offs on resource-constrained hardware. We propose lm-Meter, the first lightweight, online latency profiler tailored for on-device LLM inference. lm-Meter captures fine-grained, real-time latency at both phase (e.g., embedding, prefill, decode, softmax, sampling) and kernel levels without auxiliary devices. We implement lm-Meter on commercial mobile platforms and demonstrate its high profiling accuracy with minimal system overhead, e.g., only 2.58% throughput reduction in prefill and 0.99% in decode under the most constrained Powersave governor. Leveraging lm-Meter, we conduct comprehensive empirical studies revealing phase- and kernel-level bottlenecks in on-device LLM inference, quantifying accuracy-efficiency trade-offs, and identifying systematic optimization opportunities. lm-Meter provides unprecedented visibility into the runtime behavior of LLMs on constrained platforms, laying the foundation for informed optimization and accelerating the democratization of on-device LLM systems. Code and tutorials are available at https://github.com/amai-gsu/LM-Meter. --- ### Downsized and Compromised?: Assessing the Faithfulness of Model Compression Authors: Moumita Kamal, Douglas A. Talbert Published: 2025-10-07 Categories: cs.LG arXiv: https://arxiv.org/abs/2510.06125v1 In real-world applications, computational constraints often require transforming large models into smaller, more efficient versions through model compression. While these techniques aim to reduce size and computational cost without sacrificing performance, their evaluations have traditionally focused on the trade-off between size and accuracy, overlooking the aspect of model faithfulness. This limited view is insufficient for high-stakes domains like healthcare, finance, and criminal justice, where compressed models must remain faithful to the behavior of their original counterparts. This paper presents a novel approach to evaluating faithfulness in compressed models, moving beyond standard metrics. We introduce and demonstrate a set of faithfulness metrics that capture how model behavior changes post-compression. Our contributions include introducing techniques to assess predictive consistency between the original and compressed models using model agreement, and applying chi-squared tests to detect statistically significant changes in predictive patterns across both the overall dataset and demographic subgroups, thereby exposing shifts that aggregate fairness metrics may obscure. We demonstrate our approaches by applying quantization and pruning to artificial neural networks (ANNs) trained on three diverse and socially meaningful datasets. Our findings show that high accuracy does not guarantee faithfulness, and our statistical tests detect subtle yet significant shifts that are missed by standard metrics, such as Accuracy and Equalized Odds. The proposed metrics provide a practical and more direct method for ensuring that efficiency gains through compression do not compromise the fairness or faithfulness essential for trustworthy AI. ---