Beacon is a new agentic visual reasoning model that improves multimodal large language models (MLLMs) by better deciding when to use tools and how to use them. It introduces two key concepts—Mode Adaptiveness, which ensures tools are invoked only when necessary, and Tool Effect, which measures the net benefit of tool use— and trains the model with supervised fine‑tuning and reinforcement learning that rewards necessity-aware decisions and expands capability through expert hints. Across 13 benchmarks, Beacon outperforms other open‑source models, achieving the highest average score and the largest net tool‑gain on diagnostic tests.
By Qixun Wang, Yang Shi, Letian Cheng, Zhuoran Zhang, Yan He, Yuqi Tang, Qi Zhang, Xinlei Yu, Ruizhe Chen, Tianrun Xu, Yuanxing Zhang, Pengfei Wan, Haotian Wang, Xianghua Ying
PARSEE-VAD is a training‑free online video anomaly detection framework that separates semantic evidence acquisition from score‑state evolution. It uses Proposition‑Aware Reasoning to extract structured propositional evidence from the current causal window and selectively activates more specific queries, while Streaming Evidence Escalation maps this evidence into a compact score‑domain event state and propagates only the bounded state to maintain temporal continuity. Experiments on four benchmarks show strong performance with reduced specialist computation and sparse score‑state propagation, supporting a current‑first principle for streaming multimodal inference.
By Ji Wang, Shuangqing Zhang, Guo-Sen Xie, Fang Zhao
PhysWAM is a unified world-action model for autonomous driving that jointly denoises multiview video, metric depth, and ego motion using a flow‑matching transformer. It introduces Coupled Point Projection (CPP), a geometric constraint that aligns generated depth points with LiDAR data after applying the predicted SE(3) ego motion, thereby enforcing physical consistency. At inference, trajectory selection uses a simple label‑free consensus rule, and the model demonstrates strong planning performance, zero‑shot transfer to unseen environments, and accurate, temporally coherent depth and video predictions.
By Dhruv Parikh, Fengcheng Yu, Quankai Gao, Jiawei Yang, Junjie Ye, Maulik Bhatt, Thang Vu, Charles Ochoa, Rowan McAllister, Igor Vasiljevic, Rajgopal Kannan, Viktor Prasanna, Vitor Guizilini, Yue Wang
The paper investigates whether input ablations on predictive models can reliably reveal the importance of information for explaining human sequential choice behavior. Using two synthetic bandit tasks with known generating policies, the authors compare GRUs, Transformers, a fine‑tuned LLaMA, and cognitive models under varied reward contributions. They find that while neural models can predict choices well, their responses to ablations often diverge from the true generating process, indicating that predictive accuracy alone does not guarantee faithful model ablations.
By Hanbo Xie
RelayVSR introduces a streaming video super‑resolution framework that combines a large generative model, which produces reference latents for sparse keyframes, with a lightweight Dual‑Memory Video Transformer that super‑resolves every frame using these references and low‑resolution input. The method employs Video‑Aware Reference Optimization (VARO), a reinforcement‑learning strategy that optimizes both system‑level video quality and reference‑level keyframe fidelity, outperforming direct joint training. On 1080p video, RelayVSR achieves 29.29 FPS with modest GPU memory usage, significantly faster and more efficient than the FlashVSR‑Tiny baseline.
By Xijun Wang, Xin Li, Zirui Lang, Suhang Yao, Haoran Li, Zhibo Chen
AdvancedMathBench is a new benchmark suite that evaluates large language models on advanced mathematical proof generation and verification. It includes ProverBench, with 245 problems from undergraduate to doctoral qualifying‑exam levels, and VerifierBench, which tests models’ ability to judge proof validity using 888 expert‑annotated trajectories. The suite features an automatic verification pipeline trained on expert data, and results show that even state‑of‑the‑art models perform poorly, highlighting a gap between generation and verification skills.
By Lingkai Kong, Zijian Wu, Yuzhe Gu, Haiteng Zhao, Zhouqi Hua, Wenyong Huang, Shuang Sun, Zhicheng Xiong, Xiaotian Zhang, Shuya Zhao, Yan Wang, Disheng Xu, Wenwei Zhang, Kai Chen
The study trains ten open‑weight large language models (LLMs) to predict their own accuracy on factual multiple‑choice questions before answering. Results show that the models’ confidence signals split into two distinct patterns: early in training, confidence aligns with output consistency (how concentrated the answer distribution is), while later, it aligns with true accuracy but only on data similar to the training set. This indicates that calibration training may not universally teach LLMs to detect their own errors.
By Nicolas Yax, Stefano Palminteri, Pierre-Yves Oudeyer
StateTape introduces a new framework for long‑horizon coding agents that rewrites the agent’s context as the code repository changes, rather than letting the context grow with every observation. It models the repository as a symbol‑level code graph, using a tape to mark symbols altered by each write and a manager model to resolve stale records. The authors provide theoretical analysis, a new benchmark called TraceBench, and empirical results showing higher resolve rates across six agents and three edit‑heavy benchmarks with minimal computational overhead.
By Ziyang Yu, Liang Zhao, Bowen Zhu, Hasibul Haque
The paper introduces the first systematic study of zero‑knowledge (ZK)‑friendly quantization for large language models (LLMs). It defines what makes a quantization scheme suitable for ZK proof generation and evaluates nine models, including Qwen2.5‑14B and Qwen3‑30B‑A3B, across various weight, activation, and nonlinear lookup precisions. Findings reveal that activation precision is more critical than weight precision, nonlinear lookup approximations can dominate utility loss, and that reducing bit‑width or lookup size does not always lead to proportional proving cost savings, highlighting the need for operator‑aware precision selection.
By Taeung Yoon, Yupeng Zhang, Xiaojing Liao
HyperSAM is a promptable foundation model for hyperspectral remote sensing that integrates a data‑centric synthesis pipeline with a spectral adaptation architecture based on Segment Anything Model 3 (SAM3). The model generates full‑spectrum hyperspectral cubes from high‑resolution multispectral imagery using a physics‑informed abundance‑transfer generator, and employs SAM3‑derived pseudo‑masks for object‑centric supervision. With a frozen SAM3 RGB branch, a trainable hyperspectral encoder, ControlNet‑style feature injection, and a mixture‑of‑experts mask refiner, HyperSAM demonstrates strong generalization across diverse hyperspectral tasks such as classification, anomaly detection, change detection, target detection, and airborne oil‑spill mapping.
By Li Pang, Xinqiao Wu, Jing Yao, Pedram Ghamisi, Jun Zhou, Zhengchao Chen, Deyu Meng, Xiangyong Cao
The paper introduces Reliable Parallel Decoding (RPD) for masked diffusion language models, addressing the unreliability of committing multiple predictions from a single forward pass. Diagnostics reveal that confidence alone is insufficient, as confident end‑sequence predictions can preempt necessary upstream computations, and downstream predictions degrade with upstream uncertainty. RPD selects candidates based on layer‑wise stability and final confidence, committing them under an entropy budget while deferring uncertain predictions, achieving superior throughput and competitive accuracy on LLaDA and Dream benchmarks.
By Zhenghao He, Bohan Liu, Guangzhi Xiong, Aidong Zhang
The paper introduces a controlled benchmark for evaluating large language models (LLMs) on key‑value pair extraction from documents with varying levels of OCR noise. It tests 136 configurations across five instruction‑tuned open‑weight LLMs, three datasets, and four text‑quality conditions, using deterministic decoding to generate 17,688 document‑level inferences. The study finds that clean‑text performance does not reliably predict real‑world robustness, model rankings can reverse under noisy conditions, and few‑shot demonstrations do not always improve accuracy, highlighting reliability risks in OCR‑to‑LLM pipelines.
By Zahra Anvari
CORE-BREW is a new multi‑bit watermarking method for large language models that uses log‑likelihood ratios for soft‑decision decoding, targeting a fixed hit rate to calibrate the watermark channel. It introduces entropy‑aware erasures to reduce perturbations in low‑entropy contexts and combines likelihood‑based scoring with soft‑decision list decoding to better exploit token‑level reliability. Experiments on open‑source LLMs show that CORE‑BREW improves detection robustness and payload recovery compared to the BREW baseline while keeping false‑positive rates low and maintaining translation quality metrics close to unwatermarked text.
By Joeun Kim, HoEun Kim, Young-Sik Kim
FastVR is a streaming video restoration framework that uses a one‑step diffusion model to achieve strong restoration quality and temporal consistency while processing 1080p video at 11 FPS on a single H20 GPU. It addresses efficiency bottlenecks by combining a lightweight VAE with chunk‑wise causal attention, and improves inference speed and restoration quality through velocity consistency regularization and continuous trajectory learning during training. Experiments demonstrate that FastVR outperforms diffusion baselines in efficiency and achieves state‑of‑the‑art performance on both synthetic and real‑world benchmarks.
By Xiaoxu Chen, Qin Yang, Haoran Bai, Sibin Deng, Ying Chen
Spatial-OPSD is a label‑free self‑improvement framework for vision‑language models that leverages spatial priors such as depth, 3D relations, and camera geometry to provide dense token‑level supervision. During training, a privileged teacher uses these priors while the student learns from only the original visual‑language input, and a recursive round‑wise scheme allows repeated self‑improvement without moving the teacher. Across four VLM families, one round of Spatial‑OPSD improves the five‑benchmark average, and three rounds push a strong spatially specialized model to the open‑source frontier, achieving the highest average among open models and best results on three of five spatial reasoning benchmarks.
By Zhenyu Liu, Zhangquan Chen, Keyi Chen, Mingze Sun, Xiang An, Haodong Jing, Ruqi Huang
The paper investigates a new failure mode in relational in-context learning called support‑set target leakage, where target‑derived features appear only in the labeled support set and not in the query. The authors create 14 synthetic leaker types across 20 columns, evaluate a frozen relational encoder on RelBench databases, and use Integrated Gradients to identify and mitigate the most harmful leakers. Results show that target‑table leakers degrade performance most, while one‑ and two‑hop leakers are less consistently exploited, and that IG can partially recover performance by removing high‑ranked leakers.
By Roshan Reddy Upendra, Alexandre Dorais, Joe Meyer, Andrew Pouret, Anastasios Lambrianos Stappas, Dinesh Katupputhur Ramprasath, Tom Palczewski, Minghua Li
arXiv:2609.34879v2 Announce Type: replace
Abstract: Tool agents use large language models to act through external tools, yet successfully executed calls can still leave user requests unfulfilled. Too...
By Xiang Xia, Cheng Yan, Wuyang Zhang, Fan Xu, Zhijun Fan, Shuyuan Zhang, Yanyong Zhang
arXiv:2605.30327v2 Announce Type: replace-cross
Abstract: Frontier reasoning models are produced by post-training base language models with reinforcement learning. Recent work has challenged this by...
By Felix Zhou, Anay Mehrotra, Quanquan C. Liu
The paper introduces the concept of protocol divergence, showing that identical nominal missing rates can lead to vastly different learning regimes in incomplete multi‑view clustering. It critiques existing evaluation practices that ignore observation structure and proposes CRAFT, a train‑once framework that fuses observed views with mask‑aware attention, enabling efficient deployment across multiple missing‑view protocols. Experiments on CUB, MultiFashion, and other benchmarks demonstrate CRAFT’s superior performance and significant computational savings through checkpoint reuse.
By Haolu Liu, Xiyue Wang, Xuanting Xie, Liangjian Wen, Zhao Kang
arXiv:2609.35694v2 Announce Type: replace
Abstract: Continuous diffusion generates complete reasoning solutions through iterative refinement in latent space. We introduce the Continuous Embedding Dif...
By Xiang Cheng