Constrained Dynamic Gaussian Splatting (CDGS) addresses the memory challenges of Dynamic Gaussian Splatting by framing reconstruction as a budget‑constrained optimization problem. It introduces a differentiable budget controller that fuses geometric, motion, and perceptual cues to enforce a strict, user‑defined Gaussian budget during training. CDGS further splits the representation into static and dynamic branches, allocates capacity based on motion complexity, and employs a three‑phase training strategy and hybrid compression to achieve high‑quality rendering with over 3× model compression compared to Ex4DGS.
By Zihan Zheng, Zhenlong Wu, Xuanxuan Wang, Houqiang Zhong, Xiaoyun Zhang, Qiang Hu, Guangtao Zhai, Wenjun Zhang
The paper introduces DORS, a diffusion-based framework for removing objects from dense scenes. It employs Dynamic Attention Routing, combining Instance-Filtered Attention to block misleading semantic cues from similar objects and Context-Guided Routing to preserve visual consistency. A new benchmark, DOR-Bench, is also presented, and experiments show DORS outperforms existing methods in reducing incomplete removal and duplicate artifacts.
By Haitong Tang, Haipeng Liu, Yang Wang
C-LoRA introduces a continual learning approach for pre‑trained visual models that uses a single shared Low‑Rank Adaptation (LoRA) adapter. It employs a learnable routing matrix split into a stability component (R_base) and a plasticity component (R_delta) to balance preserving past knowledge and adapting to new tasks. This design eliminates the need for expanding modules or complex inference fusion while maintaining competitive performance on multiple benchmarks.
By Xin Zhang, Liang Bai, Xian Yang
The paper introduces PEBench, a unified benchmark for evaluating prompt enhancers (PEs) in video generation across text-to-video, image-to-video, and reference-to-video tasks. It contains 1,100 expert-verified cases, 1,005 visual assets, and 35 fine-grained tasks, and employs a fact-extraction and rubric-based assessment framework covering 24 criteria. Evaluation of various PE methods shows a trend toward intent-preserving cinematic planning, with certain methods excelling in coverage or coherence, and human validation confirms that PEBench scores align with expert judgments and downstream video quality.
By Yawen Shao, Yubo Zhu, Ziyun Dai, Zixun Fang, Kai Zhu, Zeyinzi Jiang, Yufeng Ai, Siyang Sun, Haolan Xue, Yu Shang, Yuxiang Bao, Zoubin Bi, Jingming Luo, Jie Xiao, Chaojie Mao, Zhehan Kan, Hongchen Luo, Yu Liu, Sheng Zhong, Wei Tong, Xueyang Fu, Yang Cao, Wei Zhai, Zheng-Jun Zha
The paper introduces OcuBench, a comprehensive benchmark for eyeglass reflection removal that includes 10,280 synthetic pairs, 732 real-input pseudo-pairs, and 458 real-world test images, enabling both paired evaluation and assessment beyond generated supervision. It also proposes OcuFlow, an ocular-adaptive pixel MeanFlow framework that uses geometry-adaptive representation and one-step pMF to focus on reflection-obscured ocular regions while preserving native-resolution details. Experiments show OcuFlow consistently outperforms baselines in reflection removal quality, ocular fidelity, and efficiency, achieving 67.32% of selections in a blind user study, six times the next-best share.
By Tao Liu, Youwei Pang, Kailai Zhou, Jiaming Zuo, Hanqi Liu, Wei Ji, Peng-Tao Jiang, Xiaofeng Liu, Weisi Lin, Xiaoqi Zhao
Opera is a verbal critic framework designed for long‑horizon coding agents, treating each correction as a persistent note that is followed until the problem is resolved. It uses periodic and event‑driven triggers to diagnose issues, audits feedback against visible evidence, and tracks subsequent actions to distinguish compliance from true resolution. In tests, Opera improves the resolve rate of non‑critic agents by up to 12.4, 15.0, and 8.9 percentage points on three benchmarks and provides on‑policy training data that boosts a model’s performance without requiring a critic at inference time.
By Kai Mei, Zhiyuan Hu, Yutong Dai, Juntao Tan, Yifan Zhang, Dingjie Song, Dimitris N. Metaxas, Silvio Savarese, Ran Xu, Zeyuan Chen
FA-Bench is an open framework that standardizes forced alignment evaluation by fixing transcript normalization, data splits, phone mapping, and scoring procedures. It offers two tracks: one with reference transcripts and another with recognizer outputs, evaluated on clean and four degraded audio conditions using 21 open models and 9 commercial APIs. The benchmark scores every utterance boundary with a tolerance‑based F1 metric, revealing systematic timing biases such as Whisper’s 150 ms early alignment and commercial APIs’ 50 ms late alignment.
By Wei Chu, Yuanzhe Dong, Ke Tan, Dong Han, Yichao Zhou, Ruchao Fan, Bingshen Mu, Jingbei Li, Vishwas Shetty, Sarthak Bisht, Ziyue Qiu, Massa Baali, Rita Singh, Bhisha Raj
The paper introduces the Latent Core Tokenizer (LCT), a language‑agnostic method that first discovers reusable linguistic units using Minimum Description Length, entropy‑based boundary signals, and morphotactic constraints before building a shared vocabulary. With a 200K‑token vocabulary across 104 languages, LCT shows lower fertility and higher MorphScore than BPE, Unigram, and parity‑aware BPE, while keeping tokenization cost comparable across languages. On four multilingual downstream benchmarks, LCT outperforms the baselines by 1.48, 1.83, and 2.00 aggregate points, demonstrating that compression alone does not guarantee representation quality and underscoring the role of morphology‑driven structural discovery.
By Felermino D. M. A. Ali, Millicent Ochieng, Ogbemi Ekwejunor-Etchie, Ade Famoti, Jacki O'Neill, Debjit Paul
MiniVer‑V is a new benchmark for short‑video fact‑checking that contains 195 videos, each annotated with a supported, refuted, or insufficient verdict and 5,510 multimodal evidence units (visual keyframes, speech transcripts, and web‑retrieved sources). The authors propose a two‑layer verification framework that first checks claim‑video consistency using internal evidence, then determines the factual verdict with external corroboration, and employs a sufficiency‑driven greedy search to assemble only the minimal evidence needed for a confident decision. Using Claude Sonnet 4, the method achieves a Macro‑F1 of 0.510 with an average of 4.5 evidence units—about 16 % of the full set—while matching the full‑evidence baseline and improving detection of insufficient cases.
By Leran Chen, Lingnan Kong, Zile Cai
The paper introduces SAGE, a lightweight method that guides decoder attention in vision-language models away from prompt‑invariant attention sinks (PIS) toward query‑dependent regions of interest (ROIs). By using token‑aligned ROI masks from vision backbones like CLIP, ViT, and DINOv3, SAGE improves visual grounding and reduces hallucinations across various VLM families. Experiments show consistent gains on public vision‑language benchmarks, especially in fine‑grained visual discrimination tasks where localized evidence is essential.
By Jeonghyo Song, YoungJoon Yoo
SkillGraph introduces a skill library framework that models reusable skills as nodes in a directed graph, with typed edges representing prerequisite, enhancement, and co-occurrence relationships. When presented with a new task, the system retrieves an ordered subgraph of relevant skills, guiding multi-step decision making. The graph is continuously refined through agent trajectories and reinforcement learning, enabling simultaneous improvement of the skill library and the agent policy, and achieving state‑of‑the‑art results on ALFWorld, WebShop, and several search‑augmented QA tasks.
By Xiaoyuan Li, Moxin Li, Keqin Bao, Yubo Ma, Wenjie Wang, Dayiheng Liu, Fuli Feng
CiteVQA is a benchmark that evaluates multimodal large language models on both answering document questions and providing element‑level bounding‑box citations for the evidence used. It contains 1,897 questions across 711 PDFs in seven domains and two languages, with ground‑truth citations generated via an automated pipeline and strict quality control. The evaluation metric, Strict Attributed Accuracy (SAA), rewards a prediction only when both the answer and the cited region are correct, revealing widespread attribution hallucination in current models.
By Dongsheng Ma, Jiayu Li, Zhengren Wang, Yijie Wang, Jiahao Kong, Weijun Zeng, Jutao Xiao, Jie Yang, Bangrui Xu, Yuhan Wang, Bin Wang, Conghui He
SERL‑SQL introduces a selective execution‑grounded reinforcement learning framework for multi‑turn Text‑to‑SQL agents. It samples on‑policy interaction trajectories and employs a training‑only teacher to re‑score student actions using execution feedback, converting the teacher‑student likelihood gap into bounded, masked weights that reweight GRPO advantages only on SQL and tool‑action tokens. Experiments on BIRD, Spider, and cross‑domain benchmarks show competitive performance, achieving 76.56% execution accuracy on BIRD‑Dev and 89.92% on Spider‑Test, and the reward‑based selection strategy approaches the oracle Best‑of‑N upper bound.
By Tao Liu, Tao Feng, Xiangheng Li, Jinwang Song, Yifan Li, Xiaoqing Cheng, Dixuan Zhang, Siquan Li, Lin Lan, Hongying Zan, Kunli Zhang, Chao Wu
Fluid-Gen-Zero is a training-free framework that generates physics-aware fluid-object interaction videos by separating physical reasoning from appearance synthesis. It uses a physics simulator for motion dynamics and a pretrained video generator for visual quality, coordinated through a two-level agentic workflow involving a vision-language model for planning and latent-space guidance for simulation signals. The method improves simulation alignment and fluid flow accuracy across multiple backbones while maintaining perceptual quality, as shown by quantitative metrics and human preference studies.
By Hong Huang, Yuqiu Liu, Chenyu You, Daniel Martin, Chuhang Zou, Wuyang Chen
The paper introduces a new paradigm for multimodal emotion understanding that moves beyond surface-level cue-label associations by incorporating cognitive appraisal reasoning. It presents CogEmo-40K, a large instruction‑tuning dataset that elicits evidence‑grounded reasoning across six appraisal dimensions; CogEmo‑MoE, a compact sparse multimodal large language model with interleaved MoE blocks for appraisal‑specific adaptation; and CogEmo‑Bench, a benchmark featuring an Appraisal Evidence Quality Score (AEQS) to evaluate why emotions arise. Experiments demonstrate that this perception‑to‑appraisal approach improves cross‑domain generalization and aligns model reasoning more closely with human cognitive processes.
By Jia Li, Yichao He, Yangchen Yu, Qiankun Li, Xinyi Li, Baiyi Ye, Zhenzhen Hu, Richang Hong, Erik Cambria
The paper introduces Entropy‑Normalized Trust Region (ENTR), a method for asynchronous reinforcement learning that adjusts off‑policy ratio thresholds based on token entropy rather than a single magnitude cut‑off. By recognizing that the natural scale of the ratio is set by entropy, ENTR preserves genuine exploration while filtering out noise from low‑entropy, stale data. Experiments on long‑horizon agentic tasks and mathematical reasoning benchmarks show ENTR outperforms existing asynchronous methods, improving BrowseComp‑Plus performance by 6.9 % and enabling stable training up to 30 policy versions of staleness while matching synchronous GRPO at a 2.6× speedup.
By Guanqun Zhao, Zijun Xie, Binbin Zheng, Yehan Yang, Jiafeng Lu, Aoqi Hu, Enlei Gong, Zeyu Chen
The paper investigates whether large language models (LLMs) can automatically verify no-code bug fixes—solutions that instruct users to change settings, upgrade versions, or alter workflows—by executing them in a real browser environment. Using 322 fixes from a benchmark, the study tests three executor agents (OpenCUA‑72B, Claude Sonnet 5, and Meta’s Muse Glimmer) and finds that only 17.6% of issues could be set up and passed sanity checks, with resolution rates ranging from 14.6% to 49.7% depending on the executor. Even the best configuration resolved fewer than half of the fixes, and executor choice significantly affected outcomes, highlighting the need for execution-based verification before such fixes reach users.
By Utku Boran Torun, Veli Karakaya, Eray T\"uz\"un
DataVista is a new benchmark for data video understanding, comprising 961 real‑world videos and 6,775 evaluation questions. It is organized into a three‑level capability framework—data perception, temporal reasoning, and narrative understanding—with 10 fine‑grained question types across five topic domains. Evaluation of 19 mainstream multimodal large language models shows that the best model, Gemini‑3.1‑Pro, scores 70.0% overall accuracy, lagging far behind human experts, especially on causal reasoning and narrative structure.
By Yupeng Xie, Zhenyang Wang, Jiayi Zhu, Yinghao Tang, Zhouan Shen, Yiyu Chen, Yuyu Luo
The paper proposes a method that allows agents in multi‑agent motion planning to compute with multiple prioritizations simultaneously, avoiding reliance on domain‑specific heuristics or costly iterative searches. This general approach is applied to receding‑horizon motion planning with system dynamics, and numerical experiments show near‑optimal prioritization and performance gains over state‑of‑the‑art methods with only a slight increase in computation time. Real‑time feasibility is demonstrated on a road network with ten vehicles in a Cyber‑Physical Mobility Lab.
By Patrick Scheffe, Julius Kahle, Bassam Alrifaee
arXiv:2509. 16525v2 Announce Type: replace-cross Abstract: Machine learning models are increasingly deployed as software components that must evolve as requirements change.
By Anna Mazhar, Sainyam Galhotra