MOOSEnger is a simulation‑aware AI agent framework designed for the MOOSE ecosystem, integrating an interchangeable reasoning model with domain knowledge, revised simulation artifacts, MOOSE‑specific validation, and executable solver feedback. Its generate‑check‑repair‑run workflow uses MOOSE knowledge retrieval, HIT‑aware parsing, syntax metadata, diagnostics, and revision‑controlled authoring to bind evidence to each input revision and guide bounded repair before acceptance. Across 200 prompts, MOOSEnger raises executable success from 5% to 89.5% with GPT‑5.2 and from 0% to 76.5% with Gemma 4 31B, and a ten‑case benchmark shows all generated inputs meet semantic alignment, with eight also meeting numerical‑accuracy criteria.
By Mengnan Li, Jason Miller, Zaid Abulawi, Zachary Prince, Matt Kohl, Jack M. Cavaluzzi, Guillaume Giudicelli, Casey T. Icenhour, Alexander Lindsay, Cody Permann
The paper introduces Band-Attention Modulation Network (BAM‑Net), a face forgery detection framework that learns fine‑grained, adaptive modulation of frequency bands in the Discrete Cosine Transform spectrogram. BAM‑Net dynamically reweights anti‑diagonal frequency bands to enhance forgery‑related spectral cues while suppressing irrelevant information, then fuses this modulated frequency data with spatial features using a lightweight backbone with distance‑decayed attention. Experiments on FaceForensics++, Celeb‑DF, and DFDC show that BAM‑Net achieves state‑of‑the‑art performance and strong generalization across datasets, compression levels, and manipulation types.
By Zhida Zhang, Wenkui Yang, Xinlei Ma, Qihang Fan, Jie Cao
The paper introduces a multimodal dataset for survival prediction in resected pancreatic ductal adenocarcinoma, comprising 302 patients, 446 H&E whole-slide images, clinicopathological variables, targeted sequencing data for 154 patients, and overall survival outcomes. The authors evaluated fourteen survival‑prediction models, finding that a Ridge Cox regression on numeric clinicopathological variables achieved the highest concordance (≈0.65), while multimodal fusion of image and molecular data reached 0.619. These benchmarks provide a foundation for future research and external validation using this pancreas‑specific dataset.
By Anh-Tien Nguyen, Mawuko Tettey, Jacqueline Michelle Metsch, Teresa Zimmer, Niklas Ullrich, Mario Duker, Sandra Rungeling, Kirsten Reuter-Jessen, Tessa Rosenthal, Lena-Christin Conradi, Michael Ghadimi, Alexander Konig, Elisabeth Hessmann, Volker Ellenrieder, Philipp Strobel, Hanibal Bohnenberger, Anne-Christin Hauschild
The paper demonstrates that language models can acquire new capabilities from post‑training data even when the training text is unrelated to the target task. Using a method called Active Taskless Distillation (ATD), the authors show that a single word from a teacher model can transfer knowledge to a student model without any target‑task examples or teacher logits. Experiments on Qwen2.5-1.5B reveal significant performance gains on HumanEval+ and improvements in scientific knowledge, commonsense reasoning, and reading comprehension across various model families.
By Ziyang Zhang, Yubin Jing, Yuanhao Zeng, Yuyao Li, Haofan Wang, Yichen Gong
The study evaluates large language model (LLM) graders on two computer‑science exams, testing 171 configurations of closed‑ and open‑weights models. While the best LLM configuration achieved a mean absolute error of 1.64/35—better than the 2.61/35 error between two human graders—its performance was highly sensitive to the prompt. A short "strict grader" preamble caused most open‑weight models to exceed acceptable error thresholds or stop grading entirely, whereas fine‑tuning with a single LoRA adapter restored parity with human graders and reduced sensitivity to harsh prompts.
By Ali Habibullah, Yazan Alshoibi, Mohammad Alshiekh, Salman Khan, Naeemullah Khan
The paper introduces WST-Graph, a topology-preserving wavelet scattering front-end designed for speech deepfake detection. It reconstructs wavelet scattering paths into a sparse modulation‑carrier grid that feeds an AASIST graph backend, employing modulation‑level normalization and adaptive local attention pooling to produce fixed relative‑time representations while keeping acoustic axes intact. The resulting waveform‑to‑graph interface uses about 60% fewer trainable parameters than AASIST yet remains competitive, showing clear improvements on selected out‑of‑domain benchmarks.
By Kwok-Ho Ng, Tingting Song, Bingwen Feng, Zhihua Xia
PROOF is a benchmark that profiles the reliability of object-level facts in instruction-tuned language models by converting a frozen Wikidata snapshot into 18,486 English multiple-choice questions grounded in 11,779 semantic facts across 101 classes, 392 properties, and 14 domains. Each question includes an explicit "I don't know" option, a "No correct option" control, and nine controlled formulations, with 1,849 questions designed as no-correct-option traps. The study evaluates 18 open-weight model deployments on 166,374 prompts, revealing wide variability in factual accuracy, sensitivity to wording changes, and the impact of decoder perturbations.
By Andrei Chetvergov, Mikhail Solovev, Timofei Sivoraksha, Stepan Ukolov, Valeriia Kuschenko, Alexander Evseev, Sergey Bolovtsov
The paper tackles syntactic parsing for Urdu, a morphologically rich language, and reports state‑of‑the‑art results for both constituency and dependency parsing. It introduces four key contributions: converting the CLE‑UTB phrase structure treebank into a dependency treebank with language‑specific mapping rules, a novel sequence labeling scheme that unifies the parsing task, training contextualized word representations on a 220‑million‑token Urdu corpus, and a parsing framework that employs both single‑task and multi‑task learning. Experiments show that the multi‑task setup boosts performance, achieving an F1 score of 91.39 for constituency parsing and a labeled attachment score of 85.69 for dependency parsing, improving over previous results.
By Toqeer Ehsan, Miriam Butt, Sarmad Hussain, Hassan Alhuzali, Ali Al-Laith
The paper evaluates how different design choices—document feeding strategy, retrieval method, and execution mode—affect Vision‑Language Models (VLMs) on long‑document question answering. Experiments on two benchmarks show that a multi‑tool agent only outperforms static input when the VLM is large, that retrieval modality (image vs text) is more critical than the specific retriever, and that combining the best pipelines per question can significantly boost performance. The study highlights the trade‑offs between token efficiency, model size, and pipeline complexity for deploying VLMs on complex documents.
By Kenan E. Ak, Jay Mohta, Gwang Gook Lee, Yan Xu, Dimitrios Dimitriadis
World Action Agent (WAA) is a multi‑agent framework that lets vision‑language models (VLMs) directly pilot robots by operating within a visual action workspace. The workspace provides automatically selected contact views, editable action rehearsals, and in‑view correction to refine decisions before low‑level execution. WAA learns procedural skills from expert videos and human teaching, and its interaction traces can train smaller VLMs, achieving state‑of‑the‑art success on LIBERO‑Pro and improving out‑of‑domain performance on robosuite and Qwen3.5‑9B.
By Yehang Zhang, Haojian Huang, Yifan Chang, Jianchong Su, Bohan Zhou, Yingjie Xu, Wosong Chen, Tianhao Zhou, Chenxu Wang, Tianyi Zhang, Yangkai Wei, Wenqian Li, Shiyuan Deng, Yinchuan Li, Ying-Cong Chen, Zexi Li
PHOSA introduces MVSign, the first multi‑view Chinese sign language dataset co‑designed with Deaf experts, featuring diverse gestures and rich annotations. The authors develop a hybrid fitting pipeline for accurate SMPL‑X annotation and propose a decoupled sign avatar representation that isolates body, head, and hand components, coupled with a motion‑aware sampling strategy to handle motion blur and balance gesture diversity. Experiments show high‑fidelity visual results on MVSign, especially in detailed hand and facial regions, and good generalization to in‑the‑wild monocular sign language videos.
By Haodong Wang, Hezhen Hu, Wengang Zhou, Houqiang Li
The paper introduces FlipDir, a training‑free inference‑time technique that mitigates answer flips in vision‑language models by steering hidden states along a low‑rank subspace derived from contrastive image pairs. It employs a margin‑based gate to attenuate steering only during uncertain decoding steps, thereby restoring original predictions while keeping stable ones unchanged. The authors also present VisFlip, a benchmark framework that evaluates models across nine dataset‑variation combinations in scientific reasoning, robot‑scene understanding, and medical VQA, showing that FlipDir consistently outperforms existing methods on recovery and preservation metrics.
By Yeonsung Jung, Joonhyun Jeong, Hoang Pham, Joowon Kim, Yoonsik Park, Viet Dac Lai, Eunho Yang
The paper presents a retrieval‑augmented generation pipeline for answering regulatory compliance questions in finance. It builds a three‑stage retriever on LegalBERT and a compact 2B–12B generator served with 4‑bit quantization, achieving a Recall@10 of 0.774 on the ObliQA benchmark and improving answer quality via RAFT‑LoRA fine‑tuning. However, the adapted models fail to transfer to Australian case‑law questions, and a closed‑book model performs almost as well while lacking verifiable grounding.
By Tobias Deu{\ss}er, Abhishek Pillai, Aurelio F. Bariviera, Dhananjay Bhardwaj, Lorenz Sparrenberg, David Berghaus, Christian Bauckhage, Rafet Sifa
The paper investigates why reconstruction-based unsupervised learning can fail, identifying two failure modes: over‑reconstruction of anomalies and loss of nominal variation. Using the Pursuit of Subspaces hypothesis, it links these failures to geometric properties—join blindness from excess range and meet preference from insufficient capacity—and shows that a compact nominal union is optimal, typically requiring a nonlinear reconstruction map. The authors propose Dynamic Push and Pull, along with nested manifold carving, to learn compact representations without anomaly labels, and demonstrate improved anomaly detection on standard benchmarks, unseen image degradations, and ECG classification.
By Mehmet Yama\c{c}, Yagmur Mustu, Muhammad Numan Yousaf, Lei Xu, Marcel van Gerven
The paper introduces VeriSpeak, a benchmark of 3,879 spoken claims for evaluating fact verification in Large Audio Language Models (LALMs). It shows a clear modality gap: models that verify written claims well often fail on spoken versions, and retrieval alone offers limited improvement. Combining retrieval with explicit reasoning yields the best performance, reaching 86.1% accuracy and demonstrating the need for grounded reasoning over retrieved evidence in speech misinformation detection.
By Debajyoti Mazumder, Mamta, Abhirama Subramanyam Penamakuri
The paper proposes a method for controlling downside risk when adjusting forecasts from frozen models, such as foundation models, by combining a static corrector and an online corrector on the simplex. Using only post‑horizon losses, the approach achieves minimal deterioration (0.15%) and up to 11.5% gains across 28 forecast pairs, and consistently reduces mean MSE in day‑ahead load forecasts for seven European bidding zones. The method’s applicability is bounded by three empirical conditions related to expert speed, stream length, and outcome alignment.
By Minkyoung Kim, Hyunjung Byun, Yohan Lee, Beakcheol Jang
IatroBench is a pre‑registered benchmark that evaluates language models on clinical omission and commission harms across 60 scenarios and six models. Using a physician‑written rubric scored by Claude Opus 4.6, the study finds that models tend to withhold more information from patients than from doctors—a phenomenon termed framing‑contingent withholding—while also revealing varied patterns of omission across different models. The benchmark highlights how framing influences the amount of medical information shared by AI systems.
By David Gringras
arXiv:2503. 02781v3 Announce Type: replace-cross Abstract: Predicting clinical outcomes from preclinical data is essential for selecting safe and effective drug combinations and for reducing late-stage failures.
By Yepeng Huang, Xiaorui Su, Varun Ullanat, Intae Moon, Ivy Liang, Lindsay Clegg, Damilola Olabode, Ruthie Johnson, Nicholas Ho, Megan Gibbs, Alexander Gusev, Bino John, Marinka Zitnik
IDRBench is a benchmark designed to evaluate the interactive capabilities of deep research agents that use large language models. It introduces controlled opportunities for clarification within a common workflow, comparing autonomous and interactive trajectories by measuring task‑specific report alignment and interaction cost. Experiments on 100 tasks with seven LLMs show that interaction consistently improves alignment, though its effectiveness varies depending on the agents’ questions and feedback integration.
By Yingchaojie Feng, Qiang Huang, Xiaoya Xie, Zhaorui Yang, Jun Yu, Wei Chen, Anthony K. H. Tung
LiveMathematicianBench is a dynamic multiple‑choice benchmark for research‑level mathematical reasoning, built from recent arXiv papers published after model training cutoffs. It introduces a thirteen‑category logical taxonomy of theorem types and uses a proof‑sketch‑guided distractor pipeline to create plausible but invalid answer choices, enhancing sensitivity to genuine reasoning. Evaluation shows current large language models perform poorly, with the best model scoring 43.5% overall and only 17.6% under substitution‑resistant conditions, indicating the benchmark’s difficulty and realism.
By Linyang He, Qiyao Yu, Hanze Dong, Baohao Liao, Xinxing Xu, Micah Goldblum, Jiang Bian, Nima Mesgarani