Mohamed Bin Zayed University of Artificial Intelligence · 27 records

dateidtitleauthorsvenue↑
2026-11-202611.25179Phantom Success, Silent Failure: Auditing the Success Checkers of Simulated Manipulation BenchmarksZirui Song et al.arXiv3
2026-11-182611.21644MInfoVisQA: A Multilingual Benchmark for Information-Rich Visual Question AnsweringChangjiang Jiang et al.arXiv1
2026-10-232610.22951MambaAct: Efficient Mamba Action Experts for Flow-Matching Vision-Language-Action ModelsKhuyen Pham et al.arXiv14
2026-10-192610.22029Selection Is Not Learning: Choosing Model FuturesHaoyi Yang et al.arXiv4
2026-10-162610.21439Each Reward to Its Relevant Tokens: Reward-Specific Token-Level Credit Assignment for Multi-Reward RLSkylar Zhai et al.arXiv36
2026-10-092610.12459WorldGuide: Goal-Directed Video World Model for Procedural Task ExecutionAnkan Deria et al.arXiv18
2026-10-072610.08773AdvSim2Real : Training Web Agents Against Adaptive Prompt Injection in a Web World ModelSarim Hashmi et al.arXiv16
2026-10-062610.05894Noise Out, Bias In: Targeted Bias Injection in Diffusion Language Models via Closed-Loop Activation SteeringSarim Hashmi et al.arXiv14
2026-10-062610.05870Certification of Real Images through Calibrated Content AuthenticationSarim Hashmi et al.arXiv14
2026-10-052610.03574HyperBrowseComp: A Multilingual and Multimodal Stress Test for Web-Browsing AgentsAlham Fikri Aji et al.arXiv25
2026-10-022610.01092Ego2Act: Evaluating Goal-Directed Manipulation in Egocentric Video GenerationPatrick Amadeus Irawan et al.arXiv24
2026-10-022610.02148Omni-Embed-Mini: Binding Modalities Without Forgetting via Dense DistillationMohammed Irfan Kurpath et al.arXiv14
2026-09-292609.35718Hard Vision, Easy Vision: What GPT-6 Astra Reveals Across Computer VisionHanoona Rasheed et al.arXiv23
2026-09-292609.33855Program-Verified Self-Evolution for Vision-Language ModelsAhmed Heakl et al.arXiv10
2026-09-232609.26796Flash-dLLM: IO-Aware KV Caching and Parallel Decoding for Fast, Memory-Efficient Diffusion LLMsQuan Nguyen-Tri et al.arXiv31
2026-09-212609.21094Geometry of Values: Task Vector Composition for Ethical Preference Alignment in Language ModelsUtkarsh Agarwal et al.arXiv5
2026-09-072609.04242Training-Free Speech-Centric Omni Understanding with Frozen VLMsAnkan Deria et al.arXiv2
2026-08-312608.28192Locate Anything in Videos: Rethinking Efficient Generative Spatio-Temporal Video GroundingHanoona Rasheed et al.arXiv17
2026-07-102607.03936Can Dialects Be Steered Like Languages? Sparse Neurons and Distributed Directions in Arabic LLMsKareem Elozeiri et al.arXiv2
2026-06-302606.28525A Gravitational Interpretation of Fine-Tuning ReversionSamuele Poppi et al.arXiv2
2026-05-202605.19436CEPO: RLVR Self-Distillation using Contrastive Evidence Policy OptimizationAhmed Heakl et al.arXiv10
2026-05-202605.12623DocAtlas: Multilingual Document Understanding Across 80+ LanguagesAhmed Heakl et al.arXiv4
2026-05-192605.18719SafeDiffusion-R1: Online Reward Steering for Safe Diffusion Post-TrainingKomal Kumar et al.arXiv3
2026-05-182605.15592Efficient Image Synthesis with Sphere Latent EncoderTung Do et al.arXiv6
2026-05-132605.07654Reliable Chain-of-Thought via Prefix ConsistencyNaoto Iwase et al.arXiv1
2026-05-122605.10468Can Muon Fine-tune Adam-Pretrained Models?Xingyu Qu et al.arXiv4
2026-05-012604.27766Instruction-Guided Poetry Generation in Arabic and Its DialectsAbdelrahman Sadallah et al.arXiv3