| 2026-11-20 | 2611.25179 | Phantom Success, Silent Failure: Auditing the Success Checkers of Simulated Manipulation Benchmarks | Zirui Song et al. | arXiv | 3 | |
| 2026-11-18 | 2611.21644 | MInfoVisQA: A Multilingual Benchmark for Information-Rich Visual Question Answering | Changjiang Jiang et al. | arXiv | 1 | |
| 2026-10-23 | 2610.22951 | MambaAct: Efficient Mamba Action Experts for Flow-Matching Vision-Language-Action Models | Khuyen Pham et al. | arXiv | 14 | |
| 2026-10-19 | 2610.22029 | Selection Is Not Learning: Choosing Model Futures | Haoyi Yang et al. | arXiv | 4 | |
| 2026-10-16 | 2610.21439 | Each Reward to Its Relevant Tokens: Reward-Specific Token-Level Credit Assignment for Multi-Reward RL | Skylar Zhai et al. | arXiv | 36 | |
| 2026-10-09 | 2610.12459 | WorldGuide: Goal-Directed Video World Model for Procedural Task Execution | Ankan Deria et al. | arXiv | 18 | |
| 2026-10-07 | 2610.08773 | AdvSim2Real : Training Web Agents Against Adaptive Prompt Injection in a Web World Model | Sarim Hashmi et al. | arXiv | 16 | |
| 2026-10-06 | 2610.05894 | Noise Out, Bias In: Targeted Bias Injection in Diffusion Language Models via Closed-Loop Activation Steering | Sarim Hashmi et al. | arXiv | 14 | |
| 2026-10-06 | 2610.05870 | Certification of Real Images through Calibrated Content Authentication | Sarim Hashmi et al. | arXiv | 14 | |
| 2026-10-05 | 2610.03574 | HyperBrowseComp: A Multilingual and Multimodal Stress Test for Web-Browsing Agents | Alham Fikri Aji et al. | arXiv | 25 | |
| 2026-10-02 | 2610.01092 | Ego2Act: Evaluating Goal-Directed Manipulation in Egocentric Video Generation | Patrick Amadeus Irawan et al. | arXiv | 24 | |
| 2026-10-02 | 2610.02148 | Omni-Embed-Mini: Binding Modalities Without Forgetting via Dense Distillation | Mohammed Irfan Kurpath et al. | arXiv | 14 | |
| 2026-09-29 | 2609.35718 | Hard Vision, Easy Vision: What GPT-6 Astra Reveals Across Computer Vision | Hanoona Rasheed et al. | arXiv | 23 | |
| 2026-09-29 | 2609.33855 | Program-Verified Self-Evolution for Vision-Language Models | Ahmed Heakl et al. | arXiv | 10 | |
| 2026-09-23 | 2609.26796 | Flash-dLLM: IO-Aware KV Caching and Parallel Decoding for Fast, Memory-Efficient Diffusion LLMs | Quan Nguyen-Tri et al. | arXiv | 31 | |
| 2026-09-21 | 2609.21094 | Geometry of Values: Task Vector Composition for Ethical Preference Alignment in Language Models | Utkarsh Agarwal et al. | arXiv | 5 | |
| 2026-09-07 | 2609.04242 | Training-Free Speech-Centric Omni Understanding with Frozen VLMs | Ankan Deria et al. | arXiv | 2 | |
| 2026-08-31 | 2608.28192 | Locate Anything in Videos: Rethinking Efficient Generative Spatio-Temporal Video Grounding | Hanoona Rasheed et al. | arXiv | 17 | |
| 2026-07-10 | 2607.03936 | Can Dialects Be Steered Like Languages? Sparse Neurons and Distributed Directions in Arabic LLMs | Kareem Elozeiri et al. | arXiv | 2 | |
| 2026-06-30 | 2606.28525 | A Gravitational Interpretation of Fine-Tuning Reversion | Samuele Poppi et al. | arXiv | 2 | |
| 2026-05-20 | 2605.19436 | CEPO: RLVR Self-Distillation using Contrastive Evidence Policy Optimization | Ahmed Heakl et al. | arXiv | 10 | |
| 2026-05-20 | 2605.12623 | DocAtlas: Multilingual Document Understanding Across 80+ Languages | Ahmed Heakl et al. | arXiv | 4 | |
| 2026-05-19 | 2605.18719 | SafeDiffusion-R1: Online Reward Steering for Safe Diffusion Post-Training | Komal Kumar et al. | arXiv | 3 | |
| 2026-05-18 | 2605.15592 | Efficient Image Synthesis with Sphere Latent Encoder | Tung Do et al. | arXiv | 6 | |
| 2026-05-13 | 2605.07654 | Reliable Chain-of-Thought via Prefix Consistency | Naoto Iwase et al. | arXiv | 1 | |
| 2026-05-12 | 2605.10468 | Can Muon Fine-tune Adam-Pretrained Models? | Xingyu Qu et al. | arXiv | 4 | |
| 2026-05-01 | 2604.27766 | Instruction-Guided Poetry Generation in Arabic and Its Dialects | Abdelrahman Sadallah et al. | arXiv | 3 | |