Cheers: Decoupling Patch Details from Semantic Representations Enables Unified Multimodal Comprehension and Generation Paper • 2603.12793 • Published 14 days ago • 37
HybridStitch: Pixel and Timestep Level Model Stitching for Diffusion Acceleration Paper • 2603.07815 • Published 19 days ago • 10
Can Vision-Language Models Solve the Shell Game? Paper • 2603.08436 • Published 18 days ago • 39
From Sparse to Dense: Multi-View GRPO for Flow Models via Augmented Condition Space Paper • 2603.12648 • Published 15 days ago • 14
Video Streaming Thinking: VideoLLMs Can Watch and Think Simultaneously Paper • 2603.12262 • Published 15 days ago • 30
MM-CondChain: A Programmatically Verified Benchmark for Visually Grounded Deep Compositional Reasoning Paper • 2603.12266 • Published 15 days ago • 19
V-Bridge: Bridging Video Generative Priors to Versatile Few-shot Image Restoration Paper • 2603.13089 • Published 14 days ago • 13
Visual-ERM: Reward Modeling for Visual Equivalence Paper • 2603.13224 • Published 14 days ago • 21
OmniForcing: Unleashing Real-time Joint Audio-Visual Generation Paper • 2603.11647 • Published 15 days ago • 31
Understanding by Reconstruction: Reversing the Software Development Process for LLM Pretraining Paper • 2603.11103 • Published 16 days ago • 8 • 2
BeyondSWE: Can Current Code Agent Survive Beyond Single-Repo Bug Fixing? Paper • 2603.03194 • Published 24 days ago • 56