KI-Forschungspapiere Täglich

Mask^2DiT: Dual Mask-based Diffusion Transformer for Multi-Scene Long Video Generation

Qwen2.5-Omni Technical Report

Dita: Scaling Diffusion Transformer for Generalist Vision-Language-Action Policy

Wan: Open and Advanced Large-Scale Video Generative Models

Open Deep Search: Democratizing Search with Open-source Reasoning Agents

Gemma 3 Technical Report

LEGO-Puzzles: How Good Are MLLMs at Multi-Step Spatial Reasoning?

Unconditional Priors Matter! Improving Conditional Generation of Fine-Tuned Diffusion Models

Gemini Robotics: Bringing AI into the Physical World

GenHancer: Imperfect Generative Models are Secretly Strong Vision-Centric Enhancers

BizGen: Advancing Article-level Visual Text Rendering for Infographics Generation

LogQuant: Log-Distributed 2-Bit Quantization of KV Cache with Superior Accuracy Preservation

MCTS-RAG: Enhancing Retrieval-Augmented Generation with Monte Carlo Tree Search

AccVideo: Accelerating Video Diffusion Model with Synthetic Dataset

Unlocking Efficient Long-to-Short LLM Reasoning with Model Merging

Attention IoU: Examining Biases in CelebA using Attention Maps

ADS-Edit: A Multimodal Knowledge Editing Dataset for Autonomous Driving Systems

ViLBench: A Suite for Vision-Language Process Reward Modeling

Image as an IMU: Estimating Camera Motion from a Single Motion-Blurred Image

Sparse Logit Sampling: Accelerating Knowledge Distillation in LLMs

Beyond Words: Advancing Long-Text Image Generation via Multimodal Autoregressive Models

DINeMo: Learning Neural Mesh Models with no 3D Annotations

Self-Supervised Learning of Motion Concepts by Optimizing Counterfactuals

RecTable: Fast Modeling Tabular Data with Rectified Flow

Trajectory Balance with Asynchrony: Decoupling Exploration and Learning for Fast, Scalable LLM Post-Training

UniHDSA: A Unified Relation Prediction Approach for Hierarchical Document Structure Analysis

PathoHR: Breast Cancer Survival Prediction on High-Resolution Pathological Images

RONA: Pragmatically Diverse Image Captioning with Coherence Relations