KI-Forschungspapiere Täglich

SeedVR2: One-Step Video Restoration via Diffusion Adversarial Post-Training

ComfyUI-Copilot: An Intelligent Assistant for Automated Workflow Development

Video World Models with Long-term Spatial Memory

RoboRefer: Towards Spatial Referring with Reasoning in Vision-Language Models for Robotics

Diagonal Batching Unlocks Parallelism in Recurrent Memory Transformers for Long Contexts

Qwen3 Embedding: Advancing Text Embedding and Reranking Through Foundation Models

Surfer-H Meets Holo1: Cost-Efficient Web Agent Powered by Open Weights

The Common Pile v0.1: An 8TB Dataset of Public Domain and Openly Licensed Text

VideoREPA: Learning Physics for Video Generation through Relational Alignment with Foundation Models

Aligning Latent Spaces with Flow Priors

VideoMathQA: Benchmarking Mathematical Reasoning via Multimodal Understanding in Videos

AV-Reasoner: Improving and Benchmarking Clue-Grounded Audio-Visual Counting for MLLMs

Inference-Time Hyper-Scaling with KV Cache Compression

Unfolding Spatial Cognition: Evaluating Multimodal Models on Visual Simulations

SparseMM: Head Sparsity Emerges from Visual Concept Responses in MLLMs

StreamBP: Memory-Efficient Exact Backpropagation for Long Sequence Training of LLMs

EOC-Bench: Can MLLMs Identify, Recall, and Forecast Objects in an Egocentric World?

Search Arena: Analyzing Search-Augmented LLMs

MINT-CoT: Enabling Interleaved Visual Tokens in Mathematical Chain-of-Thought Reasoning

Revisiting Depth Representations for Feed-Forward 3D Gaussian Splatting

Evaluation is All You Need: Strategic Overclaiming of LLM Reasoning Capabilities Through Evaluation Design

FlexPainter: Flexible and Multi-View Consistent Texture Generation

Language-Image Alignment with Fixed Text Encoders

Autoregressive Images Watermarking through Lexical Biasing: An Approach Resistant to Regeneration Attack

FreeTimeGS: Free Gaussians at Anytime and Anywhere for Dynamic Scene Reconstruction

SkyReels-Audio: Omni Audio-Conditioned Talking Portraits in Video Diffusion Transformers

Geometry-Editable and Appearance-Preserving Object Compositon

Kinetics: Rethinking Test-Time Scaling Laws

Scaling Laws for Robust Comparison of Open Foundation Language-Vision Models and Datasets

MedAgentGym: Training LLM Agents for Code-Based Medical Reasoning at Scale

Contextual Integrity in LLMs via Reasoning and Reinforcement Learning

Rectified Point Flow: Generic Point Cloud Pose Estimation

Micro-Act: Mitigate Knowledge Conflict in Question Answering via Actionable Self-Reasoning

FEAT: Full-Dimensional Efficient Attention Transformer for Medical Video Generation

Images are Worth Variable Length of Representations

RobustSplat: Decoupling Densification and Dynamics for Transient-Free 3DGS

MARBLE: Material Recomposition and Blending in CLIP-Space

FlowDirector: Training-Free Flow Steering for Precise Text-to-Video Editing

Perceptual Decoupling for Scalable Multi-modal Reasoning via Reward-Optimized Captioning

BEVCALIB: LiDAR-Camera Calibration via Geometry-Guided Bird's-Eye View Representations

Diffusion-Based Generative Models for 3D Occupancy Prediction in Autonomous Driving

PATS: Proficiency-Aware Temporal Sampling for Multi-View Sports Skill Assessment

Watermarking Degrades Alignment in Language Models: Analysis and Mitigation

Rethinking Whole-Body CT Image Interpretation: An Abnormality-Centric Approach

SViMo: Synchronized Diffusion for Video and Motion Generation in Hand-object Interaction Scenarios

What do self-supervised speech models know about Dutch? Analyzing advantages of language-specific pre-training