KI-Forschungspapiere Täglich

REINFORCE++: A Simple and Efficient Approach for Aligning Large Language Models

Cosmos World Foundation Model Platform for Physical AI

LLaVA-Mini: Efficient Image and Video Large Multimodal Models with One Vision Token

Sa2VA: Marrying SAM2 with LLaVA for Dense Grounded Understanding of Images and Videos

MotionBench: Benchmarking and Improving Fine-grained Video Motion Understanding for Vision Language Models

Diffusion as Shader: 3D-aware Video Diffusion for Versatile Video Generation Control

PPTAgent: Generating and Evaluating Presentations Beyond Text-to-Slides

OpenOmni: Large Language Models Pivot Zero-shot Omnimodal Alignment across Language with Real-time Self-Aware Emotional Speech Synthesis

Magic Mirror: ID-Preserved Video Generation in Video Diffusion Transformers

Dolphin: Closed-loop Open-ended Auto-research through Thinking, Practice, and Feedback

MoDec-GS: Global-to-Local Motion Decomposition and Temporal Interval Adjustment for Compact Dynamic 3D Gaussian Splatting

Segmenting Text and Learning Their Rewards for Improved RLHF in Language Model

Graph-Aware Isomorphic Attention for Adaptive Dynamics in Transformers

MagicFace: High-Fidelity Facial Expression Editing with Action-Unit Control

Generalizable Origin Identification for Text-Guided Image-to-Image Diffusion Models