KI-Forschungspapiere Täglich

Paper2Code: Automating Code Generation from Scientific Papers in Machine Learning

Step1X-Edit: A Practical Framework for General Image Editing

RefVNLI: Towards Scalable Evaluation of Subject-driven Text-to-image Generation

Breaking the Modality Barrier: Universal Embedding Learning with Multimodal LLMs

Perspective-Aware Reasoning in Vision-Language Models via Mental Imagery Simulation

QuaDMix: Quality-Diversity Balanced Data Selection for Efficient LLM Pretraining

Token-Shuffle: Towards High-Resolution Image Generation with Autoregressive Models

DiMeR: Disentangled Mesh Reconstruction Model

Process Reward Models That Think

Boosting Generative Image Modeling via Joint Image-Feature Synthesis

DyMU: Dynamic Merging and Virtual Unmerging for Efficient VLMs

TimeChat-Online: 80% Visual Tokens are Naturally Redundant in Streaming Videos

3DV-TON: Textured 3D-Guided Consistent Video Try-on via Diffusion Models

IberBench: LLM Evaluation on Iberian Languages

ViSMaP: Unsupervised Hour-long Video Summarisation by Meta-Prompting

Distilling semantically aware orders for autoregressive image generation

Dynamic Camera Poses and Where to Find Them

Interpretable non-linear dimensionality reduction using gaussian weighted linear transformation