KI-Forschungspapiere Täglich

Not All Language Model Features Are Linear

DeepSeek-Prover: Advancing Theorem Proving in LLMs through Large-Scale Synthetic Data

ReVideo: Remake a Video with Motion and Content Control

Dense Connector for MLLMs

LiteVAE: Lightweight and Efficient Variational Autoencoders for Latent Diffusion Models

Distributed Speculative Inference of Large Language Models

DiM: Diffusion Mamba for Efficient High-Resolution Image Synthesis

Thermodynamic Natural Gradient Descent

Improved Distribution Matching Distillation for Fast Image Synthesis

Visual Echoes: A Simple Unified Transformer for Audio-Visual Generation

AlignGPT: Multi-modal Large Language Models with Adaptive Alignment Capability

RectifID: Personalizing Rectified Flow with Anchored Classifier Guidance

CamViG: Camera Aware Image-to-Video Generation with Multimodal Transformers

Semantica: An Adaptable Image-Conditioned Diffusion Model

NeRF-Casting: Improved View-Dependent Appearance with Consistent Reflections

Neural Directional Encoding for Efficient and Accurate View-Dependent Appearance Modeling

Tele-Aloha: A Low-budget and High-authenticity Telepresence System Using Sparse RGB Cameras