Daily Papers

One-Step Residual Shifting Diffusion for Image Super-Resolution via Distillation

Survey on Evaluation of LLM-based Agents

Stop Overthinking: A Survey on Efficient Reasoning for Large Language Models

Unleashing Vecset Diffusion Model for Fast Shape Generation

Cosmos-Reason1: From Physical Common Sense To Embodied Reasoning

Scale-wise Distillation of Diffusion Models

JARVIS-VLA: Post-Training Large-Scale Vision Language Models to Play Visual Games with Keyboards and Mouse

Reinforcement Learning for Reasoning in Small LLMs: What Works and What Doesn't

DiffMoE: Dynamic Token Selection for Scalable Diffusion Transformers

InfiniteYou: Flexible Photo Recrafting While Preserving Your Identity

Why Do Multi-Agent LLM Systems Fail?

Inside-Out: Hidden Factual Knowledge in LLMs

Plug-and-Play 1.x-Bit KV Cache Quantization for Video Large Language Models

Fin-R1: A Large Language Model for Financial Reasoning through Reinforcement Learning

MathFusion: Enhancing Mathematic Problem-solving of LLM through Instruction Fusion

LHM: Large Animatable Human Reconstruction Model from a Single Image in Seconds

SynCity: Training-Free Generation of 3D Worlds

M3: 3D-Spatial MultiModal Memory

1000+ FPS 4D Gaussian Splatting for Dynamic Scene Rendering

CaKE: Circuit-aware Editing Enables Generalizable Knowledge Learners

Ultra-Resolution Adaptation with Ease

Expert Race: A Flexible Routing Strategy for Scaling Diffusion Transformer with Mixture of Experts

XAttention: Block Sparse Attention with Antidiagonal Scoring

Tokenize Image as a Set

MagicMotion: Controllable Video Generation with Dense-to-Sparse Trajectory Guidance

NuiScene: Exploring Efficient Generation of Unbounded Outdoor Scenes

SALT: Singular Value Adaptation with Low-Rank Transformation

Zero-1-to-A: Zero-Shot One Image to Animatable Head Avatars Using Video Diffusion

BigO(Bench) -- Can LLMs Generate Code with Controlled Time and Space Complexity?

Agents Play Thousands of 3D Video Games

CLS-RL: Image Classification with Rule-Based Reinforcement Learning

Towards Unified Latent Space for 3D Molecular Latent Diffusion Modeling

Sonata: Self-Supervised Learning of Reliable Point Representations

Uni-3DAR: Unified 3D Generation and Understanding via Autoregression on Compressed Spatial Tokens

MotionStreamer: Streaming Motion Generation via Diffusion-based Autoregressive Model in Causal Latent Space

Make Your Training Flexible: Towards Deployment-Efficient Video Models

See-Saw Modality Balance: See Gradient, and Sew Impaired Vision-Language Balance to Mitigate Dominant Modality Bias

MagicID: Hybrid Preference Optimization for ID-Consistent and Dynamic-Preserved Video Customization

UVE: Are MLLMs Unified Evaluators for AI-Generated Videos?

Improving Autoregressive Image Generation through Coarse-to-Fine Token Prediction

Deceptive Humor: A Synthetic Multilingual Benchmark Dataset for Bridging Fabricated Claims with Humorous Content

VideoRFSplat: Direct Scene-Level Text-to-3D Gaussian Splatting Generation with Flexible Pose and Multi-View Joint Modeling

Where do Large Vision-Language Models Look at when Answering Questions?

Painting with Words: Elevating Detailed Image Captioning with Benchmark and Alignment Learning

AIMI: Leveraging Future Knowledge and Personalization in Sparse Event Forecasting for Treatment Adherence

TikZero: Zero-Shot Text-Guided Graphics Program Synthesis

GASP: Unifying Geometric and Semantic Self-Supervised Pre-training for Autonomous Driving

Why Personalizing Deep Learning-Based Code Completion Tools Matters