Saining Xie

What matters for Representation Alignment: Global Information or Spatial Structure?

Diffusion Transformers with Representation Autoencoders

Cambrian-S: Towards Spatial Supersensing in Video

Spatial Mental Modeling from Limited Views

GaussianLens: Localized High-Resolution Reconstruction via On-Demand Gaussian Densification

AutoCode: LLMs as Problem Setters for Competitive Programming

BlenderFusion: 3D-Grounded Visual Editing and Generative Compositing

BLIP3-o: A Family of Fully Open Unified Multimodal Models—Architecture, Training and Dataset

Transfer between Modalities with MetaQueries

Representation Alignment for Generation: Training Diffusion Transformers Is Easier Than You Think

On Scaling Up 3D Gaussian Splatting Training

Meta CLIP 2: A Worldwide Scaling Recipe

Dynamic Test-Time Compute Scaling in Control Policy: Difficulty-Aware Stochastic Interpolant Policy

AuroraCap: Efficient, Performant Video Detailed Captioning and a New Benchmark

Deconstructing Denoising Diffusion Models for Self-Supervised Learning

DiffusionGuard: A Robust Defense Against Malicious Diffusion-based Image Editing

PISA Experiments: Exploring Physics Post-Training for Video Diffusion Models by Watching Stuff Drop

SFT Memorizes, RL Generalizes: A Comparative Study of Foundation Model Post-training