Zhangyang Wang

Neon: Negative Extrapolation From Self-Training Improves Image Generation

$\nabla$-Reasoner: LLM Reasoning via Test-Time Gradient Descent in Latent Space

E3D-Bench: A Benchmark for End-to-End 3D Geometric Foundation Models

RAPID$^3$: Tri-Level Reinforced Acceleration Policies for Diffusion Transformer

DD-Ranking: Rethinking the Evaluation of Dataset Distillation

LLMs Can Get "Brain Rot"!

Baleen: Self‑Interpretable, Robust SSMs with Stochastic Selective Memory

Self-Discriminative Optimization for Video Diffusion Models

Why RL Updates Look Sparse: An Implicit Compass Drives Optimization Bias

Fantastic Reasoning Behaviors and Where to Find Them: Unsupervised Discovery of the Reasoning Process

Data Efficient Any Transformer-to-Mamba Distillation via Attention Bridge

Enhance-A-Video: Better Generated Video for Free

COPER: Agentic Context Significantly Improves and Stabilizes LLM in Multi-Player Game

Understanding the Mixture-of-Experts with Nadaraya-Watson Kernel

Revisiting Spectral Representations in Generative Diffusion Models

SAGE: Fast, Generalizable and Photorealistic 3D Human Reconstruction from a Single Image

Stable-SPAM: How to Stably Train Large Language Models in 4-Bit

Towards Personalized Parameter Generation via Data-Conditioned Mapping

POPS: Recovering Unlearned Multi-Modality Knowledge in MLLMs with Fine-tuning and Prompt-based Attacks

Make Optimization Once and for All with Fine-grained Guidance

ResearchGPT: Benchmarking and Training LLMs for End-to-End Computer Science Research Workflows

Drag-and-Drop LLMs: Zero-Shot Prompt-to-Weights

4K4DGen: Panoramic 4D Generation at 4K Resolution

SPIN-Bench: How Well Do LLMs Plan Strategically and Reason Socially?

LoX: Low-Rank Extrapolation Robustifies LLM Safety Against Fine-tuning

Graph-KV: Breaking Sequence via Injecting Structural Biases into Large Language Models

HALoS: Hierarchical Asynchronous Local SGD over Slow Networks for Geo-Distributed Large Language Model Training

On-the-Fly Adaptive Distillation of Transformer to Dual-State Linear Attention for Long-Context LLM Serving

Rethinking Addressing in Language Models via Contextualized Equivariant Positional Encoding

Transformers Provably Learn Two-Mixture of Linear Classification via Gradient Flow

LLaMaFlex: Many-in-one LLMs via Generalized Pruning and Weight Sharing

SEAL: Steerable Reasoning Calibration of Large Language Models for Free

Scaling Up Parameter Generation: A Recurrent Diffusion Approach

REPA Works Until It Doesn’t: Early-Stopped, Holistic Alignment Supercharges Diffusion Training

SAS: Simulated Attention Score

Can Test-Time Scaling Improve World Foundation Model?

More is Less: The Pitfalls of Multi-Model Synthetic Preference Data in DPO Safety Alignment

From Low Rank Gradient Subspace Stabilization to Low-Rank Weights: Observations, Theories, and Applications

HD-Painter: High-Resolution and Prompt-Faithful Text-Guided Image Inpainting with Diffusion Models

SPAM: Spike-Aware Adam with Momentum Reset for Stable LLM Training

Understanding and Mitigating Bottlenecks of State Space Models through the Lens of Recency and Over-smoothing

Rethinking Addressing in Language Models via Contextualized Equivariant Positional Encoding

One Token Embedding Is Enough to Deadlock Your Large Reasoning Model

Cavia: Camera-controllable Multi-view Video Diffusion with View-Integrated Attention

PIPA: Preference Alignment as Prior-Informed Statistical Estimation

On the Provable Separation of Scales in Maximal Update Parameterization

R-Sparse: Rank-Aware Activation Sparsity for Efficient LLM Inference

Cavia: Camera-controllable Multi-view Video Diffusion with View-Integrated Attention

StreamingT2V: Consistent, Dynamic, and Extendable Long Video Generation from Text

From GaLore to WeLore: How Low-Rank Weights Non-uniformly Emerge from Low-Rank Gradients

Q-GaLore: Quantized GaLore with INT4 Projection and Layer-Adaptive Low-Rank Gradients

CamCo: Camera-Controllable 3D-Consistent Image-to-Video Generation

Fantastic Experts and How to Find Them: A Multi-Dimensional Study for Experts-Level Sparsification in Mixture-of-Experts

OscillationInversion: Understand the structure of Large Flow Model through the Lens of Inversion Method