Zhangyang Wang

~Zhangyang_Wang1

论文总数

32.5

年均投稿

5.9

平均评分

接收情况43/65

会议分布

ICLR

NeurIPS

ICML

COLM

发表论文 (65 篇)

202533 篇

6.5

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

ICLR 2025Poster

5.5

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

ICML 2025Poster

6.0

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

ICLR 2025Poster

6.4

Scaling Up Parameter Generation: A Recurrent Diffusion Approach

NeurIPS 2025Poster

5.5

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

ICLR 2025Poster

6.0

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

ICLR 2025Poster

6.5

SEAL: Steerable Reasoning Calibration of Large Language Models for Free

COLM 2025Poster

7.0

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

COLM 2025Poster

6.6

Rethinking Addressing in Language Models via Contextualized Equivariant Positional Encoding

ICML 2025Poster

5.3

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

ICLR 2025withdrawn

7.0

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

NeurIPS 2025Poster

4.8

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

ICLR 2025Rejected

4.3

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

ICLR 2025Rejected

6.0

Rethinking Addressing in Language Models via Contextualized Equivariant Positional Encoding

ICLR 2025Rejected

6.0

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

ICLR 2025Poster

5.0

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

ICLR 2025Rejected

4.3

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

ICLR 2025withdrawn

5.5

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

ICML 2025Poster

7.0

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

ICML 2025Poster

7.4

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

COLM 2025Poster

5.8

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

ICLR 2025Rejected

5.0

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

ICLR 2025withdrawn

6.3

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

COLM 2025Poster

7.5

4K4DGen: Panoramic 4D Generation at 4K Resolution

ICLR 2025Spotlight

6.1

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

ICML 2025Poster

6.3

Can Test-Time Scaling Improve World Foundation Model?

COLM 2025Poster

6.4

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

NeurIPS 2025Poster

6.4

SAS: Simulated Attention Score

NeurIPS 2025Poster

7.8

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

NeurIPS 2025Poster

202432 篇

5.7

Training Dynamics of Transformers to Recognize Word Co-occurrence via Gradient Flow Analysis

NeurIPS 2024Poster

6.7

Doubly Robust Instance-Reweighted Adversarial Training

ICLR 2024Poster

3.0

Prompt-Free Diffusion: Taking “Text” out of Text-to-Image Diffusion Models

ICLR 2024withdrawn

6.5

Principled Architecture-aware Scaling of Hyperparameters

ICLR 2024Poster

6.3

Data Distillation Can Be Like Vodka: Distilling More Times For Better Quality

ICLR 2024Poster

4.3

Fill with Anything: High-Resolution and Prompt-Faithful Image Completion

ICLR 2024withdrawn

7.3

Polynomial Width is Sufficient for Set Representation with High-dimensional Features

ICLR 2024Poster

4.3

Multi-Concept T2I-Zero: Tweaking Only The Text Embeddings And Nothing Else

ICLR 2024withdrawn

4.0

Forget-Me-Not: Learning to Forget in Text-to-Image Diffusion Models

ICLR 2024withdrawn

6.8

Social Reward: Evaluating and Enhancing Generative AI through Million-User Feedback from an Online Creative Community

ICLR 2024Spotlight

6.0

AlphaPruning: Using Heavy-Tailed Self Regularization Theory for Improved Layer-wise Pruning of Large Language Models

NeurIPS 2024Poster

6.0

Diffusion4D: Fast Spatial-temporal Consistent 4D generation via Video Diffusion Models

NeurIPS 2024Poster

6.8

Compressing LLMs: The Truth is Rarely Pure and Never Simple

ICLR 2024Poster

7.3

Efficient-3Dim: Learning a Generalizable Single-image Novel-view Synthesizer in One Day

ICLR 2024Poster

5.8

Safe and Robust Watermark Injection with a Single OoD Image

ICLR 2024Poster

6.7

Junk DNA Hypothesis: A Task-Centric Angle of LLM Pre-trained Weights through Sparsity

ICLR 2024Rejected

5.0

(Dynamic) Prompting might be all you need to repair Compressed LLMs

ICLR 2024Rejected

7.5

DP-OPT: Make Large Language Model Your Privacy-Preserving Prompt Engineer

ICLR 2024Spotlight

6.8

LightGaussian: Unbounded 3D Gaussian Compression with 15x Reduction and 200+ FPS

NeurIPS 2024Spotlight

5.6

($\texttt{PEEP}$) $\textbf{P}$redicting $\textbf{E}$nzym$\textbf{e}$ $\textbf{P}$romiscuity with its Molecule Mate – an Attentive Metric Learning Solution

ICLR 2024Rejected

5.3

Sparse MoE as a New Treatment: Addressing Forgetting, Fitting, Learning Issues in Multi-Modal Multi-Task Learning

ICLR 2024Rejected

6.2

Latent 3D Graph Diffusion

ICLR 2024Poster

6.0

$\textit{Read-ME}$: Refactorizing LLMs as Router-Decoupled Mixture of Experts with System Co-Design

NeurIPS 2024Poster

5.6

Expressive Gaussian Human Avatars from Monocular RGB Video

NeurIPS 2024Poster

4.8

PAIR Diffusion: A Comprehensive Multimodal Object-Level Image Editor

ICLR 2024withdrawn

5.7

Found in the Middle: How Language Models Use Long Contexts Better via Plug-and-Play Positional Encoding

NeurIPS 2024Poster

5.8

Large Spatial Model: End-to-end Unposed Images to Semantic 3D

NeurIPS 2024Poster

5.7

Sparse Cocktail: Every Sparse Pattern Every Sparse Ratio All At Once

ICLR 2024Rejected

3.7

Drag View: Generalizable Novel View Synthesis with Unposed Imagery

ICLR 2024withdrawn

6.0

Outlier Weighed Layerwise Sparsity (OWL): A Missing Secret Sauce for Pruning LLMs to High Sparsity

ICLR 2024Rejected

5.0

SteinDreamer: Variance Reduction for Text-to-3D Score Distillation via Stein Identity

ICLR 2024Rejected

4.8

Taming Mode Collapse in Score Distillation for Text-to-3D Generation

合作者 (20)

AJAY KUMAR JAISWAL7 篇

Tianlong Chen6 篇

Zhangyang Wang

发表论文 (65 篇)

202533 篇

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

On the Provable Separation of Scales in Maximal Update Parameterization

PIPA: Preference Alignment as Prior-Informed Statistical Estimation

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

One Token Embedding Is Enough to Deadlock Your Large Reasoning Model

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

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

Scaling Up Parameter Generation: A Recurrent Diffusion Approach

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

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

SEAL: Steerable Reasoning Calibration of Large Language Models for Free

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

Rethinking Addressing in Language Models via Contextualized Equivariant Positional Encoding

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

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

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

Rethinking Addressing in Language Models via Contextualized Equivariant Positional Encoding

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

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

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

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

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

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

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

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

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

4K4DGen: Panoramic 4D Generation at 4K Resolution

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

Can Test-Time Scaling Improve World Foundation Model?

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

SAS: Simulated Attention Score

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

202432 篇

Training Dynamics of Transformers to Recognize Word Co-occurrence via Gradient Flow Analysis

Doubly Robust Instance-Reweighted Adversarial Training

Prompt-Free Diffusion: Taking “Text” out of Text-to-Image Diffusion Models

Principled Architecture-aware Scaling of Hyperparameters

Data Distillation Can Be Like Vodka: Distilling More Times For Better Quality

Fill with Anything: High-Resolution and Prompt-Faithful Image Completion

Polynomial Width is Sufficient for Set Representation with High-dimensional Features

Multi-Concept T2I-Zero: Tweaking Only The Text Embeddings And Nothing Else

Forget-Me-Not: Learning to Forget in Text-to-Image Diffusion Models

Social Reward: Evaluating and Enhancing Generative AI through Million-User Feedback from an Online Creative Community

AlphaPruning: Using Heavy-Tailed Self Regularization Theory for Improved Layer-wise Pruning of Large Language Models

Diffusion4D: Fast Spatial-temporal Consistent 4D generation via Video Diffusion Models

Compressing LLMs: The Truth is Rarely Pure and Never Simple

Efficient-3Dim: Learning a Generalizable Single-image Novel-view Synthesizer in One Day

Safe and Robust Watermark Injection with a Single OoD Image

Junk DNA Hypothesis: A Task-Centric Angle of LLM Pre-trained Weights through Sparsity

(Dynamic) Prompting might be all you need to repair Compressed LLMs

DP-OPT: Make Large Language Model Your Privacy-Preserving Prompt Engineer

LightGaussian: Unbounded 3D Gaussian Compression with 15x Reduction and 200+ FPS

($\texttt{PEEP}$) $\textbf{P}$redicting $\textbf{E}$nzym$\textbf{e}$ $\textbf{P}$romiscuity with its Molecule Mate – an Attentive Metric Learning Solution

Sparse MoE as a New Treatment: Addressing Forgetting, Fitting, Learning Issues in Multi-Modal Multi-Task Learning

Latent 3D Graph Diffusion

$\textit{Read-ME}$: Refactorizing LLMs as Router-Decoupled Mixture of Experts with System Co-Design

Expressive Gaussian Human Avatars from Monocular RGB Video

PAIR Diffusion: A Comprehensive Multimodal Object-Level Image Editor

Found in the Middle: How Language Models Use Long Contexts Better via Plug-and-Play Positional Encoding

Large Spatial Model: End-to-end Unposed Images to Semantic 3D

Sparse Cocktail: Every Sparse Pattern Every Sparse Ratio All At Once

Drag View: Generalizable Novel View Synthesis with Unposed Imagery

Outlier Weighed Layerwise Sparsity (OWL): A Missing Secret Sauce for Pruning LLMs to High Sparsity

SteinDreamer: Variance Reduction for Text-to-3D Score Distillation via Stein Identity

Taming Mode Collapse in Score Distillation for Text-to-3D Generation