Shen

papers

787

total citations

papers (27)

UniCTokens: Boosting Personalized Understanding and Generation via Unified Concept Tokens

NEURIPS 2025arXiv

citations

OpenCUA: Open Foundations for Computer-Use Agents

NEURIPS 2025arXiv

citations

Initializing Models with Larger Ones

ICLR 2024arXiv

citations

On Reasoning Strength Planning in Large Reasoning Models

NEURIPS 2025arXiv

citations

FaceShot: Bring Any Character into Life

ICLR 2025arXiv

citations

SPORTU: A Comprehensive Sports Understanding Benchmark for Multimodal Large Language Models

ICLR 2025arXiv

citations

USE: Universal Segment Embeddings for Open-Vocabulary Image Segmentation

CVPR 2024arXiv

citations

Cross-Modal Safety Mechanism Transfer in Large Vision-Language Models

ICLR 2025arXiv

citations

AdaptiveStep: Automatically Dividing Reasoning Step through Model Confidence

ICML 2025arXiv

citations

Epistemic Alignment: A Mediating Framework for User-LLM Knowledge Delivery

COLM 2025arXiv

citations

What Do Latent Action Models Actually Learn?

NEURIPS 2025arXiv

citations

DoF: A Diffusion Factorization Framework for Offline Multi-Agent Reinforcement Learning

ICLR 2025

citations

CoreGuard: Safeguarding Foundational Capabilities of LLMs Against Model Stealing in Edge Deployment

NEURIPS 2025arXiv

citations

ResKoopNet: Learning Koopman Representations for Complex Dynamics with Spectral Residuals

ICML 2025arXiv

citations

Improving Zero-Shot Generalization for CLIP with Variational Adapter

ECCV 2024

citations

Multi-Modal View Enhanced Large Vision Models for Long-Term Time Series Forecasting

NEURIPS 2025arXiv

citations

Sensitivity-Constrained Fourier Neural Operators for Forward and Inverse Problems in Parametric Differential Equations

ICLR 2025arXiv

citations

STEP: A Unified Spiking Transformer Evaluation Platform for Fair and Reproducible Benchmarking

NEURIPS 2025arXiv

citations

Layer as Puzzle Pieces: Compressing Large Language Models through Layer Concatenation

NEURIPS 2025arXiv

citations

VL-SAE: Interpreting and Enhancing Vision-Language Alignment with a Unified Concept Set

NEURIPS 2025arXiv

citations

Fully Spiking Neural Networks for Unified Frame-Event Object Tracking

NEURIPS 2025arXiv

citations

Shen

papers (27)

Solving Inverse Problems with Latent Diffusion Models via Hard Data Consistency

Matcher: Segment Anything with One Shot Using All-Purpose Feature Matching

VCP-CLIP: A visual context prompting model for zero-shot anomaly segmentation

Long-Term TalkingFace Generation via Motion-Prior Conditional Diffusion Model

RMB: Comprehensively benchmarking reward models in LLM alignment

Depth Any Video with Scalable Synthetic Data

UniCTokens: Boosting Personalized Understanding and Generation via Unified Concept Tokens

OpenCUA: Open Foundations for Computer-Use Agents

Initializing Models with Larger Ones

On Reasoning Strength Planning in Large Reasoning Models

FaceShot: Bring Any Character into Life

SPORTU: A Comprehensive Sports Understanding Benchmark for Multimodal Large Language Models

USE: Universal Segment Embeddings for Open-Vocabulary Image Segmentation

Cross-Modal Safety Mechanism Transfer in Large Vision-Language Models

AdaptiveStep: Automatically Dividing Reasoning Step through Model Confidence

Epistemic Alignment: A Mediating Framework for User-LLM Knowledge Delivery

What Do Latent Action Models Actually Learn?

DoF: A Diffusion Factorization Framework for Offline Multi-Agent Reinforcement Learning

CoreGuard: Safeguarding Foundational Capabilities of LLMs Against Model Stealing in Edge Deployment

ResKoopNet: Learning Koopman Representations for Complex Dynamics with Spectral Residuals

Improving Zero-Shot Generalization for CLIP with Variational Adapter

Multi-Modal View Enhanced Large Vision Models for Long-Term Time Series Forecasting

Sensitivity-Constrained Fourier Neural Operators for Forward and Inverse Problems in Parametric Differential Equations

STEP: A Unified Spiking Transformer Evaluation Platform for Fair and Reproducible Benchmarking

Layer as Puzzle Pieces: Compressing Large Language Models through Layer Concatenation

VL-SAE: Interpreting and Enhancing Vision-Language Alignment with a Unified Concept Set

Fully Spiking Neural Networks for Unified Frame-Event Object Tracking

papers (27)

Solving Inverse Problems with Latent Diffusion Models via Hard Data Consistency

Matcher: Segment Anything with One Shot Using All-Purpose Feature Matching

VCP-CLIP: A visual context prompting model for zero-shot anomaly segmentation

Long-Term TalkingFace Generation via Motion-Prior Conditional Diffusion Model

RMB: Comprehensively benchmarking reward models in LLM alignment

Depth Any Video with Scalable Synthetic Data

UniCTokens: Boosting Personalized Understanding and Generation via Unified Concept Tokens

OpenCUA: Open Foundations for Computer-Use Agents

Initializing Models with Larger Ones

On Reasoning Strength Planning in Large Reasoning Models

FaceShot: Bring Any Character into Life

SPORTU: A Comprehensive Sports Understanding Benchmark for Multimodal Large Language Models

USE: Universal Segment Embeddings for Open-Vocabulary Image Segmentation

Cross-Modal Safety Mechanism Transfer in Large Vision-Language Models

AdaptiveStep: Automatically Dividing Reasoning Step through Model Confidence

Epistemic Alignment: A Mediating Framework for User-LLM Knowledge Delivery

What Do Latent Action Models Actually Learn?

DoF: A Diffusion Factorization Framework for Offline Multi-Agent Reinforcement Learning

CoreGuard: Safeguarding Foundational Capabilities of LLMs Against Model Stealing in Edge Deployment

ResKoopNet: Learning Koopman Representations for Complex Dynamics with Spectral Residuals

Improving Zero-Shot Generalization for CLIP with Variational Adapter

Multi-Modal View Enhanced Large Vision Models for Long-Term Time Series Forecasting

Sensitivity-Constrained Fourier Neural Operators for Forward and Inverse Problems in Parametric Differential Equations

STEP: A Unified Spiking Transformer Evaluation Platform for Fair and Reproducible Benchmarking

Layer as Puzzle Pieces: Compressing Large Language Models through Layer Concatenation

VL-SAE: Interpreting and Enhancing Vision-Language Alignment with a Unified Concept Set

Fully Spiking Neural Networks for Unified Frame-Event Object Tracking