Yang You

MoNE: Replacing Redundant Experts with Lightweight Novices for Structured Pruning of MoE

ASCIIEval: Benchmarking Models' Visual Perception in Text Strings via ASCII Art

K-Sort Eval: Efficient Preference Evaluation for Visual Generation via Corrected VLM-as-a-Judge

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

FOCUS: Efficient Keyframe Selection for Long Video Understanding

DD-Ranking: Rethinking the Evaluation of Dataset Distillation

MoST: Mixing Speech and Text with Modality-Aware Mixture of Experts

Unbiased Gradient Low-Rank Projection

Training Variable Long Sequences With Data-centric Parallel

POME: Post Optimization Model Edit via Matrix Orthogonalization

Enhance-A-Video: Better Generated Video for Free

SRDiffusion: Accelerate Diffusion Inference via Sketching-Rendering Cooperation

Region-Adaptive Sampling for Diffusion Transformers

QuantGen: Parameter Generation for Controllable Model Quantization

Faster Vision Mamba is Rebuilt in Minutes via Merged Token Re-training

Towards Personalized Parameter Generation via Data-Conditioned Mapping

On-the-Fly Data Augmentation via Gradient-Guided and Sample-Aware Influence Estimation

SortedRL: Accelerating RL Training for LLMs through Online Length-aware Scheduling

SeedFT: Structure-Preserving Fusion for Multi-Seed LLM Fine-Tuning

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

MixEval-X: Any-to-any Evaluations from Real-world Data Mixture

ElasticMM: Efficient Multimodal LLMs Serving with Elastic Multimodal Parallelism

Real-Time Video Generation with Pyramid Attention Broadcast

Neural-Driven Image Editing

MERIT: Maximum-normalized Element-wise Ratio for Language Model Large-batch Training

Scaling Up Parameter Generation: A Recurrent Diffusion Approach

StarTrail: Concentric Ring Sequence Parallelism for Efficient Near-Infinite-Context Transformer Model Training

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

Ensemble Debiasing Across Class and Sample Levels for Fairer Prompting Accuracy

Can a Large Language Model be a Gaslighter?

Train Small, Infer Large: Memory-Efficient LoRA Training for Large Language Models

DSP: Dynamic Sequence Parallelism for Multi-Dimensional Transformers

Sparse MeZO: Less Parameters for Better Performance in Zeroth-Order LLM Fine-Tuning

Visual Perception in Text Strings

Sparse MeZO: Less Parameters for Better Performance in Zeroth-Order LLM Fine-Tuning

MERIT: Maximum-normalized Element-wise Ratio for Language Model Large-batch Training

SeedLoRA: A Fusion Approach to Efficient LLM Fine-Tuning

Dynamic Diffusion Transformer

SeedLoRA: A Fusion Approach to Efficient LLM Fine-Tuning

Info-Coevolution: An Efficient Framework for Data Model Coevolution

Unsupervised Learning for Class Distribution Mismatch

DSP: Dynamic Sequence Parallelism for Multi-Dimensional Transformers

Let the Rule Speak: Enhancing In-context Learning Debiasing with Interpretability

COBias and Debias: Minimizing Language Model Pairwise Accuracy Bias via Nonlinear Integer Programming

Prioritize Alignment in Dataset Distillation

Recurrent Diffusion for Large-Scale Parameter Generation

Emphasizing Discriminative Features for Dataset Distillation in Complex Scenarios

Trusted and Interactive Clustering for Time-Series Data

A Closer Look at Time Steps is Worthy of Triple Speed-Up for Diffusion Model Training

EnvBridge: Bridging Diverse Environments with Cross-Environment Knowledge Transfer for Embodied AI

Generating GFlowNets as You Wish with Diffusion Process

Conditional LoRA Parameter Generation

SOP-Agent: Empower General Purpose AI Agent with Domain-Specific SOPs

EfficientSkip: Efficiently Transforming Dense LLMs into Sparse Variants