Xiu Li

SpeakerVid-5M: A Large-Scale High-Quality Dataset for Audio-Visual Dyadic Interactive Human Generation

Photon: Speedup Volume Understanding with Efficient Multimodal Large Language Models

Align-Then-stEer: Adapting the Vision-Language Action Models through Unified Latent Guidance

PreciseCache: Precise Feature Caching for Efficient and High-fidelity Video Generation

Stochastic Self-Guidance for Training-Free Enhancement of Diffusion Models

Dual-Robust Cross-Domain Offline Reinforcement Learning Against Dynamics Shifts

Stabilizing Knowledge, Promoting Reasoning: Dual-Token Constraints for RLVR

Cross-domain Offline Policy Adaptation with Dynamics- and Value-aligned Data Filtering

DVT-LLaVA: Vision-Language Model Personalization with Disentangled Visual Tuning

Attention as a Compass: Efficient Exploration for Process-Supervised RL in Reasoning Models

Cross-Domain Offline Policy Adaptation via Selective Transition Correction

PROF: An LLM-based Reward Code Preference Optimization Framework for Offline Imitation Learning

Unified Single Transformer for Multimodal Video Understanding and Generation

Temporal Difference Learning with Constrained Initial Representations

EgoPlan: Towards Effective Embodied Agents via Egocentric Planning

HEAR: High-frequency Enhanced Autoregressive Modeling for Identity-Preserving Image Generation

StoryCtrl: Customized Story Visualization with Fine-Grained Control

Reti-Diff: Illumination Degradation Image Restoration with Retinex-based Latent Diffusion Model

LoRA-Gen: Specializing Large Language Model via Online LoRA Generation

MindOmni: Unleashing Reasoning Generation in Vision Language Models with RGPO

Cross-Domain Offline Policy Adaptation with Optimal Transport and Dataset Constraint

SAM-R1: Leveraging SAM for Reward Feedback in Multimodal Segmentation via Reinforcement Learning

ADG: Ambient Diffusion-Guided Dataset Recovery for Corruption-Robust Offline Reinforcement Learning

Linear Differential Vision Transformer: Learning Visual Contrasts via Pairwise Differentials

Taming Rectified Flow for Inversion and Editing

LoRA-Gen: Specializing Language Model via Online LoRA Generation

InstantSwap: Fast Customized Concept Swapping across Sharp Shape Differences

Revisiting Multi-Agent World Modeling from a Diffusion-Inspired Perspective

Decentralized Transformers with Centralized Aggregation are Sample-Efficient Multi-Agent World Models

UniQA: Unified Vision-Language Pre-training for Image Quality and Aesthetic Assessment

InterDance: Reactive 3D Dance Generation with Realistic Duet Interactions

HyDance: A Novel Hybrid Dance Generation Network with temporal and frequency features

VLP: Vision-Language Preference Learning for Embodied Manipulation

Gamma: Toward Generic Image Assessment with Mixture of Assessment Experts

Offline-to-Online Reinforcement Learning with Prioritized Experience Selection

Towards Efficient LLM Grounding for Embodied Multi-Agent Collaboration

Egocentric Vision Language Planning

InteractiveCOT: Aligning Dynamic Chain-of-Thought Planning for Embodied Decision-Making