Kun Gai

10

Papers

132

Total Citations

Papers (10)

Improving Video Generation with Human Feedback

LEARN: Knowledge Adaptation from Large Language Model to Recommendation for Practical Industrial Application

FullDiT: Video Generative Foundation Models with Multimodal Control via Full Attention

Granularity-Adaptive Spatial Evidence Tokenization for Video Question Answering

Video-LaVIT: Unified Video-Language Pre-training with Decoupled Visual-Motional Tokenization

Towards Precise Scaling Laws for Video Diffusion Transformers

Imbalance in Balance: Online Concept Balancing in Generation Models

Joint Optimization of Tree-based Index and Deep Model for Recommender Systems

State Regularized Policy Optimization on Data with Dynamics Shift

KuaiSim: A Comprehensive Simulator for Recommender Systems