First-authored paper: Streamlined Inference for Video LDMs (NeurIPS 2024)
First-authored paper: ToP-ViM (NeurIPS 2024)
First-authored paper: SF-V (NeurIPS 2024)
First-authored paper: ToR-SSM (EMNLP 2024)
First-authored paper: LOTUS (DAC 2024)
First-authored paper: Mobile-SR (ECCV 2022)
First-authored paper: MOC (ICCAD 2023)
Co-authored paper D3HR accepted by ICML 2025
Co-authored paper Condense accepted by DAC 2023
Co-authored paper All-in-One accepted by ICCAD 2022
Co-authored paper accepted by ICCV 2021
Preprint: 'Taming Diffusion Transformer for Efficient Mobile Video Generation in Seconds' achieves >10 FPS on iPhone 16 Pro Max via compact VAE, tri-level pruning, and adversarial step distillation
Preprint: 'H3AE' proposes a high-compression, high-speed, high-quality autoencoder for video diffusion models