Happy to share that I have successfully defended my thesis @UCLA earlier this month. I will join @sirbayes’s team @GoogleDeepMind to continue working on (latent-variable) generative models!
Thesis: escholarship.org/uc/item/85f7z0…
Sirui Xie
@SiruiXie
- Congratulations to the best paper award winners
- Thanks @_akhaliq for the tweet. EMD matches the teacher diffusion model’s marginal distribution by making the old EM great again. Check out our paper at: arxiv.org/abs/2405.16852EM Distillation for One-step Diffusion Models While diffusion models can learn complex distributions, sampling requires a computationally expensive iterative process. Existing distillation methods enable efficient sampling, but have notable limitations, such as
- 📢 Excited to share EM Distillation (EMD), a maximum likelihood method that distills pretrained diffusion models to one-step generators. EMD gracefully interpolates between mode-seeking and mode-covering KL to better capture the teacher's distribution. arxiv.org/abs/2405.16852
