UW-Madison-Lee-Lab / SFT-PG

Code for "Optimizing DDPM Sampling with Shortcut Fine-Tuning" (https://arxiv.org/abs/2301.13362), ICML 2023
MIT License
24 stars 5 forks source link

Finetuning other samplers? #1

Open tgxs002 opened 1 year ago

tgxs002 commented 1 year ago

Impressed by this paper. It's very interesting to take advantage of the Markov property of DDPM and tune it via RL. I'm curious whether this method is applicable to some recent deterministic sampling methods, like DDIM?

yingfan-bot commented 1 year ago

Hi, thanks for your interest in our work! We're currently working on deterministic sampling methods as an extension. Please stay tuned!