search

Zeju1997 / oft

Official implementation of "Controlling Text-to-Image Diffusion by Orthogonal Finetuning".
https://oft.wyliu.com/
MIT License
280 stars 14 forks source link

dreambooth lora training slower than dreambooth oft? #8

Open xiaohaipeng opened 1 year ago

xiaohaipeng commented 1 year ago

in papaer,oft use less params to train,why train slower than lora image