Could you provide data to reproduce the Kiriko results.

cloneofsimo / lora

Using Low-rank adaptation to quickly fine-tune diffusion models.

https://arxiv.org/abs/2106.09685

Apache License 2.0

6.94k stars 479 forks source link

Could you provide data to reproduce the Kiriko results. #196

Open guyuchao opened 1 year ago

guyuchao commented 1 year ago

Hi, I am trying to train LoRA on game characters, but the results are quite poor than your Kiriko results. Could you provide the data your trained on or some details about dataset size, and show some typical examples?

guyuchao commented 1 year ago

BTW, do you use face segmentation in this experiment?