search

timothybrooks / instruct-pix2pix

Other
6.1k stars 527 forks source link

The more training steps, the worse the result #102

Closed yftongbupt closed 1 year ago

yftongbupt commented 1 year ago

I use instruct pix2pix to train a style transfer model, my dataset has about 8,000 image pairs and the instruction is just "change the style to ……". I found that the more steps the model trains, the worse the inference result to get and the result changes to irrelevant to the original image. Such as I want to change a photo to anime style, but I get a twisted figure. Does anyone have similar question or know why?

lsabrinax commented 1 year ago

I have the same problem

crisz94 commented 4 months ago

@yftongbupt @lsabrinax hey guys, I have a similiar problem as well.. Did you guys find any clues about this question?

caojiehui commented 2 months ago

I have the same problem...