So what do we conclude from this
So what do we conclude from this?
Funet becomes your character
Fany noise will be denoised as your character
Fdont prompt anything
Fand generate
Fyou will see yourself
AI cant wait to experiment with that
Fif enough trained :d
AOk, so 200 steps were used
C
Cwow
Cwowowowow
C
C200 epoch
Awow
Awhat did you use for prompt?
Cphoto of man art by tomer hanuka
A
C
Cphoto of man art by craig paton
F
Cinteresting
Ftry photo of castle
C
Fu didnt train any tench?
Fthis is not expected
C
Cit sort of is, its affecting photos
C
CI think its impacted by weight look
Fye unet will become yourself
Fi tink you didnt that much overtrianed
Fthat is why
Idid you use blink to do the tags for the dataset?
Cyeah its not overtrained
Ftags are not important
Fhe didnt do tags training
CI did this
Fwhen text encoder training 0
Fif code is working as expected
Ftags are not used at all
Fcaptions
Fonly unet trained
Cso text files are irrelevant?
Fyep
Fbecause they are not used
Fin unet
Ftraining
Cyup
Funet trains only based on image and noise
Cok so then