hmm, @Furkan Gözükara SECourses can you install tokenizer without problems? I think I have to reins
hmm, @Furkan Gözükara SECourses can you install tokenizer without problems? I think I have to reinstall torch again
FDidn't test recently
Cwont work
FI can test in 30 min
Cdont worry, I got it!
C
C12 images
Cbecause we do not have a big dataset, we dont need files and all files use same 75 tokens
CHere are the class images being generated with the sampe prompt.
Theoretically, if I fine tune with proper weights
Theoretically, if I fine tune with proper weights
Cthe same prompt should in next generation have more red hair
Cand at least, not disturb the images
Cooooh
Fyou are really making experimetns
Fthat others didnt :d
CIs that bad?
C
CIm trying 
Fvery low loss looks good
Fcurrently i am focused instruct-pix2pix
CWell I hope I keep you entertained 
Cactually
Cthose are REALLY low numbers
C
Cthats what normally people get
Fyes really low
Flearning super good
C0.12 avg
0.085 avg
0.085 avg
C30% weakerr? Probably less because the upper you have to take into consideration warmup I think?
Fnot sure about that
Fmaybe opposite
Coh wow
CSeed before tuning
Cohwx woman
Cafter fine tuning ohwx woman ceteris paribus
C
Cbefore tuning, after tuning same seed
CBefore and after
I@Furkan Gözükara SECourses in your last video, when you set class images per instance to 48, and 200 epochs - how many total training steps did you get?
Iand how long was your training process?
Fclass images count don't affect your training process speed or steps count
Fit depends on your cpu but it would get calculated like this
Flets say you have 10 training images
F1 epoch would mean 10 + 10 (this comes from classiication images) now in same bucket so 20 steps
Flets say your gpu can process 1 it second so 1 step second
Fso 200 epoch would mean 200 * 20 = 4000 steps = 4000 seconds
Iso what settings would you use for 34 instance images?
CInstance images: 10 = I
Epoch: I+I(CI)
36+36(
Epoch: I+I(CI)
36+36(
Ffor dreambooth i think you can test same. just save every 5 epoch. also try lower training rate if it gets too fast overtrained
Fso like
F50 classification for example for each instance
F1 batch 1 gradient default 0.000002 learn rate