Software Engineering Courses (SECourses)•2y ago

I guess the base model only knows DDD extra large boobs 😄 But I dont want these silicone boobs in m

I guess the base model only knows DDD extra large boobs

But I dont want these silicone boobs in my instance model

crapthings•4/5/24, 12:47 PM

is this overfitting?

MMorpheus SX343 I guess the base model only knows DDD extra large boobs 😄 But I dont want these...

Kallamamran•4/5/24, 12:56 PM

That's the problem

Original message was deleted

Lelics•4/5/24, 2:45 PM

fake scam

NNeo Not sure who has responded but I train LoRAs (presently using Kohya, will attemp...

magical•4/5/24, 6:37 PM

the color coding is very interesting if it works it's amazing really. funny that the dude did all this work just to generate female genitals

magical•4/5/24, 7:09 PM

so does onetrainer masked training do this as well? can you mask parts of an image and describe them individually?

magical•4/5/24, 7:11 PM

hmmm if it is like in this example you can't color code it but only can mask one part of an image https://i.imgur.com/rqouPi9.png

Imgur

magical•4/5/24, 7:12 PM

so you'd have to mask the same image multiple times but that wouldn't work either because it wouldn't learn everything's relation to each other

Ccrapthings @Dr. Furkan Gözükara

Furkan Gözükara SECourses•4/5/24, 7:34 PM

ye please remove some images causes problem some times

TTuanny what settings do i adjust in supir so that it doesnt hallucinate as much. i did...

Furkan Gözükara SECourses•4/5/24, 7:35 PM

reduce text cfg

KKallamamran @Dr. Furkan Gözükara Is it possible to add a checkbox or something to make it po...

Furkan Gözükara SECourses•4/5/24, 7:42 PM

we already have that

Furkan Gözükara SECourses•4/5/24, 7:42 PM

which script are you using?

ada0436•4/5/24, 7:51 PM

Onetrainer question: Have 83 images - in TRAINING tab do I change Epochs 200 to 830? also what do I put under Stop Training After tab? 830? Thx

ada0436•4/5/24, 7:52 PM

used tier1_15.4GB fast scipt - RTX 4080

NNeo I'm having great success with the text guidance set to 4 or lower to remove arti...

Furkan Gözükara SECourses•4/5/24, 8:00 PM

good tips

NAMU•4/5/24, 8:00 PM

@Dr. Furkan Gözükara I'm watching your OneTrainer presets tutorial vid, and in the video it shows tier2_SDXL, but in this post (https://www.patreon.com/posts/96028218) it only has downloads for tier1. Just want to make sure I'm downloading the latest preset configs.

Patreon

OneTrainer Stable Diffusion XL (SDXL) Fine Tuning Best Presets | SE...

Get more from SECourses: Tutorials, Guides, Resources, Training, MidJourney, Voice Clone, TTS, ChatGPT, GPT, LLM, Scripts by Furkan Gözükara on Patreon

Aada0436 Onetrainer question: Have 83 images - in TRAINING tab do I change Epochs 200 to ...

Furkan Gözükara SECourses•4/5/24, 8:00 PM

if you make it 830 epochs

Furkan Gözükara SECourses•4/5/24, 8:00 PM

it would be 830 * 83 steps

NNAMU @Dr. Furkan Gözükara I'm watching your OneTrainer presets tutorial vid, and in t...

Furkan Gözükara SECourses•4/5/24, 8:00 PM

ye now we have better presets

Furkan Gözükara SECourses•4/5/24, 8:00 PM

upgraded

Furkan Gözükara SECourses•4/5/24, 8:01 PM

that is beauty of my patreon scripts. they get better :d

NAMU•4/5/24, 8:01 PM

Definitely! But should you update the Patreon list to include those better scripts? lol

NAMU•4/5/24, 8:01 PM

I can't find them.

NNAMU Definitely! But should you update the Patreon list to include those better scrip...

Furkan Gözükara SECourses•4/5/24, 8:01 PM

better scripts are already in post :d

NAMU•4/5/24, 8:01 PM

Is tier1 better than tier2?

Furkan Gözükara SECourses•4/5/24, 8:01 PM

currently all tiers are equal but i plan to publish more commercial wise videos on higher tier

Furkan Gözükara SECourses•4/5/24, 8:02 PM

like style

NNAMU Is tier1 better than tier2?

Furkan Gözükara SECourses•4/5/24, 8:02 PM

yes

Furkan Gözükara SECourses•4/5/24, 8:02 PM

tier 1 is better

NAMU•4/5/24, 8:02 PM

got it, I was thinking it was the opposite, thanks.

NNAMU got it, I was thinking it was the opposite, thanks.

Furkan Gözükara SECourses•4/5/24, 8:02 PM

i see. ye it may be confusing. you are welcome

Furkan Gözükara SECourses•4/5/24, 8:02 PM

maybe i should tier like SS S A

NAMU•4/5/24, 8:03 PM

lol. the military uses the same tiers (tier 1 is best), but it's always been confusing to me, because what happens when you improve tier 1? move to tier 0? lol

ada0436•4/5/24, 8:03 PM

so for 83 images - what should i set epoch to?

NNAMU lol. the military uses the same tiers (tier 1 is best), but it's always been con...

Furkan Gözükara SECourses•4/5/24, 8:03 PM

true :d

Aada0436 so for 83 images - what should i set epoch to?

Furkan Gözükara SECourses•4/5/24, 8:03 PM

well i think max 200

Furkan Gözükara SECourses•4/5/24, 8:04 PM

and save frequent epochs

Furkan Gözükara SECourses•4/5/24, 8:04 PM

because depending on your dataset

Furkan Gözükara SECourses•4/5/24, 8:04 PM

it may get overtrained

ada0436•4/5/24, 8:04 PM

ok - tried 200 before but seemed low quality - will restart again

Aada0436 ok - tried 200 before but seemed low quality - will restart again

Furkan Gözükara SECourses•4/5/24, 8:04 PM

what are you training?

ada0436•4/5/24, 8:05 PM

faces

Furkan Gözükara SECourses•4/5/24, 8:05 PM

if still was low quality either dataset is not good

Furkan Gözükara SECourses•4/5/24, 8:05 PM

different person?=

ada0436•4/5/24, 8:05 PM

same face

NAMU•4/5/24, 8:05 PM

When using regularization image sets, can I just use the set for the resolution I use all the time? I almost never render at 512x for example, always 1024 or higher. Or should I still use all the sets?

Aada0436 same face

Furkan Gözükara SECourses•4/5/24, 8:06 PM

ok then your dataset either not good

Furkan Gözükara SECourses•4/5/24, 8:06 PM

or you made mistake