Software Engineering Courses (SECourses)•8mo ago

<@205854764540362752> I am not 'urging' u to make a video now but I would like to know if Kohya can

@Furkan Gözükara SECourses I am not 'urging' u to make a video now but I would like to know if Kohya can be used to train HiDream model (with my personal pics) like Flux ? Can I do it in Massed Compute ?

Escanor•5/10/25, 5:50 PM

If I train model A for 3200 steps and model B for 5000 steps, both having the same base model, would the result of generating an image be the same for both if for model B I use a checkpoint that's near 3200 steps like model A (save every x epoch/steps)?

And if I train a model for 3200 steps and then retrain it for another 1800 steps for a total of 5000 steps, will that model be as efficient as if I trained a model to 5000 steps directly?

Thank you

EEscanor Yeah, I'll first try to finetune with only 8 images, will try to find the best #...

Furkan Gözükara SECourses•5/11/25, 11:30 PM

great

TTrumpYouAll <@205854764540362752> I am not 'urging' u to make a video now but I would like t...

Furkan Gözükara SECourses•5/11/25, 11:30 PM

that is also on my list but so many stuff

EEscanor If I train model A for 3200 steps and model B for 5000 steps, both having the sa...

Furkan Gözükara SECourses•5/11/25, 11:31 PM

yes it should be same 3200 + 1800 = 5000 directly

Furkan Gözükara SECourses•5/11/25, 11:31 PM

since our learning rates are all constant

Escanor•5/13/25, 3:38 PM

@Furkan Gözükara SECourses Is the kohya dreambooth sdxl 48gb config usable with the kohya sdxl lora extract configuration that you have in one of your patreon posts (not sure if I can post image here)?

I am asking because if I leave the default settings of Clamp Quantile at 0.99 and Minimum difference to 0.01, I think it says the text encoder does not need to be extracted or something similar and when I try to use the lora, it simply does not show anything of my trained model compared to if I use the settings in your image of 0.1 for Clamp Quantile and 0.00001 for Minimum difference where now it works fine.

It just made me wonder if that screenshot of the kohya sdxl lora extract settings was up to date with the latest 48gb config of kohya dreambooth settings!

I did see a recent post for lora extract, but it was for flux so I wasn't sure

Thank you for the great support!

EEscanor <@205854764540362752> Is the kohya dreambooth sdxl 48gb config usable with the k...

Furkan Gözükara SECourses•5/13/25, 4:37 PM

make it 0.000001 to extract

Furkan Gözükara SECourses•5/13/25, 4:37 PM

also flux post is updated too but flux extract not very good

JAV_Pomoe•5/14/25, 1:54 AM

Doctor, could you please tell me the best settings for LoRA training for SDXL? I saw your previous post about fine tuning... Have you already posted about LoRA training?

JJAV_Pomoe Doctor, could you please tell me the best settings for LoRA training for SDXL? I...

Furkan Gözükara SECourses•5/14/25, 2:12 AM

we have dreambooth

Furkan Gözükara SECourses•5/14/25, 2:12 AM

i recommend after that extract lora

FFurkan Gözükara SECourses i recommend after that extract lora

JAV_Pomoe•5/14/25, 2:32 AM

I see. I'll give it a try. Thank you as always.

Escanor•5/14/25, 11:15 AM

@JAV_Pomoe Personally, on my end and from what I see in all the guides, I get better results by training either with kohya_ss dreambooth or onetrainer and then extracting the lora from the finetuned model with koyha_ss. The resulting extracted lora seems better to me than one training directly trained as such. Plus, you also have the finetuned model you can keep too

It's crucial to use the good settings for extracting the lora and finetuning the model though. For extracting the LORA, the settings can be found in an image at the very bottom of this patreon post

However, do note that in the image, the Clamp Quantile and Minimum difference don't have red rectangels around them, but they do have different values than the default and you have to use those precisely.

Good luck!

Patreon

Full Workflow For Newbie Stable Diffusion Trainers For SD 1.5 Model...

Get more from SECourses: FLUX, Tutorials, Guides, Resources, Training, Scripts on Patreon

uisato•5/14/25, 4:25 PM

Kinestasis Stop Motion / Hyperlapse - [new WAN 2.1 LORAs]

https://www.youtube.com/watch?v=3rYPtYxq_3c&ab_channel=uisato

YouTubeuisato

Kinestasis Stop Motion / Hyperlapse - [WAN 2.1 LORAs Project Files]

Technique consisting in a fine-tuned AI model [WAN 2.1 / txt2vid]. Hundreds of hours of training and testing. Still far from good, but hopefully getting somewhere. I'm a huge fan of this technique.

Music by @klsr-av

You can access these new WAN 2.1 LORAs, through my Patreon profile.

#stopmotion #machinelearning #design #aiart

petecemetery•5/14/25, 10:42 PM

looking for some advice on this regularization workflow for training this flux concept. my concept im training is faces pressed against glass. So i have 75 images of faces like the above. Then I have captioned each one without any mention of pressing against glass, and have generated regularization images with those captions so that each training image has a corresponding regularization image to help isolate the concept. With this workflow, do i want to keep repeats on both training and reg images at 1?

petecemetery•5/14/25, 10:45 PM

also as long as my txt files are in the same folder as my training images , they should be used to train right?

petecemetery•5/14/25, 10:50 PM

im using the lora config for this 1voteup

Ppetecemetery looking for some advice on this regularization workflow for training this flux c...

Furkan Gözükara SECourses•5/14/25, 10:54 PM

you dont need reg images

Furkan Gözükara SECourses•5/14/25, 10:54 PM

make sure every person in dataset is different

Furkan Gözükara SECourses•5/14/25, 10:54 PM

it should work perfect

petecemetery•5/14/25, 10:54 PM

ok great, thanks

Ppetecemetery ok great, thanks

Furkan Gözükara SECourses•5/14/25, 11:04 PM

you are welcome

EEscanor <@1320367778317865031> Personally, on my end and from what I see in all the guid...

JAV_Pomoe•5/15/25, 7:32 AM

Thank you so much for the great information!

Stav•5/15/25, 11:13 PM

Hi, how many epochs are recommended for a 15 pictures training with dreamboth?

SStav Hi, how many epochs are recommended for a 15 pictures training with dreamboth?

Furkan Gözükara SECourses•5/15/25, 11:33 PM

at least 200

Furkan Gözükara SECourses•5/15/25, 11:33 PM

but you can go up to 300 and compare

FFurkan Gözükara SECourses at least 200

Stav•5/16/25, 11:42 AM

Thank you doctor

Escanor•5/16/25, 1:48 PM

Honestly I'm always wondering how to know the number of epochs and such. Should we just aim for a certain # of total steps and play with the # of epochs and repeats?

For example, with 15 images 200-300 epochs but how many images repeats? Or is it better 1 epoch and tons of repeats?

Escanor•5/16/25, 1:50 PM

I also wonder if with like 15 portrait images we should be using captions or not

Escanor•5/16/25, 1:58 PM

Also wondering because the sdxl kohya 48gb config uses 1 epoch

skyrrr•5/16/25, 4:45 PM

Hey, for flux full finetuning is there a solution faster than L40S for the 48gb config file ? Thanks

EEscanor Honestly I'm always wondering how to know the number of epochs and such. Should ...

Furkan Gözükara SECourses•5/16/25, 5:30 PM

repeat is always 1 for flux

Furkan Gözükara SECourses•5/16/25, 5:30 PM

we had to use repeat for sdxl to use more reg images due to how kohya bad coded reg images logic

EEscanor Also wondering because the sdxl kohya 48gb config uses 1 epoch 🤔

Furkan Gözükara SECourses•5/16/25, 5:31 PM

because with sdxl we want to use more reg images

Furkan Gözükara SECourses•5/16/25, 5:31 PM

and only way is more repeat with kohya

Sskyrrr Hey, for flux full finetuning is there a solution faster than L40S for the 48gb ...

Furkan Gözükara SECourses•5/16/25, 5:32 PM

what do you mean?_

Escanor•5/16/25, 5:42 PM

Ahh I see. Then for sdxl is it better to use onetrainer instead in your opinion? Considering I have only 8 portrait images and the 5000 reg images of your patreon. Also unsure if I should use captions or not

Thank you

FFurkan Gözükara SECourses what do you mean?_

skyrrr•5/16/25, 6:40 PM

I mean would an H100 run the training significantly faster than the L40S ? I'm currently at 4.8sec/iteration

EEscanor Ahh I see. Then for sdxl is it better to use onetrainer instead in your opinion?...

Furkan Gözükara SECourses•5/16/25, 9:12 PM

yes onetrainer also has EMA

Furkan Gözükara SECourses•5/16/25, 9:12 PM

which works good for SDXL

Sskyrrr I mean would an H100 run the training significantly faster than the L40S ? I'm c...

Furkan Gözükara SECourses•5/16/25, 9:12 PM

yes it is faster

Furkan Gözükara SECourses•5/16/25, 9:12 PM

are you batch size 7?

Furkan Gözükara SECourses•5/16/25, 9:13 PM

4.8 sec is low for H100

Escanor•5/16/25, 9:23 PM

Ok for onetrainer, will try.

Is FLUX really that much better than SDXL though even for just 8 training images? Wondering because it's so slow to run on my m2 macbook pro..

EEscanor Ok for onetrainer, will try. Is FLUX really that much better than SDXL though ...

Furkan Gözükara SECourses•5/16/25, 9:24 PM

very very better

Furkan Gözükara SECourses•5/16/25, 9:25 PM

so much difference

Escanor•5/16/25, 9:27 PM

Urgh ok. Because on mac I need to use a weird flux model or else tons of errors. Model is flux1-dev-Q8_0.gguf

EEscanor Urgh ok. Because on mac I need to use a weird flux model or else tons of errors....

Furkan Gözükara SECourses•5/16/25, 9:40 PM

i would use mac if only i was developing an app for iphone :d

Escanor•5/16/25, 9:41 PM

Meh, for developing from backend to frontend it's great, ai stuff it's shit lol

Escanor•5/16/25, 9:41 PM

Compatibility is atrocious

<@205854764540362752> I am not 'urging' u to make a video now but I would like to know if Kohya can

Similar Threads