Software Engineering Courses (SECourses)•16mo ago

I want to try this but I have some questions, layes=bloks what's the diference between single and

I want to try this but I have some questions, layes=bloks what's the diference between single and double blocks? you are testing with single blocks or layers? can I test with 7 and 20 using a higher lr, let us know your findings, thanks

�𝗡𝗶𝘁𝗶𝗻 𝗣𝗮𝗿𝗲𝘃𝗮 what network dim and alpha are you using?

Furkan Gözükara SECourses•9/17/24, 3:56 PM

128x128

KKallamamran @Dr. Furkan Gözükara I can't seem to find that captioning editor GUI you've made...

Furkan Gözükara SECourses•9/17/24, 3:58 PM

https://www.patreon.com/posts/108992085

Patreon

Image Captioning Editor Gradio APP - Edit Your Captions Super Easy ...

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

Ddsienra I want to try this but I have some questions, layes=bloks what's the diference...

Furkan Gözükara SECourses•9/17/24, 3:58 PM

that is what i am testing now :d

Furkan Gözükara SECourses•9/17/24, 3:58 PM

done 16 trainings so far training 8 more

FFurkan Gözükara SECourses 128x128

�

𝗡𝗶𝘁𝗶𝗻 𝗣𝗮𝗿𝗲𝘃𝗮•9/17/24, 4:07 PM

You can try 256 if you don't see good results

�𝗡𝗶𝘁𝗶𝗻 𝗣𝗮𝗿𝗲𝘃𝗮 You can try 256 if you don't see good results

Furkan Gözükara SECourses•9/17/24, 4:07 PM

yes can be improved nice idea to test

Darth Teethius•9/17/24, 4:55 PM

looks like individual block training might have been implemented in the sd3 flux1 kohya ss branch last night

xlads•9/17/24, 4:55 PM

So fine tuning is better the Lora over all you think @Dr. Furkan Gözükara ?

Darth Teethius•9/17/24, 4:56 PM

DDarth Teethius looks like individual block training might have been implemented in the sd3 flux...

Furkan Gözükara SECourses•9/17/24, 4:56 PM

i am testing already

Xxlads So fine tuning is better the Lora over all you think @Dr. Furkan Gözükara ?

Furkan Gözükara SECourses•9/17/24, 4:56 PM

yep

xlads•9/17/24, 4:57 PM

You have a video on how to fine tune training?

Ra25•9/17/24, 5:22 PM

Can someone confirm Flux Lora Extraction in completely broken on Kohya ?

Ra25•9/17/24, 5:25 PM

I did a finetune of flux dev, results are amazing. I use Extract Flux LoRA tab to extract the Lora from the finetuned model. When I use the extracted Lora it makes no difference with not using Lora. @Dr. Furkan Gözükara Did you try the Extract Flux LoRA feature ?

xlads•9/17/24, 5:40 PM

I’m confused on what fine tuning is, do we basically do the same training as in this video https://youtu.be/nySGu12Y05k?feature=shared but with the configuration from the new update on your patreon?

YouTubeSECourses

FLUX LoRA Training Simplified: From Zero to Hero with Kohya SS GUI ...

Ultimate Kohya GUI FLUX LoRA training tutorial. This tutorial is product of non-stop 9 days research and training. I have trained over 73 FLUX LoRA models and analyzed all to prepare this tutorial video. The research still going on and hopefully the results will be significantly improved and latest configs and findings will be shared. Please wat...

RRa25 Can someone confirm Flux Lora Extraction in completely broken on Kohya ?

Neo•9/17/24, 6:10 PM

I'd also like to know. Though I was waiting on Kohya to refine it a bit more before delving into fine tunes.

RRa25 Can someone confirm Flux Lora Extraction in completely broken on Kohya ?

smoke•9/17/24, 6:11 PM

Same here. Do you mean extracting a lora from a finetuned checkpoint?

Neo•9/17/24, 6:12 PM

Yeah

smoke•9/17/24, 6:12 PM

For me the extracted lora looks nothing like my subject, nothing like the checkpoint

smoke•9/17/24, 6:12 PM

I guess it is broken

Ra25•9/17/24, 7:26 PM

@smoke yeah, thx for the confirmation

smoke•9/17/24, 7:28 PM

Should probably raise an issue in sd-scripts Github since it is broken

Ra25•9/17/24, 7:35 PM

There's already one https://github.com/kohya-ss/sd-scripts/issues/1496

GitHub

Extracted flux lora not working · Issue #1496 · kohya-ss/sd-scripts

2 Issues sometimes it works, but it looks nothing like the fine tuned model. Doesnt work and i get the error below During handling of the above exception, another exception occurred: Traceback (mos...

smoke•9/17/24, 7:36 PM

Ah yeah it definitely is broken

Ra25•9/17/24, 7:36 PM

But i'm not sure it's the same issue

Ra25•9/17/24, 7:37 PM

Because they say it works at 1.5 strength

Ra25•9/17/24, 7:37 PM

But for me at 2.0 there's was no difference

smoke•9/17/24, 7:37 PM

Well for me it is, the extracted lora does not look like my finetuned checkpoint. At 1.5 strength it starts to look more like it but it created a lot of other issues and graininess, so definitely not the best approach

smoke•9/17/24, 7:37 PM

Atleast for me

Ra25•9/17/24, 7:38 PM

Comfy ?

smoke•9/17/24, 7:38 PM

I tried in Comfy yeah but I believe the same issue happened in forge

Darth Teethius•9/17/24, 8:41 PM

has anyone got the only training specific layers yet to see a speed increase? im getting same speed trying to training only 2 layers as i do when i train all

DDarth Teethius has anyone got the only training specific layers yet to see a speed increase? im...

xlads•9/17/24, 8:52 PM

what are layers?

Xxlads I’m confused on what fine tuning is, do we basically do the same training as in ...

Furkan Gözükara SECourses•9/17/24, 10:07 PM

yes true

Furkan Gözükara SECourses•9/17/24, 10:07 PM

exactly same

Furkan Gözükara SECourses•9/17/24, 10:07 PM

just use new config

Furkan Gözükara SECourses•9/17/24, 10:08 PM

checkpoints will be 23.8 gb