Software Engineering Courses (SECourses)•16mo ago

I’m confused on what fine tuning is, do we basically do the same training as in this video https://y

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