Software Engineering Courses (SECourses)•11mo ago

@Dr. Furkan Gözükara where do you place the .txt captioning files when training a flux checkpoint wi

@Dr. Furkan Gözükara where do you place the .txt captioning files when training a flux checkpoint with your latest dreambooth setup?

I have captions made from JoyCaptions, do i just leave them in the input dataset folder?

Mm3dia_offline @Dr. Furkan Gözükara where do you place the .txt captioning files when training ...

Furkan Gözükara SECourses•2/17/25, 10:12 PM

yes same folder

Furkan Gözükara SECourses•2/17/25, 10:12 PM

image1.png

Furkan Gözükara SECourses•2/17/25, 10:12 PM

image1.txt

phoenix•2/18/25, 1:37 AM

@Furkan Gözükara SECourses is captioning better while training on flux+dreambooth? does it give better results?

m3dia_offlineOP•2/18/25, 5:11 AM

i don't understand this error when i hit start training in dreambooth kohya, does anyone know why I have this "data not found error"?

my kohya prepared data directory looks correct to me :/

Pphoenix @Furkan Gözükara SECourses is captioning better while training on flux+dreamboot...

Furkan Gözükara SECourses•2/18/25, 9:51 AM

nope

Furkan Gözükara SECourses•2/18/25, 9:51 AM

i prefer activation token + class token

Furkan Gözükara SECourses•2/18/25, 9:52 AM

like ohwx man

Mm3dia_offline i don't understand this error when i hit start training in dreambooth kohya, doe...

Furkan Gözükara SECourses•2/18/25, 9:52 AM

dataset error

phoenix•2/18/25, 9:52 AM

Ok so I shouldn't add captions right?

Furkan Gözükara SECourses•2/18/25, 9:52 AM

watch this section of this video

Furkan Gözükara SECourses•2/18/25, 9:52 AM

https://youtu.be/nySGu12Y05k

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...

Furkan Gözükara SECourses•2/18/25, 9:52 AM

17:08 Dataset preparation, emphasizing image captioning, cropping, and resizing

Furkan Gözükara SECourses•2/18/25, 9:52 AM

@m3dia_offline

Pphoenix Ok so I shouldn't add captions right?

Furkan Gözükara SECourses•2/18/25, 9:52 AM

yes i dont recommend

phoenix•2/18/25, 9:56 AM

@Furkan Gözükara SECourses i have 4 A6000 total 192 gb ram the training config will automatically distribute the load among all 4 right?

phoenix•2/18/25, 9:57 AM

so the speed will be quiet fast for 48 gb 1 batch model?

Pphoenix @Furkan Gözükara SECourses i have 4 A6000 total 192 gb ram the training config w...

Furkan Gözükara SECourses•2/18/25, 10:00 AM

nope

Furkan Gözükara SECourses•2/18/25, 10:00 AM

you need to use multi config

Furkan Gözükara SECourses•2/18/25, 10:00 AM

and only lora works

Pphoenix @Furkan Gözükara SECourses i have 4 A6000 total 192 gb ram the training config w...

Furkan Gözükara SECourses•2/18/25, 10:00 AM

this video shows logic and how to multi gpu https://youtu.be/-uhL2nW7Ddw

YouTubeSECourses

Blazing Fast & Ultra Cheap FLUX LoRA Training on Massed Compute & R...

Unlock the power of FLUX LoRA training, even if you're short on GPUs or looking to boost speed and scale! This comprehensive guide takes you from novice to expert, showing you how to use Kohya GUI for creating top-notch FLUX LoRAs in the cloud. We'll cover everything: maximizing quality, optimizing speed, and finding the best deals. With our exc...

phoenix•2/18/25, 10:03 AM

https://www.reddit.com/r/StableDiffusion/s/kd3aq9o7Vy

Can you do a tutorial on this please

From the StableDiffusion community on Reddit: Some Real(ly AI-Gener...

Explore this post and more from the StableDiffusion community

phoenix•2/18/25, 10:04 AM

They are claiming ultra photo realistic it would be really helpful

phoenix•2/18/25, 10:04 AM

For generating photos of person

Pphoenix https://www.reddit.com/r/StableDiffusion/s/kd3aq9o7Vy Can you do a tutorial on...

Furkan Gözükara SECourses•2/18/25, 10:16 AM

these more looks like prompting related

Furkan Gözükara SECourses•2/18/25, 10:16 AM

but let me test on myself new trained model an example

phoenix•2/18/25, 10:21 AM

Yes please and let us know the results

Pphoenix Yes please and let us know the results 🙂

Furkan Gözükara SECourses•2/18/25, 10:34 AM

how does this look didnt try prompt engineering just upscale of dreambooth model

Furkan Gözükara SECourses•2/18/25, 10:34 AM

1019-photograph_of_ohwx_man_riding_a_powerful-Lion_8e_06-32770944-3.png

phoenix•2/18/25, 11:59 AM

hi @Furkan Gözükara SECourses should I change the learning rate and max train steps for this configuration i am using 48gb batch 1 model and have 214 images ,let me know if i should change or is this perfect?

Pphoenix hi @Furkan Gözükara SECourses should I change the learning rate and max train st...

Furkan Gözükara SECourses•2/18/25, 12:04 PM

you dont need to change anything

Furkan Gözükara SECourses•2/18/25, 12:05 PM

but 300 epoch for 214 images can be overkill

Furkan Gözükara SECourses•2/18/25, 12:05 PM

still you are saving every 10 epoch so you can compare and find best checkpoint

�

𝗡𝗶𝘁𝗶𝗻 𝗣𝗮𝗿𝗲𝘃𝗮•2/18/25, 12:19 PM

@Furkan Gözükara SECourses getting close to 5 minutes on 1 H100 full lora -

Furkan Gözükara SECourses•2/18/25, 12:31 PM

NVME linked machine can use 48 GB VRAM

amazing

Furkan Gözükara SECourses•2/18/25, 12:31 PM

on massed compute

phoenix•2/18/25, 12:35 PM

@Furkan Gözükara SECourses how much time will this take? or should I do 50 epochs and one checkpoint per 5 epochs?

MetaKan•2/18/25, 12:42 PM

does kohya supports traning for Flux Pro?

MMetaKan does kohya supports traning for Flux Pro?

�

𝗡𝗶𝘁𝗶𝗻 𝗣𝗮𝗿𝗲𝘃𝗮•2/18/25, 12:45 PM

flux pro is not open source and only available via API

Pphoenix @Furkan Gözükara SECourses how much time will this take? or should I do 50 epoch...

Furkan Gözükara SECourses•2/18/25, 12:58 PM

it will take 114 hours

Furkan Gözükara SECourses•2/18/25, 12:58 PM

yes do 50 epochs

MMetaKan does kohya supports traning for Flux Pro?

Furkan Gözükara SECourses•2/18/25, 12:58 PM

flux pro not published so not supporting

Furkan Gözükara SECourses•2/18/25, 2:02 PM

here an important tips : https://www.reddit.com/r/SECourses/comments/1isdbw4/when_you_see_your_gpu_watt_usage_is_below_maximum/

From the SECourses community on Reddit: When you see your GPU watt ...

Explore this post and more from the SECourses community

Furkan Gözükara SECourses•2/18/25, 2:02 PM

6 new experiments going on https://www.reddit.com/r/SECourses/comments/1iscy60/new_6_experiments_started_to_find_out_hopefully/

From the SECourses community on Reddit: New 6 experiments started t...

Explore this post and more from the SECourses community

Furkan Gözükara SECourses•2/18/25, 2:58 PM

updating my swarmui kaggle notebook it is working better than before

phoenix•2/18/25, 3:09 PM

@Furkan Gözükara SECourses for lora you told the formula to caluclate LR but for flux the LR is fixed to 0.000004 , i am doing 100 epichs and 10 checkpoints on H100

phoenix•2/18/25, 3:10 PM

so should I keep the LR same 0.0000004

phoenix•2/18/25, 3:10 PM

phoenix•2/18/25, 3:13 PM

@Furkan Gözükara SECourses I am getting this error when I click on start training

@Dr. Furkan Gözükara where do you place the .txt captioning files when training a flux checkpoint wi

Similar Threads