Software Engineering Courses (SECourses)•16mo ago

This is my task manager before training

Xxlads This is my task manager before training

druhl•9/7/24, 6:17 PM

Even close the task manager and google chrome windows. Just use nvitop to monitor.

xladsOP•9/7/24, 6:17 PM

Wow okay

xladsOP•9/7/24, 6:17 PM

thanks

xladsOP•9/7/24, 6:19 PM

wondering if Rank_12_7500MB_7_68_Second_IT.json will produce lower training givingit a shot now

Xxlads This is my task manager before training

druhl•9/7/24, 6:19 PM

You have like 7,592 MB available in this situation

xladsOP•9/7/24, 7:12 PM

Just started training again, see how it goes

Phuong Bui•9/7/24, 7:24 PM

I got this error when training lora on runpod, how do I fix it?
Traceback (most recent call last):
File "/workspace/kohya_ss/sd-scripts/flux_train_network.py", line 519, in <module>
trainer.train(args)
File "/workspace/kohya_ss/sd-scripts/train_network.py", line 441, in train
self.post_process_network(args, accelerator, network, text_encoders, unet)
File "/workspace/kohya_ss/sd-scripts/flux_train_network.py", line 170, in post_process_network
self.train_t5xxl = network.train_t5xxl
File "/workspace/kohya_ss/venv/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1729, in getattr
raise AttributeError(f"'{type(self).name}' object has no attribute '{name}'")
AttributeError: 'LoRANetwork' object has no attribute 'train_t5xxl'
Traceback (most recent call last):
File "/workspace/kohya_ss/venv/bin/accelerate", line 8, in <module>
sys.exit(main())
File "/workspace/kohya_ss/venv/lib/python3.10/site-packages/accelerate/commands/accelerate_cli.py", line 48, in main
args.func(args)
File "/workspace/kohya_ss/venv/lib/python3.10/site-packages/accelerate/commands/launch.py", line 1106, in launch_command
simple_launcher(args)
File "/workspace/kohya_ss/venv/lib/python3.10/site-packages/accelerate/commands/launch.py", line 704, in simple_launcher
raise subprocess.CalledProcessError(returncode=process.returncode, cmd=cmd)
subprocess.CalledProcessError: Command '['/workspace/kohya_ss/venv/bin/python', '/workspace/kohya_ss/sd-scripts/flux_train_network.py', '--config_file', '/workspace/kq/model/config_lora-20240907-192115.toml']' returned non-zero exit status 1.
19:21:35-742791 INFO Training has ended.

Kakachia777 💎•9/7/24, 7:43 PM

Which is the best way to train on Runpod Custom lora for FLUX? Is it yet again using PyTorch 2.2-2.4?

DEXperience Solutions•9/7/24, 7:49 PM

hi everyone

DEXperience Solutions•9/7/24, 7:52 PM

Necesito una ayuda, de alguien que me oriente como entrenar un Lora para lograr que los siguientes
dibujos animados sean dibujados en diferentes contexto de fondo. Segun lo que se le coloque en el prompt.
estuve entrenando un https://replicate.com/ostris/flux-dev-lora-trainer/train es decir un Lora sobre Flux dev, pero sin exito alguno. Por favor que alguien se apiade de mi y me de una luz en el camino, llevo dias probando diferente cosas.
----------------
I need help from someone who can guide me on how to train a LoRA to make the following cartoon characters be drawn in different background contexts, based on what is specified in the prompt. I have been training a LoRA using https://replicate.com/ostris/flux-dev-lora-trainer/train, that is, a LoRA on Flux dev, but without success. Please, can someone take pity on me and give me a light in this path? I’ve been trying different things for days without any success.

DDEXperience Solutions Necesito una ayuda, de alguien que me oriente como entrenar un Lora para lograr ...

�

𝗡𝗶𝘁𝗶𝗻 𝗣𝗮𝗿𝗲𝘃𝗮•9/7/24, 7:55 PM

are you trying to make NFTs?

�𝗡𝗶𝘁𝗶𝗻 𝗣𝗮𝗿𝗲𝘃𝗮 are you trying to make NFTs?

DEXperience Solutions•9/7/24, 7:57 PM

No, I already have the NFTs, I just want to make fun images based on them.

�𝗡𝗶𝘁𝗶𝗻 𝗣𝗮𝗿𝗲𝘃𝗮 are you trying to make NFTs?

DEXperience Solutions•9/7/24, 8:05 PM

Could you guide me on which technology to use?

DDEXperience Solutions Could you guide me on which technology to use?

�

𝗡𝗶𝘁𝗶𝗻 𝗣𝗮𝗿𝗲𝘃𝗮•9/7/24, 8:06 PM

how many images are you training on?

�𝗡𝗶𝘁𝗶𝗻 𝗣𝗮𝗿𝗲𝘃𝗮 how many images are you training on?

DEXperience Solutions•9/7/24, 8:08 PM

800 imgs

DDEXperience Solutions 800 imgs

�

𝗡𝗶𝘁𝗶𝗻 𝗣𝗮𝗿𝗲𝘃𝗮•9/7/24, 8:09 PM

I think you can test with a smaller dataset first are all the characters different?

DEXperience Solutions•9/7/24, 8:12 PM

I just tried training with only 4 images, but from the same country, which have more similarity. I was thinking of training a LoRA for each country and using it dynamically.

Do you think it could work? But I only have 4 images per country, and I still don’t know if I should remove the backgrounds from my .png files before training? In how many steps?

DDEXperience Solutions I just tried training with only 4 images, but from the same country, which have ...

�

𝗡𝗶𝘁𝗶𝗻 𝗣𝗮𝗿𝗲𝘃𝗮•9/7/24, 8:17 PM

I think you should try with 20 images and in the caption files include a different trigger word for each distinct character (make sure to add this at the starting of the caption) and then just describe the background like green background, etc and train for 3000 steps and test the lora and take it from there

�𝗡𝗶𝘁𝗶𝗻 𝗣𝗮𝗿𝗲𝘃𝗮 I think you should try with 20 images and in the caption files include a differe...

DEXperience Solutions•9/7/24, 8:21 PM

I’ll try that. Do you think it’s better to train it with images without backgrounds?

Luciani•9/7/24, 8:22 PM

Hi, can please someone tell to avoid vertical lines rendered in almost every image. I´m using Forge... The Lora I trained exactly from the instruction video of Dr. Furkan -- for now with 100 epochs

DDEXperience Solutions I’ll try that. Do you think it’s better to train it with images without backgrou...

�

𝗡𝗶𝘁𝗶𝗻 𝗣𝗮𝗿𝗲𝘃𝗮•9/7/24, 8:23 PM

no need to remove the background

Xxlads This is my task manager before training

Furkan Gözükara SECourses•9/7/24, 9:16 PM

you should try to get 0.4 gb

PPhuong Bui I got this error when training lora on runpod, how do I fix it? Traceback (most ...

Furkan Gözükara SECourses•9/7/24, 9:17 PM

use our config dont change it dont load into dreamboth tab even a once

KKakachia777 💎Which is the best way to train on Runpod Custom lora for FLUX? Is it yet again u...

Furkan Gözükara SECourses•9/7/24, 9:17 PM

i think not matters since we install torch 2.4

Furkan Gözükara SECourses•9/7/24, 9:17 PM

but use the instructions version pytorch 2.1 template

DDEXperience Solutions Necesito una ayuda, de alguien que me oriente como entrenar un Lora para lograr ...

Furkan Gözükara SECourses•9/7/24, 9:18 PM

here a tutorial for you : 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...

Furkan Gözükara SECourses•9/7/24, 9:18 PM

also i am preparing style training example on hugging face

LLuciani Hi, can please someone tell to avoid vertical lines rendered in almost every ima...

Furkan Gözükara SECourses•9/7/24, 9:20 PM

upscale adding it

Furkan Gözükara SECourses•9/7/24, 9:20 PM

use supir for upscale

Furkan Gözükara SECourses•9/7/24, 9:20 PM

https://youtu.be/OYxVEvDf284

YouTubeSECourses

Complete Guide to SUPIR Enhancing and Upscaling Images Like in Sci-...

In this video, I explain how to 1 click install and use the most advanced image upscaler / enhancer in the world that is both commercially and open source available. The upscaler that I am going to introduce you is open source #SUPIR and the model is free to use. SUPIR upscaler is many times better than both paid Topaz AI and Magnific AI and you...

Luciani•9/7/24, 9:21 PM

unfortunately it also happens with no upscaling / also with lower resolution like 1024x512
When I run the same prompt and seed without the lora there are no vertical lines
Is my Lora "broken"?

FFurkan Gözükara SECourses use our config dont change it dont load into dreamboth tab even a once

Phuong Bui•9/7/24, 9:22 PM

i use a40 what rank should i use ?

LLuciani unfortunately it also happens with no upscaling / also with lower resolution lik...

Furkan Gözükara SECourses•9/7/24, 9:25 PM

then very likely

PPhuong Bui i use a40 what rank should i use ?

Furkan Gözükara SECourses•9/7/24, 9:25 PM

rank 1 best

Furkan Gözükara SECourses•9/7/24, 9:25 PM

but it is too slow if you need speed

Furkan Gözükara SECourses•9/7/24, 9:25 PM

rank 2

Furkan Gözükara SECourses•9/7/24, 9:26 PM

i feel like with consistent dataset no captions worked better than captions

Furkan Gözükara SECourses•9/7/24, 9:26 PM

for style for flux training

Furkan Gözükara SECourses•9/7/24, 9:26 PM

let me share link with you

Luciani•9/7/24, 9:27 PM

i did not use any caption files and also no reference images -- just 100 epochs with repeat set to 1 having 132 images to train

Luciani•9/7/24, 9:29 PM

however i notice the vertical lines are "lesser" with lower initial resolution like 1024x512 (instead of something like 1400x700)

Luciani•9/7/24, 9:29 PM

what generation resolution do you recommend in common?

Furkan Gözükara SECourses•9/7/24, 9:33 PM

https://huggingface.co/MonsterMMORPG/3D-Cartoon-Style-FLUX/resolve/main/Fixed-Consistent-Training-Dataset-Results-Grid-50700x15500px.jpg

LLuciani however i notice the vertical lines are "lesser" with lower initial resolution l...

Furkan Gözükara SECourses•9/7/24, 9:34 PM

make all train iamges 1024x1024 and generate 1024x1024

Furkan Gözükara SECourses•9/7/24, 9:34 PM

then you can try multi resolution

LLuciani i did not use any caption files and also no reference images -- just 100 epochs ...

Furkan Gözükara SECourses•9/7/24, 10:00 PM

currently people reporting multi resolution with kohya not working very good for flux - i am not sure - but as i said first go with 1024x1024 and once you have some decent results you can try multi res training and compare

Luciani•9/7/24, 10:06 PM

or maybe it is because of my pics are being oversized for training? Or is it rather the mixed aspect ratio?

LLuciani or maybe it is because of my pics are being oversized for training? Or is it rat...

Furkan Gözükara SECourses•9/7/24, 10:09 PM

mixed aspect ratio

Furkan Gözükara SECourses•9/7/24, 10:09 PM

you have too many

Furkan Gözükara SECourses•9/7/24, 10:09 PM

and too little number of images

This is my task manager before training

Similar Threads