4x gpu works. i trained straight 40 hours
4x gpu works. i trained straight 40 hours
Fbut he tries to train like 10000 images i dont know if that is broken
Fye dont change LR 
For maybe 0.00075
Dlol, I remember setting it back to 0.00005, but it defaulted to 0.0001 for some reason.
DI'm wondering whether that will break my training 
DBoth the GPUs are running fine though
Fnope should work fine
Fnice
XThis is my task manager before training
DEven close the task manager and google chrome windows. Just use nvitop to monitor.
XWow okay
Xthanks
Xwondering if Rank_12_7500MB_7_68_Second_IT.json will produce lower training givingit a shot now
DYou have like 7,592 MB available in this situation
XJust started training again, see how it goes
PI 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.
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.
KWhich is the best way to train on Runpod Custom lora for FLUX? Is it yet again using PyTorch 2.2-2.4?
Dhi everyone
DNecesito 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.
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.
�are you trying to make NFTs?
DNo, I already have the NFTs, I just want to make fun images based on them.
DCould you guide me on which technology to use?
�how many images are you training on?
D800 imgs
�I think you can test with a smaller dataset first are all the characters different?
DI 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?
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?
�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
DI’ll try that. Do you think it’s better to train it with images without backgrounds?
LHi, 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
�no need to remove the background
Fyou should try to get 0.4 gb
Fuse our config dont change it dont load into dreamboth tab even a once
Fi think not matters since we install torch 2.4
Fbut use the instructions version pytorch 2.1 template
FYouTubeSECourses
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...
Falso i am preparing style training example on hugging face
Fupscale adding it
Fuse supir for upscale
FYouTubeSECourses
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...
Lunfortunately 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"?
When I run the same prompt and seed without the lora there are no vertical lines
Is my Lora "broken"?
Pi use a40 what rank should i use ?
Fthen very likely
Frank 1 best
Fbut it is too slow if you need speed
Frank 2
Fi feel like with consistent dataset no captions worked better than captions
Ffor style for flux training
Flet me share link with you
Li did not use any caption files and also no reference images -- just 100 epochs with repeat set to 1 having 132 images to train
