Massed compute or runpod?
Massed compute or runpod?
M@Dr. Furkan Gözükara my results with kohya
of walking dead star andrew lincoln
of walking dead star andrew lincoln
DIs it possible to train on Kohya for flux already?
�Yeah scroll up, the Dr. has been testing it extensively all day yesterday with good results
DHey everyone!
I'm brand new to AI art and trying to wrap my head around all these terms. I'm trying to decide between using something like Kohya or a simpler platform like Ostris AI.
Could someone give me a quick rundown of the pros and cons of each? Also, any resources or beginner guides you recommend would be much appreciated!
I'm brand new to AI art and trying to wrap my head around all these terms. I'm trying to decide between using something like Kohya or a simpler platform like Ostris AI.
Could someone give me a quick rundown of the pros and cons of each? Also, any resources or beginner guides you recommend would be much appreciated!
Mai toolkit noob friendly, kohya with the best settings will be better
Csince we cant train the text encoder right now, it makes no sense to caption the training dataset with tags like ohwx right? what should be used instead?
VEvery day there is something new to flux...
Ngreat work!!
Fmassed compute
Fyep i am sharing results too
BWhat resolutions / aspect ratios does flux support for training data, is it the same as SDXL?
Mmy pod proxy address gave me this, dr
Mdo i have to run it again ?
Mi saw some issues on github, and i seems very common
Fyep
Fyou can see this master kohya tutorial
FYouTubeSECourses
In this tutorial, you will learn how to install Automatic1111 Web UI for SDXL. How to use LoRAs with Automatic1111 SD Web UI. How to install Kohya SS GUI scripts to do Stable Diffusion training. How to train LoRAs on SDXL model with least amount of VRAM using settings. All of the details, tips and tricks of Kohya trainings. How to do x/y/z plot ...
Fit still make sense according to the situation
Fbecause when it is training unet it uses those caption vectors
Fwe cache them
Fthey are working really hard
Frented 12 gb gpu to test 12 gb flux lora training 
Ftesting flux training on rtx 3060
J...i hope you are also thinking of us lonely 4090 users 
Fi rent another machine for 24 gb gpus :d
Fso on 1 machine 48 gb on 1 machine 24 gb running
FPatreon
Get more from SECourses: Tutorials, Guides, Resources, Training, FLUX, MidJourney, Voice Clone, TTS, ChatGPT, GPT, LLM, Scripts by Furkan Gözü on Patreon
Fby the way 12 gb still fails i think kohya not properly testing
Show do you do for get that clean expresions?
Fi will try with my training lets see capability of flux
Fuploading newest trainings for testing
V@Dr. Furkan Gözükara ifound kohya does not work at the moment, right?
Vbecause json cannot be loaded from the local disk
Fit works i am treaining with kohya
�I've finally was able to get Kohya to work thanks to your instructions and with your configs, so thank you @Dr. Furkan Gözükara 
�by the way are you running all of your trainings concurrently through one single instance of Kohya? I know you rent the multiple GPUs but can you control them a certain way with their ID like you did with SwarmUI ?
Fyes via command line arguments
Fexport CUDA_VISIBLE_DEVICES=3
Fand also in accelerator settings
F--gpu_ids=1
MTurboEdit: Instant text-based image editing
�That's amazing, thank you Dr.! I'll be testing it out myself soon to see how I can train multiple Loras at once. Would be awesome if you could demonstrate this process in your next Kohya Flux tutorial video
Fyes i will show
FStarting new 8x A6000 GPU on Massed Compute
More tests on the way for FLUX ultimate LoRA workflow - for 48 GB, 24 GB and 12 GB GPUs
I should test fine tuning too
More tests on the way for FLUX ultimate LoRA workflow - for 48 GB, 24 GB and 12 GB GPUs
I should test fine tuning too
FHow to switch kohya massed compute accurate branch
D
DI'm going to start training with one of your configs, what config do you recommend me to start for a rtx 3090?
Fi havent tested new ones still testing currently among those best one is 1e_04_fp8_bf16_accelerate_full_bf16_32_rank
Fmake sure that you have like 500 mb vram usage before starting training
