default shift (base 0.5 / max 1.15) ?
default shift (base 0.5 / max 1.15) ?
Fah from your image i dont know what are they 
Kif discord doesn't resize the image maybe this is more useful. Seem pretty spot on with epoch 200 and lora_strenght 1.0 at 1.1 if starts mutilate hands.
Fye i use lora str 1
FPatreon
Get more from SECourses: Tutorials, Guides, Resources, Training, FLUX, MidJourney, Voice Clone, TTS, ChatGPT, GPT, LLM, Scripts by Furkan Gözü on Patreon
KOkey, but that is not for lora. check that out another time as it was a 48hr run with 100 images in a data set. for this lora.
Kon a RTX 3090 ti
Fye 100 images 100 epochs would take a lot of time
K100 images and 200 epochs 
KShall I use this checkpoint weight or the previous one (epoch 175) and lower the learning rate (by how much?) and do another 200 epochs?
Kow, wait used the worng check point. that's from an old run
Knot that it matter the sweet spot is still around epoch 200 and lora strength 0.95-1.0
XI was driving down in my car today and couldn’t help but notice all the billboard ads and how ai is revolutionizing that entire market
Fyep i also see so many ads are ai made
KSame problem
K
KI don't get it
LHey I used the supir from the videos tutorial and somehow it takes really long time for 50 steps with rtx 4090 any suggestion why?
KSo I ran the manual accelerate config and set GPU:1 there. Now it works
KTakes forever though
Lim getting this error with supir:( some one know why is that?
DOut of memory. How much VRAM and RAM do you have?
Loh ok i fixed it thanks it seems that too many things was using my vram
DI am trying to follow the massed compute FLUX LORA tutorial. I downloaded FLUX models, added them to stableswarmUI folders correctly. But now I get error which says "ERROR: self-start comfy-UI-0 on port 7823 failed. "
Lwhat does it do? should i enable it?
Lin the supir upscaler
Fif not using shared vram
Fit is its speed
Fstart as tiled
Fand fp8
Fnot much worthy to play with
Fwe already have 2 configs you can test them
Z@Dr. Furkan Gözükara using SUPIR and getting these errors, any suggestions? decoderF
unsupported embed per head: 512
is not supported because:
max(query.shape[-1] != value.shape[-1]) > 128
xFormers wasn't build with CUDA support
attn_bias type is <class 'NoneType'>
operator wasn't built - see python -m xformers.info for more info
is not supported because:
max(query.shape[-1] != value.shape[-1]) > 256
xFormers wasn't build with CUDA support
operator wasn't built - see python -m xformers.info for more info
is not supported because:
xFormers wasn't build with CUDA support
operator wasn't built - see python -m xformers.info for more info
is not supported because:
max(query.shape[-1] != value.shape[-1]) > 32
xFormers wasn't build with CUDA support
dtype=torch.bfloat16 (supported: {torch.float32})
operator wasn't built - see python -m xformers.info` for more infounsupported embed per head: 512
Fdid you do a fresh install of latest version_
Fthis is xformers incompability
Fmy installer shouldnt have this issue
Zi'm using arch on my local system
Fxformers only for nvidia gpus afaik
Show do I add realistic scars, do I need to create a LORA for that?
Fyou can search for better prompting too
Fbut lora may work better
Fth
Fthis config failed so bad
Fman i need 16 gpus having machine :d
Swhy you want to be bald bro?
SI tried facial scar, knife cut scar etc, it draws a bloody mess lol, but I want a cool scar like I'm some ancient warrior
Fprobably you need lora
Ftesting
Z@Dr. Furkan Gözükara Have you tried training a LORA for an object? I'm finding when i go to generate the image using the LORA i have to describe the object in detail for it to retain most of the objects details
