perhaps I will when I get some cash, can i still do something basic now
perhaps I will when I get some cash, can i still do something basic now
Fyou can do sdxl training atm. until get ram try to learn concepts
Fdont forget this guys
Falso you can do training on massed compute
Ri taught it used mainly vram
Fit uses but flux is too big
Fand has to use ram since you dont have 48 gb gpu
Rwell it is fast system ram, even if not vram
Fwith flux a newer technique being used
Fblock swapping
Fit i just implemented like 1 month ago
Rwell i not getting a new gpu anytime soon as much as i would love one. Perhaps i should get my feet wet before i invest that dough
Ris it okay to have a111 on HDD, or should i move it my SSD
Fyou should reinstall
Fyou cant move venv
Fbut you can move models
Rso i should just do a git clone since nothing is really setup in it yet
Fi also have auto installer if you are patreon sub
Rshould my datasets be on ssd as well. There isn't alot of room on it.
AHi Doc... can you explain this to the 5yrs child I am?... I edited the josn and pointed to the Kaggle notebook; unfortunately after the deployment, the kohya instance did'nt show the base model I put in the json file (hosted on my HF account)...
Fnope it can be anywhere
For models
Fusing kaggle hard because it is free
FYouTubeSECourses
Notebook updated to V5 with huge improvements. If you are having trouble to download models onto your cloud platforms such as RunPod, Google Colab, Kaggle, Massed Compute or anywhere, this tutorial is made for you. Moreover, you can upload your files / models into your private Hugging Face repository and save them permanently privately and downl...
Fyou need to download into kaggle
Fthen you need to give its path
Rso far your the only one who seems to making good resources for this @Furkan Gözükara SECourses
Xdoes Gradient accumulate steps make lora quality bad?
�if all training pics are 2048x2048, should I keep them that way, or reduce to 1024x1024?
RHow do I install xformers
Fit is increasing batch size
Fdont do it unless you know what you are doing
R
FYouTubeSECourses
Welcome to the Pillars of AI Application Libraries: How to Install Tutorial! 
Are you ready to embark on a journey of installing AI libraries and applications with ease? 
In this video, we'll guide you through the process of installing Python, Git, Visual Studio C++ Compile tools, and FFmpeg on your Windows 10 machine. 
We'll also show you ho...
Are you ready to embark on a journey of installing AI libraries and applications with ease? In this video, we'll guide you through the process of installing Python, Git, Visual Studio C++ Compile tools, and FFmpeg on your Windows 10 machine. We'll also show you ho...Rokay i look at that video
Xyeah that's what I want to do, since 12GB vram cant handle 2batch size with prodigy I was gonna add some accu steps, but saw some people say that it does lower the quality
Fbigger batch size = lowers quality
Ftrue
Rdo i need s pefic version of torch or would the most recent work
ROkay xformers is installing now, thanks
Rdo i use cuda 11.8 or 12.4
Ryour video used 11.8
KI have a general question for anyone who might know: When working with existing Flux checkpoints, should I:
Take an existing checkpoint and train my concepts on top of it, or
Merge my checkpoint (containing my concepts) with another existing checkpoint?
Will both approaches yield the same results? I'm wondering if training on top of an existing checkpoint is better, or if it's just a waste of time when I could simply merge checkpoints instead
Take an existing checkpoint and train my concepts on top of it, or
Merge my checkpoint (containing my concepts) with another existing checkpoint?
Will both approaches yield the same results? I'm wondering if training on top of an existing checkpoint is better, or if it's just a waste of time when I could simply merge checkpoints instead
Fcuda 12.4
Fnow we recommend that
Ftrain top of it
F= best results
Rbut nothing works if in use 12.4
Fmy newer scripts are all working