Hello, Hello, Quick question does anyone know is there a Formula for Embedding Learning Rate and Ba
Hello, Hello, Quick question does anyone know is there a Formula for Embedding Learning Rate and Batch Size for best results
AI am trying to train Textual Inversion on my RTXA5000 card for Stable diffusion 2.1 and not getting quite good results.
Cyou are improving
Yhow are u guys getting good results? every model i download looks, kinda ok ;/ i think it's my prompts, idk.
Frunpod io runs on a linux os
Fhello. could you install directly in C drive? still looks to me as path permission error
Fsure i can help. but need more info
Fvery good congrats
Fthank you. any more suggestions?
FI would first test 1 batch size 1 gradient and default learning rate
Fyou tried that?
Fthe batch size and gradient actually changes how model learns . it is mini batches vs full batches. which one better is debated
AYeah, Tried with 2 Vectors but seems very inflexible
Av5 for 20 input images working slyly better but not quite there yet
Fye it requires experimenting
AI'll try more combinations and see what it comes up with.
GHi! I think somehow the models didn't download properly, they were only just over 1gb when I did a git pull, but 5gb when I downloaded each individually
Fah so fixed now?
Fwhen you do git pull
Fyou have to wait a lot
GI followed your suggestion of watching the ethernet, but must have biffed it
GThank's for you video btw, super helpful
Fglad to help
Gand thanks for taking the time to respond!
HI think you need to install git LFS to download large files
HWith git
Fit still downloads
Fbut doesnt show progress
Fby the way i installed gitbash
Fmaybe that is why i am able to
Bhow to use controlnet in colab ?
Fif there is a colab supporting automatic1111 could be possible but i havent tested yet
Fi plan to show in runpodio
Fguys give me more ideas about runpod
FGreetings everyone. In this video I am going to Show how to use Automatic1111 Web UI for stable diffusion tasks on runpod io like you are using it on your computer. I will cover many topics such as how to upload and download files quickly, how to delete directories, how to install and run extensions, how to quickly download and run custom models, how to do dreambooth training on sd 1.5 and sd 2.1 versions, how to do fine tuning via dreambooth, how to do textual inversion training. I will also explain how their pricing system works, how you can use bidding, how you can transfer your files from 1 pod to another pod. I will also demonstrate how you can use new famous controlnet on runpod io.
HHeres a couple..
- Blackblaze storage option/setup .. they seem to reccomend but I couldnt get it right
- how to Download custom Repos in a pod (Kohya ect)
- Controlnet setup
- How to use RunCtrl cmd
Fawesome ty
IBackblaze is possible - pretty strait forward - they have a tutorial in Their blog
II’d go over creating some simple Jupiter notebooks to make importing exporting data easier
Q@Dr. Furkan Gözükara
MFigured this out! I was setting the guidance scale to 1. 
https://diffusionmagic.web.app/ is working great now! If anybody wants to take ControlNet + Scribbling for a spin, give it a shot!
https://diffusionmagic.web.app/ is working great now! If anybody wants to take ControlNet + Scribbling for a spin, give it a shot!A new Flutter project.
SThankyou !!! I was able to do every step in your video, except I only downloaded only one model "control_sd15_mlsd.pth", but still don't have ControlNet tab on my web ui
SShould I have Xformers version? it says N/A for me
S@Dr. Furkan Gözükara
Fhey
NPlease help me, I have installed Python but the system still reports that it can't find it. I'm stuck here, can anyone with experience help me?
Fdid you install the extension and restart cmd?
Fit can work without xformers just slower
Fsure i can help
