Software Engineering Courses (SECourses)•3y ago

sounds good ill downgrade before training again, i ended up finishing my model w/o xformers and it s

sounds good ill downgrade before training again, i ended up finishing my model w/o xformers and it still didnt take too long and the results are phenominal

CCelsius sounds good ill downgrade before training again, i ended up finishing my model w...

Furkan Gözükara SECourses•2/17/23, 10:59 AM

yes actually without xformers

Furkan Gözükara SECourses•2/17/23, 10:59 AM

it is better

Furkan Gözükara SECourses•2/17/23, 10:59 AM

xformers reduces quality

Furkan Gözükara SECourses•2/17/23, 10:59 AM

but you need like 24 gb vram for that

FFurkan Gözükara SECourses but you need like 24 gb vram for that

CelsiusOP•2/17/23, 11:00 AM

for real, it saturated most of my 4090's vram

FFurkan Gözükara SECourses Greetings everyone. In this video I am going to Show how to use Automatic1111 We...

bettercallaul•2/17/23, 11:11 AM

F for Respect, sir. thank you very much for your contribution to AI community

CCelsius for real, it saturated most of my 4090's vram

Furkan Gözükara SECourses•2/17/23, 11:25 AM

true. for sd 2.1 i had to enable xformers on runpod

Furkan Gözükara SECourses•2/17/23, 11:25 AM

currently post editing runpod tutorial

Furkan Gözükara SECourses•2/17/23, 11:25 AM

Bbettercallaul F for Respect, sir. thank you very much for your contribution to AI community

Furkan Gözükara SECourses•2/17/23, 11:25 AM

thank you so much for great comment. if you are able to become patron i would appreciate very much that

Manicipal•2/17/23, 11:45 AM

Firstly thanks for the top videos. Super useful. I'm a little confused by ControlNet, does anyone know how to access this via API yet?

Colored Tonalities•2/17/23, 11:51 AM

1girl, breathing_fire, burning, embers, explosion, fiery_hair, fiery_wings, fire, flame, lips, looking_at_viewer, molten_rock, moon, orange_theme, red_moon, red_sky, red_theme, solo, sun, sunset, torch 1girl, breathing_fire, burning, embers, explosion, fiery_hair, fiery_wings, fire, flame, lips, looking_at_viewer, molten_rock, moon, orange_theme, red_moon, red_sky, red_theme, solo, sun, sunset, torch

MManicipal Firstly thanks for the top videos. Super useful. I'm a little confused by Contro...

Furkan Gözükara SECourses•2/17/23, 12:08 PM

i think you can use their native script they have api there

Furkan Gözükara SECourses•2/17/23, 12:09 PM

https://youtu.be/YJebdQ30UZQ

YouTubeSoftware Engineering Courses - SE Courses

Transform Your Sketches into Masterpieces with Stable Diffusion Con...

Our Discord : https://discord.gg/HbqgGaZVmr. In this video, You will see the new amazing Stable Diffusion technology - #ControlNet. If I have been of assistance to you and you would like to show your support for my work, please consider becoming a patron on

https://www.patreon.com/SECourses

Playlist of #StableDiffusion Tutorials, Automatic111...

Omer•2/17/23, 12:10 PM

Hi I'm trying to use Shivam colab with a custom mode ( Realisticvision 1.3) from hugging face. It has all the diffusers data. In the colab, I can run the inference without problem and generate images, but when I try to train it exits with errors. I commented out the "fp16 revison" flag and changed the VAE parameter to " --pretrained_vae_name_or_path="SG161222/Realistic_Vision_V1.3/vae" "

I got "subprocess.CalledProcessError: Command '['/usr/bin/python3', 'train_dreambooth.py', '--pretrained_model_name_or_path=SG161222/Realistic_Vision_V1.3', '--pretrained_vae_name_or_path=SG161222/Realistic_Vision_V1.3/vae', '--output_dir=/content/Rv2_weights/zwx']' returned non-zero exit status 1.
"
Do you know what the issue is? Please see the screenshots for more details. Thanks!

OOmer Hi I'm trying to use Shivam colab with a custom mode ( Realisticvision 1.3) from...

Furkan Gözükara SECourses•2/17/23, 12:14 PM

they have only main branch

Furkan Gözükara SECourses•2/17/23, 12:14 PM

https://huggingface.co/SG161222/Realistic_Vision_V1.3/tree/main

Furkan Gözükara SECourses•2/17/23, 12:15 PM

change revision to main

Furkan Gözükara SECourses•2/17/23, 12:15 PM

and i dont know if your vae path correct

Omer•2/17/23, 12:17 PM

Ok thanks a lot!! so I just change the revision parameter to --revision="main" . This is the model " https://huggingface.co/SG161222/Realistic_Vision_V1.3/tree/main". Thank again!

OOmer Ok thanks a lot!! so I just change the revision parameter to --revision="main" ...

Furkan Gözükara SECourses•2/17/23, 12:18 PM

yep

Furkan Gözükara SECourses•2/17/23, 12:18 PM

also not sure about your vae path

Furkan Gözükara SECourses•2/17/23, 12:18 PM

did you watch this video? https://youtu.be/kIyqAdd_i10

YouTubeSoftware Engineering Courses - SE Courses

Stable Diffusion Google Colab, Continue, Directory, Transfer, Clone...

Our Discord : https://discord.gg/HbqgGaZVmr. This is the video where you will learn how to use Google Colab for Stable Diffusion. If I have been of assistance to you and you would like to show your support for my work, please consider becoming a patron on

https://www.patreon.com/SECourses

Playlist of Stable Diffusion Tutorials, #Automatic1111...

Михаил Кислов•2/17/23, 12:18 PM

Hi all. I am writing through google translator.

Are there ways to merge two lore files into one? Not 50/50, but 100/100 to make one big file.

Furkan Gözükara SECourses•2/17/23, 12:18 PM

i explain there path thing very well

ММихаил Кислов Hi all. I am writing through google translator. Are there ways to merge two lor...

Furkan Gözükara SECourses•2/17/23, 12:19 PM

with lora probably not possible. it is not able to add new tokens. therefore, whatever you do will be merged into same tokens

Furkan Gözükara SECourses•2/17/23, 12:19 PM

so if you make 1 of them 100%, the other one will be 0%

FFurkan Gözükara SECourses so if you make 1 of them 100%, the other one will be 0%

Михаил Кислов•2/17/23, 12:20 PM

Are there any alternative ways? Maybe not with the help of Laura, but with the help of other utilities.

FFurkan Gözükara SECourses did you watch this video? https://youtu.be/kIyqAdd_i10

Omer•2/17/23, 12:31 PM

One more question about the VAE path. I put the screenshot of the main branch which is at "SG161222/Realistic_Vision_V1.3" at hugging face. The main branch has a folder called "vae" with the vae files.

When I add the paramater --pretrained_vae_name_or_path="SG161222/Realistic_Vision_V1.3/vae" I get an error in the colab when the cell runs, about vae not found at path. Do I need to change the path to something else? Thanks again!!

OOmer One more question about the VAE path. I put the screenshot of the main branch wh...

Furkan Gözükara SECourses•2/17/23, 1:19 PM

i think upload vae to google drive

Furkan Gözükara SECourses•2/17/23, 1:19 PM

give its path

ММихаил Кислов Are there any alternative ways? Maybe not with the help of Laura, but with the h...

Furkan Gözükara SECourses•2/17/23, 1:20 PM

i dont think

FFurkan Gözükara SECourses i think upload vae to google drive

Omer•2/17/23, 2:38 PM

Hi, I created a new model in HuggingF and added the VAE files to the main folder. I kept the original model path and just changed the VAE path to my model. It seems to work now!

It seems HuggingF paths have to be in the format username/model_name and you can't add anything else to the HF path in colab!

Thanks for the help!

nezumichan•2/17/23, 3:30 PM

Hello! I have been trying and failing to downgrade cuda and xformers to get dreambooth to work, I followed all the steps in this video (https://www.youtube.com/watch?v=O01BrQwOd-Q&ab_channel=SoftwareEngineeringCourses-SECourses), and it seems like everything worked fined, but when I try to generate an image I get an error. Can anyone try to help me?

YouTubeSoftware Engineering Courses - SE Courses

8 GB LoRA Training - Fix CUDA Version For DreamBooth and Textual In...

Our Discord : https://discord.gg/HbqgGaZVmr. This video I am showing how to downgrade CUDA and xformers version for proper training and I am showing how to do LoRA training with 8GB GPU. If I have been of assistance to you and you would like to show your support for my work, please consider becoming a patron on

https://www.patreon.com/SECourse...

Manicipal•2/17/23, 4:09 PM

Is there a Web UI available for ControlNet? Without needing to do a local install. I need to test out some prompts.

0xAdam•2/17/23, 4:33 PM

Hey guys,

I only have 6gb vram while I am trying to run Dreembooth and it won't work ofc. Is there a method to use outside service like Google for vram work or smth?

Aurora 神名河•2/17/23, 4:56 PM

What is the best xformers version to use for now sir?

Aurora 神名河•2/17/23, 4:57 PM

@Dr. Furkan Gözükara is it still the 0.0.14 version?

alenknight•2/17/23, 5:41 PM

anyone know if there's a script in Auto1111 or method to change resolutions in a batch?

Lain•2/17/23, 6:50 PM

Hi All, about Stable Diffusion installed with A1111 AutoInstaller, appear i do not have CLIP Skip Option (Stop at last layers of clip model) and ETA. Did you know how to enable it ?

Bassna•2/17/23, 8:23 PM

Any idea why my ControlNet images look great in previews when generating, but once it finishes, it ruins the color and stuff? Left is the last preview image before finishing, right is the finished image

Bassna•2/17/23, 8:23 PM

Михаил Кислов•2/17/23, 8:29 PM

Hi all. I am writing through google translator.

Does anyone know how to increase the number of iterations per second? When teaching Lora. I have a 4080, but when training, it's a miserable ~2.5 it/sec. I think it can run faster.

ММихаил Кислов Hi all. I am writing through google translator. Does anyone know how to increas...

0xAdam•2/17/23, 8:30 PM

What's ur gpu?

0xAdam•2/17/23, 8:30 PM

Hey guys,

What are the best gpu's for ai generated art? I have only 6gb right now, can barely do anything.