but you need like 24 gb vram for that
but you need like 24 gb vram for that
Cfor real, it saturated most of my 4090's vram
BF for Respect, sir. thank you very much for your contribution to AI community
Ftrue. for sd 2.1 i had to enable xformers on runpod
Fcurrently post editing runpod tutorial
F
Fthank you so much for great comment. if you are able to become patron i would appreciate very much that
MFirstly thanks for the top videos. Super useful. I'm a little confused by ControlNet, does anyone know how to access this via API yet?
C1girl, 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
Fi think you can use their native script they have api there
FYouTubeSoftware Engineering Courses - SE Courses
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...
https://www.patreon.com/SECoursesPlaylist of #StableDiffusion Tutorials, Automatic111...
OHi 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!
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!
Fthey have only main branch
Fchange revision to main
Fand i dont know if your vae path correct
OOk 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!
Falso not sure about your vae path
FYouTubeSoftware Engineering Courses - SE Courses
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...
https://www.patreon.com/SECoursesPlaylist of Stable Diffusion Tutorials, #Automatic1111...
МHi 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.
Are there ways to merge two lore files into one? Not 50/50, but 100/100 to make one big file.
Fi explain there path thing very well
Fwith lora probably not possible. it is not able to add new tokens. therefore, whatever you do will be merged into same tokens
Fso if you make 1 of them 100%, the other one will be 0%
МAre there any alternative ways? Maybe not with the help of Laura, but with the help of other utilities.
OOne 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!!
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!!
Fi think upload vae to google drive
Fgive its path
Fi dont think
OHi, 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!
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!
NHello! 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
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...
https://www.patreon.com/SECourse...MIs there a Web UI available for ControlNet? Without needing to do a local install. I need to test out some prompts.
0Hey 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?
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?
AWhat is the best xformers version to use for now sir?
A@Dr. Furkan Gözükara is it still the 0.0.14 version?
Aanyone know if there's a script in Auto1111 or method to change resolutions in a batch?
LHi 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 ?
BAny 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
B
МHi 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.
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.
0What's ur gpu?
0Hey guys,
What are the best gpu's for ai generated art? I have only 6gb right now, can barely do anything.
What are the best gpu's for ai generated art? I have only 6gb right now, can barely do anything.
М16 гб.
0Increase this numbers
МI've tried it and the speed stays the same.(
0
0I dont know then 
0Lora is kinda bad I've heard. Try without it maybe
МSo far, it gives the best learning results. In about 1 hour, with "350" in the folder name (I don't know what this parameter is called), the object begins to be very recognizable.
HIts pretty powerful. Comes down to having a strong dataset and good captioning. That in itself takes a little time to get proficient.
Fplease show screenshot of your cmd where you did downgrade