Software Engineering Courses (SECourses)•3y ago

but, I uploaded 1700 reg images for 13 instance images and the slider at 100 but it generated anothe

but, I uploaded 1700 reg images for 13 instance images and the slider at 100 but it generated another 400 and something more class images. Is it because i set the number of sample to generate at ''2''?

Bbush but, I uploaded 1700 reg images for 13 instance images and the slider at 100 but...

Furkan Gözükara SECourses•3/10/23, 4:33 PM

maybe it wasnt able to read all your class images

Furkan Gözükara SECourses•3/10/23, 4:33 PM

because this doenst make sense

bushOP•3/10/23, 4:34 PM

oh yes! that could be it. right.

Zono50•3/10/23, 5:11 PM

in stable diffusion, is there a good extension or method to basically overlay one image on top of another image?

gatsby•3/10/23, 6:14 PM

Does anyone know if there’s a fast way to install a trained model into Jupiter? It’s currently estimated to take hours for me to load it in. Just wondering if there’s a faster way?

Ggatsby Does anyone know if there’s a fast way to install a trained model into Jupiter? ...

Furkan Gözükara SECourses•3/10/23, 6:22 PM

yes

Furkan Gözükara SECourses•3/10/23, 6:22 PM

runpodctl

Furkan Gözükara SECourses•3/10/23, 6:22 PM

check chapters

Furkan Gözükara SECourses•3/10/23, 6:22 PM

https://youtu.be/QN1vdGhjcRc

YouTubeSECourses - Software Engineering Courses

Ultimate RunPod Tutorial For Stable Diffusion - Automatic1111 - Dat...

Sign up RunPod: https://bit.ly/RunPodIO. Our Discord : https://discord.gg/HbqgGaZVmr. This is the Grand Master tutorial for running Stable Diffusion via Web UI on RunPod cloud services. 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...

FFurkan Gözükara SECourses check chapters

gatsby•3/10/23, 6:34 PM

Thank you. Am checking now and after you download CTL you install a model via a url. If I’m trying to upload a model from my desktop, do I simply upload it straight into models without using the terminal? And it will automatically upload through CTL?

Mr Q•3/10/23, 6:52 PM

@Furkan Gözükara SECourses , i have a question, i was training a model in dreambooth using v1-5-pruned.ckpt as source checkpoint. after 150 epochs i stopped it and wanted to test it out, so i clicked on generate ckpt file, it gives me a savetensors file. when i use that model in the text2img it is only able to create the thing i trained, it is not combined with v1-5-pruned.ckpt anymore, is that right?

Ma7adi805•3/10/23, 6:55 PM

assert not shared.cmd_opts.lowvram, 'Training models with lowvram not possible'
AssertionError: Training models with lowvram not possible

assert not shared.cmd_opts.lowvram, 'Training models with lowvram not possible'
AssertionError: Training models with lowvram not possible

Ma7adi805•3/10/23, 6:57 PM

how can I fix this

MMr Q @Furkan Gözükara SECourses , i have a question, i was training a model in dream...

Mr Q•3/10/23, 7:00 PM

or do i need to use checkpoint merger afterwards?

MMa7adi805 ```assert not shared.cmd_opts.lowvram, 'Training models with lowvram not possibl...

justalurker•3/10/23, 7:05 PM

Check out: https://github.com/AUTOMATIC1111/stable-diffusion-webui/blob/master/modules/hypernetworks/ui.py see if that's the fix you are looking for.

Ggatsby Thank you. Am checking now and after you download CTL you install a model via a ...

Furkan Gözükara SECourses•3/10/23, 7:19 PM

yes you can use runpodctl on your pc as in your pod

MMr Q @Furkan Gözükara SECourses , i have a question, i was training a model in dream...

Furkan Gözükara SECourses•3/10/23, 7:20 PM

not anymore ckpt. saved as safetensors

MMa7adi805 ```assert not shared.cmd_opts.lowvram, 'Training models with lowvram not possibl...

Furkan Gözükara SECourses•3/10/23, 7:20 PM

yes you cant train with lowvram

Furkan Gözükara SECourses•3/10/23, 7:20 PM

your options are cloud

Furkan Gözükara SECourses•3/10/23, 7:20 PM

runpod or google colab for example

MMr Q or do i need to use checkpoint merger afterwards?

Furkan Gözükara SECourses•3/10/23, 7:20 PM

you dont need

Jjustalurker Check out: https://github.com/AUTOMATIC1111/stable-diffusion-webui/blob/master/m...

Furkan Gözükara SECourses•3/10/23, 7:20 PM

textual inversion dont work with lowvram i tested :d

Furkan Gözükara SECourses•3/10/23, 7:21 PM

hyper networks i didnt though

Furkan Gözükara SECourses•3/10/23, 7:21 PM

but probably wont be as well

Furkan Gözükara SECourses•3/10/23, 7:21 PM

generating base avatar for animation tutorial

Furkan Gözükara SECourses•3/10/23, 7:21 PM

FFurkan Gözükara SECourses you dont need

Mr Q•3/10/23, 7:21 PM

so how would i fix it that after the training i am not able to genereate for example an apple, is it overtrained?

MMr Q so how would i fix it that after the training i am not able to genereate for exa...

Furkan Gözükara SECourses•3/10/23, 7:22 PM

i didnt understand your question?

Furkan Gözükara SECourses•3/10/23, 7:22 PM

also use checkpoint comparison which i explain

Furkan Gözükara SECourses•3/10/23, 7:22 PM

in my many videos

Mr Q•3/10/23, 7:23 PM

yess i did that, but all my checkpoints arent able to generate anything else than what it was trained on. but i did use v1-5-pruned ad base model in the training

FFurkan Gözükara SECourses i didnt understand your question?

Mr Q•3/10/23, 7:24 PM

so thats why i am questioning if i should merge the model or did i do something wrong earlier?

Darkestry•3/10/23, 7:25 PM

My stable diffusion model doesnt produce any images anymore

Darkestry•3/10/23, 7:25 PM

its all black when I want to produce a sample image

Darkestry•3/10/23, 7:25 PM

How can I fix this?

MMr Q so thats why i am questioning if i should merge the model or did i do something ...

Furkan Gözükara SECourses•3/10/23, 7:28 PM

No idea what did you do so can't tell :/

DDarkestry My stable diffusion model doesnt produce any images anymore

Furkan Gözükara SECourses•3/10/23, 7:28 PM

Do fresh install always best solution

Darkestry•3/10/23, 7:31 PM

Won't I lose my model this way?

DDarkestry Won't I lose my model this way?

Furkan Gözükara SECourses•3/10/23, 7:34 PM

Are you able to generate images with other models?

Darkestry•3/10/23, 7:35 PM

I am not

Darkestry•3/10/23, 7:35 PM

Is it a dreambooth issue then or automatic1111?

FFurkan Gözükara SECourses No idea what did you do so can't tell :/

Mr Q•3/10/23, 7:37 PM

I trained a LORA model with as base model 1,5pruned.

As class images i used Photo of Man
and i had 40 images of my self as training images called Quirijn
my sample prompt was photo of quirijn man
After 150 epochs the sample images looked really good.
I stopped the training and went to txt2img
As model i selected the most recent savetensors model from my training
When i generate an image with the prompt quirijn man, it looks really good,
but the prompt "an apple on a table" is not working, it only remembers quirijn man.

How can i fix this, should i merge the model with 1,5pruned?

Furkan Gözükara SECourses•3/10/23, 7:37 PM

You can save backup of your model always

Thermidor•3/10/23, 8:57 PM

Has anyone tried merging an inpainting model with a custom dreambooth model? Is it possible?

Thermidor•3/10/23, 8:59 PM

I read on Reddit that you can do it by setting model A = inpainting model , model B = custom model , model = base 1.5 model

Thermidor•3/10/23, 8:59 PM

Then merge as difference with strength = 1

Thermidor•3/10/23, 8:59 PM

However I get: RuntimeError: Error(s) in loading state_dict for LatentDiffusion:
size mismatch for model.diffusion_model.input_blocks.0.0.weight: copying a param with shape torch.Size([320, 9, 3, 3]) from checkpoint, the shape in current model is torch.Size([320, 4, 3, 3]).

Thermidor•3/10/23, 8:59 PM

when I tried to load this model

MMr Q I trained a LORA model with as base model 1,5pruned. As class images i used Pho...

Furkan Gözükara SECourses•3/10/23, 9:01 PM

that means overtrained

TThermidor Has anyone tried merging an inpainting model with a custom dreambooth model? Is ...

Furkan Gözükara SECourses•3/10/23, 9:01 PM

in painting model probably not

but, I uploaded 1700 reg images for 13 instance images and the slider at 100 but it generated anothe

Similar Threads