Software Engineering Courses (SECourses)•3y ago

I'm looking at this post: https://www.patreon.com/posts/best-stable-1-5-90307694 - but did you only

I'm looking at this post: https://www.patreon.com/posts/best-stable-1-5-90307694 - but did you only train it for 4 epochs and 1,000 steps?

Patreon

Best Stable Diffusion 1.5 Based Models Training Config - Parameters...

Get more from SECourses: Tutorials, Guides, Resources, Training, MidJourney, Voice Clone, TTS, ChatGPT, GPT, LLM, Scripts by Furkan Gözükara on Patreon

Lland23 I'm looking at this post: https://www.patreon.com/posts/best-stable-1-5-90307694...

Furkan Gözükara SECourses•10/11/23, 10:25 PM

this is in research

Lland23 @Dr. Furkan Gözükara so, what's the BEST dreambooth model you've ever trained?

Furkan Gözükara SECourses•10/11/23, 10:25 PM

my best is sdxl dreambooth

SStefanoid9343 Hi @Dr. Furkan Gözükara I have the following message on kaggle and I cannot type...

Furkan Gözükara SECourses•10/11/23, 10:25 PM

can you private message me with screenshot?

Furkan Gözükara SECourses•10/11/23, 10:25 PM

i will update notebook which one and error is where

land23OP•10/11/23, 10:27 PM

gotcha - so this I guess https://www.youtube.com/watch?v=EEV8RPohsbw

YouTubeSECourses

How To Do Stable Diffusion XL (SDXL) DreamBooth Training (Full Fine...

Patreon post for config files : https://www.patreon.com/posts/89213064

land23OP•10/11/23, 10:27 PM

tyty! Will check it out and report back

land23OP•10/11/23, 10:38 PM

interesting - surprised euler_a is the best sample sampler

land23OP•10/11/23, 10:39 PM

also, 1000 steps for 4 epochs right? How many images did you use and how long did the whole training test take btw?

land23OP•10/11/23, 10:39 PM

Also, does choosing 16 24 or 48GB affect the output? or just the speed? What is the tradeoff

Furkan Gözükara SECourses•10/11/23, 10:46 PM

twitter is gone lol

Lland23 also, 1000 steps for 4 epochs right? How many images did you use and how long di...

Furkan Gözükara SECourses•10/11/23, 10:46 PM

i use 13 images

Lland23 Also, does choosing 16 24 or 48GB affect the output? or just the speed? What is ...

Furkan Gözükara SECourses•10/11/23, 10:46 PM

16 vs 24 different

Furkan Gözükara SECourses•10/11/23, 10:46 PM

24 vs 48 same

Furkan Gözükara SECourses•10/11/23, 10:48 PM

added this to my patreon. fixed model deload and added batch caption

Furkan Gözükara SECourses•10/11/23, 10:48 PM

https://www.patreon.com/posts/sota-very-best-90744385

Stefanoid9343•10/11/23, 11:31 PM

@Dr. Furkan Gözükara I'm using runpod and it seems that the CND_runpod.py is for control net base on SD 1.5 and not XL fo you have any update?

SStefanoid9343 @Dr. Furkan Gözükara I'm using runpod and it seems that the CND_runpod.py is for...

Furkan Gözükara SECourses•10/11/23, 11:47 PM

you are right let me update 1 sec

SStefanoid9343 @Dr. Furkan Gözükara I'm using runpod and it seems that the CND_runpod.py is for...

Furkan Gözükara SECourses•10/11/23, 11:49 PM

ok i checked again and there are 47 models

Furkan Gözükara SECourses•10/11/23, 11:49 PM

https://www.patreon.com/posts/84896373

Furkan Gözükara SECourses•10/11/23, 11:50 PM

including sdxl too

Furkan Gözükara SECourses•10/11/23, 11:50 PM

which one you think missing?

Lalalaa•10/11/23, 11:51 PM

sir are you still following the roop/face fusion project? @Dr. Furkan Gözükara

FFurkan Gözükara SECourses ok i checked again and there are 47 models

Stefanoid9343•10/12/23, 12:22 AM

I’m want to use the depth lares

Stefanoid9343•10/12/23, 12:22 AM

it seems that is not on the list

Stefanoid9343•10/12/23, 12:22 AM

for xl

LLalalaa sir are you still following the roop/face fusion project? @Dr. Furkan Gözükara

Furkan Gözükara SECourses•10/12/23, 12:23 AM

yes but i didnt have chance to cover new one

SStefanoid9343 I’m want to use the depth lares

Furkan Gözükara SECourses•10/12/23, 12:23 AM

i think there is no depth leres for sdxl. do you know it is released?

Furkan Gözükara SECourses•10/12/23, 12:24 AM

model list here

Furkan Gözükara SECourses•10/12/23, 12:24 AM

https://github.com/Mikubill/sd-webui-controlnet/discussions/2039

GitHub

[Major Update] sd-webui-controlnet 1.1.400 · Mikubill sd-webui-cont...

The extension sd-webui-controlnet has added the supports for several control models from the community. Many of the new models are related to SDXL, with several models for Stable Diffusion 1.5. The...

Furkan Gözükara SECourses•10/12/23, 12:24 AM

https://huggingface.co/lllyasviel/sd_control_collection/tree/main

lllyasviel/sd_control_collection at main

FFurkan Gözükara SECourses i think there is no depth leres for sdxl. do you know it is released?

Stefanoid9343•10/12/23, 12:28 AM

I think there is not for XL

FFurkan Gözükara SECourses yes but i didnt have chance to cover new one

Lalalaa•10/12/23, 12:31 AM

how do I select a face from many faces in a photo target using headless mode? using --reference-face-position? and enter number?

im running in colab

LLalalaa how do I select a face from many faces in a photo target using headless mode? us...

Furkan Gözükara SECourses•10/12/23, 12:44 AM

i dont know try it

Mmikemenders This could also be an interesting solution, although I haven't played with it mu...

EyeSpyBekoAI•10/12/23, 12:46 AM

Hey @mikemenders , (or anyone else interested), gettign back on this track after being busy and sick for a while. I very much would like to try this. Having a really hard time with installation though. I tried both A1111 regular download and manually downloading the insightface package via cmd. Nothing seems to work. Any particular trick you came across?

Sandstormleader•10/12/23, 1:30 AM

the auto installers by @Dr. Furkan Gözükara work very well. i suggest joining his patreon and you just have a few clicks assuming your pc is up to par

andysingal•10/12/23, 2:22 AM

How AI agent assist from text to Video: https://www.youtube.com/watch?v=R1sqO1zUhNY

YouTubeAndyAI Insights

American Civil War Story using AI #schoollife #newtechnology #amer...

The American Civil War was a defining moment in American history. It was a war that tested the ideals of the American Revolution, and it ultimately led to the creation of a more perfect union. The war was fought over the issue of slavery, and it resulted in the deaths of over 600,000 people.

AI video generators can be used to create educational...

EEyeSpyBekoAI Hey @mikemenders , (or anyone else interested), gettign back on this track after...

EyeSpyBekoAI•10/12/23, 2:41 AM

@mikemenders , nvm, I figured it out. There needed to be an additional plugin to visual studio

SSandstormleader the auto installers by @Dr. Furkan Gözükara work very well. i suggest joining hi...

EyeSpyBekoAI•10/12/23, 2:43 AM

Hey @Sandstormleader , was this for me? If so, thanks for the response.
Yup, his auto installers are great! This was for a separate extension though that I was experimenting with. At least I don't think it's in the installers.

FFurkan Gözükara SECourses i dont know

Ajay Khrolia•10/12/23, 5:32 AM

it's not either for other folders as well which exists on system. I've moved some loras to other folder.

Metaphysix•10/12/23, 5:59 AM

Is there a way to save the huggingface cache on another drive? It takes up all my space in C

TomTekSol•10/12/23, 8:24 AM

Hello guys. I just got approval for the Sagemaker Studio account and soon after I started I am continually being logged out. Any help?

Benjamin•10/12/23, 8:25 AM

@Dr. Furkan Gözükara this might also be interesting for precise captioning https://llava-vl.github.io/

LLaVA

Visual Instruction Tuning

Benjamin•10/12/23, 8:27 AM

sigh, just trained via dreambooth extension just to notice that it's broken in it's latest version... https://github.com/d8ahazard/sd_dreambooth_extension/issues/1363

GitHub

[Bug]: AttributeError: 'AttnProcessor2_0' object has no attribute '...

Is there an existing issue for this? I have searched the existing issues and checked the recent builds/commits of both this extension and the webui What happened? Training a standard 768 model the ...

mikemenders•10/12/23, 8:30 AM

How meaningful is the Tensorboard loss value? Because out of curiosity, I just looked at the Lora model in the training set that you described as having the lowest loss, and it doesn't really look like the one I'm training. However, for the similarity and this one it says big loss.

Benjamin•10/12/23, 9:21 AM

are you also planning to experiment with SDXL LoRAs @mikemenders?