Software Engineering Courses (SECourses)•3y ago

Hey guys. Thank you Furkan for your contributions! I've a question. I'm running Dreambooth training

Hey guys. Thank you Furkan for your contributions!
I've a question. I'm running Dreambooth trainings with the A111 SD extension. I've a Nvidia GeForce RTX 3060 12gb. For what I've seen from your videos, the card should be able to make 5-7 it/s right? I don't know if anyone here has the same card and does trainings...
I'm stuck in 2.5 it/s and can't go any faster, I've tried a lot of things. Do anyone here know what things I should test or check to see if there's any issue or that's the expected card speed?

AAlaBeat what.... now it worked, i haven't done anything, just stopped and started my pod...

Furkan Gözükara SECourses•5/3/23, 1:40 PM

possibly

Rrafraf another question. Batch size will affect how exactly the learning rate? I know o...

Furkan Gözükara SECourses•5/3/23, 1:42 PM

batch size makes things different. it changes how the model learns : https://www.baeldung.com/cs/epoch-vs-batch-vs-mini-batch

Baeldung on Computer Science

Differences Between Epoch, Batch, and Mini-batch | Baeldung on Comp...

Explore the differences between an epoch, a batch, and a mini-batch.

Furkan Gözükara SECourses•5/3/23, 1:42 PM

i think for teaching a face batch size 1 works best

JJoaco Hey guys. Thank you Furkan for your contributions! I've a question. I'm running...

Furkan Gözükara SECourses•5/3/23, 1:42 PM

it is for generating images not training

Furkan Gözükara SECourses•5/3/23, 1:42 PM

training slower

FFurkan Gözükara SECourses batch size makes things different. it changes how the model learns : https://www...

rafraf•5/3/23, 1:42 PM

Thanks. will read up on this

Furkan Gözükara SECourses•5/3/23, 1:43 PM

so instead of showing a picture at a time

Furkan Gözükara SECourses•5/3/23, 1:43 PM

changing weights

Furkan Gözükara SECourses•5/3/23, 1:43 PM

you show multiple images to model

Furkan Gözükara SECourses•5/3/23, 1:43 PM

change weight according to all

FFurkan Gözükara SECourses so instead of showing a picture at a time

dst19•5/3/23, 3:37 PM

Hi Furkan, saw your Lora dreambooth video and wondered if you could share the CKPT file of your trained facf

dst19•5/3/23, 3:37 PM

*face

dst19•5/3/23, 3:37 PM

I'd like to play around and see how photorealistic it is myself.

dst19•5/3/23, 3:38 PM

Refering to this video https://youtu.be/TpuDOsuKIBo

YouTubeSECourses

Generate Studio Quality Realistic Photos By Kohya LoRA Stable Diffu...

#Kohya SS web GUI DreamBooth #LoRA training full tutorial. You don't need technical knowledge to follow this tutorial. In this tutorial I have explained how to generate professional photo studio quality portrait / self images for free with Stable Diffusion training.

Our Discord server

https://bit.ly/SECoursesDiscord

If I have been of assis...

Ddst19 Hi Furkan, saw your Lora dreambooth video and wondered if you could share the CK...

Furkan Gözükara SECourses•5/3/23, 3:43 PM

My own face ckpt?

Furkan Gözükara SECourses•5/3/23, 3:44 PM

Why would you need that

Furkan Gözükara SECourses•5/3/23, 3:44 PM

By the way I did training realistic vision model

Furkan Gözükara SECourses•5/3/23, 3:44 PM

Download link already in gist file

Jcs•5/3/23, 4:36 PM

Hey! Love the tutorials. Im getting stuck at generating images in Automatic1111 getting "OutOfMemoryError: CUDA out of memory." tried a bunch of things but havent been able to solve it yet. Any ideas?

TheRustingDeadAdmin•5/3/23, 4:36 PM

means you dont have enough vram

Jcs•5/3/23, 4:37 PM

is the only real solution to use a better graphics card?

Jcs•5/3/23, 4:37 PM

or is there some settings i can use that help reduce the load on VRAM

TheRustingDeadAdmin•5/3/23, 4:37 PM

get a bigger/better card or use runpod/google collab. there are many techniques @Furkan Gözükara SECourses shows in his video many techniques to lower vram make sure to watch closely

Jcs•5/3/23, 4:38 PM

ah got it, thanks!

FFurkan Gözükara SECourses welcome what you are trying to install?

TheRustingDeadAdmin•5/3/23, 4:42 PM

Koyha ss from this video. https://www.youtube.com/watch?v=TpuDOsuKIBo&loop=0

Getting this error, I have tried many things using GPT 4 and cant get it to work.

Error no kernel image is available for execution on the device at line 167 in file D:\ai\tool\bitsandbytes\csrc\ops.cu

Error no kernel image is available for execution on the device at line 167 in file D:\ai\tool\bitsandbytes\csrc\ops.cu

I am using cuda 11.8, proper cudnn files. Card is a gtx 1070, fresh nvidia install.

TheRustingDeadAdmin•5/3/23, 4:42 PM

the installations succeeds, but training fails with this error

Hiccup•5/3/23, 5:34 PM

Could somebody help me with my vladmatic install i was having issues with diffusion storing to much vram so i deleted the venv to reset it but now im getting an issue with it not recognizing clip?

JJcs Hey! Love the tutorials. Im getting stuck at generating images in Automatic1111 ...

rafraf•5/3/23, 6:31 PM

Your resolution setting might be too high. Set it at 512x512 and try again

dave5555•5/3/23, 6:54 PM

is it possible to batch sd upscale a folder of images but with different prompts ? as if you were to load the image in to png info , send to img2img and sdupscale using prompt from png info , but in a batch process?