Software Engineering Courses (SECourses)•3y ago

also should we then use Text Encoder training or not? kept the suspence going on that twit 🙂

also should we then use Text Encoder training or not? kept the suspence going on that twit

ElisoliOP•10/8/23, 1:23 PM

looks pretty identical

Furkan Gözükara SECourses•10/8/23, 1:23 PM

this one is amazing

EElisoli also should we then use Text Encoder training or not? kept the suspence going o...

Furkan Gözükara SECourses•10/8/23, 1:23 PM

i think looks like not very necessary

Furkan Gözükara SECourses•10/8/23, 1:23 PM

but can be used too :d

rafraf•10/8/23, 1:24 PM

@Dr. Furkan Gözükara does NOT-raining text encoder save memory and speed? Is it worth turning off TE learning?

rafraf•10/8/23, 1:25 PM

Oh just noticed someone asked the same question

Rrafraf @Dr. Furkan Gözükara does NOT-raining text encoder save memory and speed? Is it ...

Furkan Gözükara SECourses•10/8/23, 1:25 PM

it is turned off by default

Furkan Gözükara SECourses•10/8/23, 1:25 PM

it increases vram slighty and reduces training speed slightly

rafraf•10/8/23, 1:26 PM

Also, where do you explain the block training --block_lr 1e-5,1e-5,1e-5,1e-5,1e-5,1e-5,1e-5,1e-5,1e-5,1e-5,1e-5,1e-5,1e-5,1e-5,1e-5,1e-5,1e-5,1e-5,1e-5,1e-5,1e-5,1e-5,1e-5

ElisoliOP•10/8/23, 1:26 PM

I had Very good success using control net Depth, Canny and openpose to upscale and bring out details in an old photo from this

Rrafraf Also, where do you explain the block training --block_lr 1e-5,1e-5,1e-5,1e-5,1e...

Furkan Gözükara SECourses•10/8/23, 1:26 PM

use config from patreon

FFurkan Gözükara SECourses it increases vram slighty and reduces training speed slightly

rafraf•10/8/23, 1:26 PM

Ok good to know

EElisoli I had Very good success using control net Depth, Canny and openpose to upscale a...

Furkan Gözükara SECourses•10/8/23, 1:26 PM

nice

ElisoliOP•10/8/23, 1:26 PM

I wish there was a better Canny model for detecting edges

ElisoliOP•10/8/23, 1:26 PM

had to mess with it a lot

EElisoli I wish there was a better Canny model for detecting edges

Furkan Gözükara SECourses•10/8/23, 1:27 PM

ye we need better sdxl controlnet

ElisoliOP•10/8/23, 1:27 PM

also waiting for QR control net for sdxl, I cant seem to properly upscale 1.5 images from 512 to higher ones without losing details

FFurkan Gözükara SECourses those models are nasty

Stefanoid9343•10/8/23, 1:34 PM

Thank you so much!

Stefanoid9343•10/8/23, 1:39 PM

@Dr. Furkan Gözükara this is the new one that replace the real vision one? https://youtu.be/AY6DMBCIZ3A?feature=shared can I do it with runpod?

YouTubeSECourses

First Ever SDXL Training With Kohya LoRA - Stable Diffusion XL Trai...

Updated for SDXL 1.0. How to install #Kohya SS GUI trainer and do #LoRA training with Stable Diffusion XL (#SDXL) this is the video you are looking for. I have shown how to install Kohya from scratch. The best parameters to do LoRA training with SDXL. How to use Kohya SDXL LoRAs with ComfyUI. How to do checkpoint comparison with SDXL LoRAs and m...