FAn amazing new feature has been added to my Gradio captioning app.
Previously, it supported batch captioning.
Now, it has been enhanced to support batch…
Previously, it supported batch captioning.
Now, it has been enhanced to support batch…
ADoes the order of prompts matter?
Example1: big beautiful sun, bright sun
Example2: bright sun, big beautiful sun
Example1: big beautiful sun, bright sun
Example2: bright sun, big beautiful sun
EI still can't figure out my why my Dreambooth training is coming out with such bad results 
guess I'm going to watch the Kohya one & see if I can do that instead. No matter what I do I can't recreate how great my first LoRA training was - and it was off 60 images when I got a good result too, doesn't make any sense. Edit: Apparently my first LoRA was also not that great, I just looked at it again & with the LoRA not applied at all vs on, there's barely any difference LUL
guess I'm going to watch the Kohya one & see if I can do that instead. No matter what I do I can't recreate how great my first LoRA training was - and it was off 60 images when I got a good result too, doesn't make any sense. Edit: Apparently my first LoRA was also not that great, I just looked at it again & with the LoRA not applied at all vs on, there's barely any difference LULELike, the result is so bad that if I just do the person's name with no LoRA, the rendered face looks almost the same as if I use the LoRA, in a given checkpoint (and it doesn't look anywhere near close enough). So it's basically doing nothing. I'm just using them as an example, once I get one face to train successfully, I have quite a few people I want to train... but gotta get one to work first 
EWhich version of Dreambooth would y'all recommend using? Cuz right now I'm just using whatever the one was on the install list under Extensions, which the good Doctor already told me I should use an older version, just not sure which to get.
NIs there anyone who offers one-on-one training or a special support tier I can access to get help?
NI'm having a difficult time following the tutorials, I've got through the installation process...
Nbut the setup / training part... I've been stuck, going from video to video and getting no where.
NPlease, send me a DM with your rates.
NI don't want to waste anymore time, I want to get started.
Runfortunately, i can't pay for your patreon. is there any other way i could access the kaggle notebooks?
EQuestion, I'm working on getting TensorRT stuff working... where do I need to put the checkpoint for the SD VAE? That wasn't explained in the video lol
Edit: i figured it out
Edit: i figured it out
SDoes anybody know what the big difference is between training with diffusers vs kohya? It looks like diffusers is a dependency of kohya. One can do training in diffusers... So what is the purpose of kohya?
Ffirst ones more important
Fcurrently if you have GPU
Fsdxl 24 gb config
Fdreambooth
Fif you dont have a free kaggle SDXL
Fdreambooth
Fkohya is more advanced
Fi am giving 1 to 1 help
Fllava running locally auto installer coming
![Digital [Starburst]](https://cdn.discordapp.com/avatars/220961909447458817/8ecbbb3132fc6e44312f3d6d2ae3df69.webp?size=40)
DWhy would ControlNet not show up in the interface even though it is installed and activated? I've tried everything I can to get it to show up, and it simply won't.
DI've disabled, enabled, installed, uninstalled, reinstalled all to no avail. The extensions sections says it is installed and activated, but it isn't visible in the Automatic1111 ui nor are the Settings available.
![Digital [Starburst]](https://cdn.discordapp.com/avatars/220961909447458817/8ecbbb3132fc6e44312f3d6d2ae3df69.webp?size=16)
Fcheck cmd
DI suspect you already knew what was wrong based on previous discussions.
DIt was my own fault for screwing up the install.
DSuch is the learning curve.
Ahi, i just updated from 8gb to 16gb 4060ti. should i update my driver from 537 to 546.17? I've heard nvidia updates are whack.. please guide @Dr. Furkan Gözükara
DI take it I should not be using automatic Nvidia driver updates then?
DI thought when you installed CUDA it installed the latest driver automatically regardless of the version of CUDA you were installing.
DI'll be putting in a new card as well.
DI've had to install SD at least 5 or 6 times so far to get it right. No promises I'll get it right this time either.
DI'm using the Patreon installers as much as possible.
DFor instance the auto-installers Dr. Gozukara designed install special models that run with Tensor. Without understanding that, you can install it, but it won't be utilized if you do not install the models.
DI wonder how many are installing TensorRT without actually using it.
DSo should I downgrade my Nvidia driver to 537 or leave it at 546.17 as it is? The new video card I will be putting in is a 4070ti 12GB. I don't know if that's important or not.
DI appear to have fixed the ControNet issue by reinstalling everything.
DI need to test it however.
Fyes i would update
Fwith 16 GB you can do SDXL DreamBooth
Fuse latest version
DCan you run Dreambooth with 12 GB?
DTruthfully I didn't even know you could buy a 4060ti with 16 GB.
DThank you for answering my question about the driver.
DOne must not mess with SD settings when one does not know what one is doing.
DHmm...then Huggingface says, "Yes, it is possible to train DreamBooth on a 16 GB GPU using techniques like 8-bit Adam, fp16 training or gradient accumulation 1. However, fine-tuning the text encoder requires more memory, so a GPU with at least 24 GB of RAM is ideal.
DSo that would be a no. I cannot run Dreambooth with 12 GB of VRAM, or at least I can't expect anything better than pathetic performance.
