Software Engineering Courses (SECourses)•3y ago

I have access to 8x 3090 systems and can not get it to work 🙈 Such a shame as I have all this proc

I have access to 8x 3090 systems and can not get it to work

Such a shame as I have all this processing power

I'll compare the Kaggle tutorial step by step with my setup

FFurkan Gözükara SECourses Click to see attachment

Stefanoid9343•11/27/23, 3:30 PM

this is midjourney killer

FFadenB I have access to 8x 3090 systems and can not get it to work 🙈 Such a shame as ...

Furkan Gözükara SECourses•11/27/23, 3:31 PM

i can make it work

Stefanoid9343•11/27/23, 3:31 PM

no commercial license to pay?

Furkan Gözükara SECourses•11/27/23, 3:31 PM

with private consultation

SStefanoid9343 no commercial license to pay?

Furkan Gözükara SECourses•11/27/23, 3:31 PM

let me look licence

Furkan Gözükara SECourses•11/27/23, 3:31 PM

https://github.com/PixArt-alpha/PixArt-alpha/blob/master/LICENSE

GitHub

PixArt-alpha/LICENSE at master · PixArt-alpha/PixArt-alpha

Fast Training of Diffusion Transformer for Photorealistic Text-to-Image Synthesis - PixArt-alpha/PixArt-alpha

Stefanoid9343•11/27/23, 3:57 PM

it seems that is free commercial use

Stefanoid9343•11/27/23, 3:57 PM

SStefanoid9343 Click to see attachment

Furkan Gözükara SECourses•11/27/23, 4:04 PM

amazing

Stefanoid9343•11/27/23, 4:06 PM

no training on this new model yet right?

DR.Siva•11/27/23, 4:11 PM

Thanks to @Dr. Furkan Gözükara , Kaggle (for this wonderful free platform) , Hugginface( for huge space and repository)

DR.Siva•11/27/23, 4:14 PM

once this project is over , next will be animatediff (hopefully you come out with similar guide for us sooon :D)

DR.Siva•11/27/23, 4:44 PM

just found lower numbered lora dont even take the prompts at all..it just isnt trained or trained too much on the training images..

AIGambino•11/27/23, 5:12 PM

Will you have a tutorial on using esrgan models @Dr. Furkan Gözükara ?

DR.Siva•11/27/23, 5:44 PM

not having much fun with dreambooth checkpoints ..

GN anyway...

Ned Shoaei•11/27/23, 5:48 PM

Hi everyone, what I noticed is that my CN on SD RunPod is this version: v1.1.313, and it doesn't update to the higher version.

Ned Shoaei•11/27/23, 5:49 PM

I've tried everthing from tutorials but didn't work

FFurkan Gözükara SECourses which automatic111 are you using?

Ned Shoaei•11/27/23, 5:51 PM

I am using RunPod's Template

John Ceed•11/27/23, 6:07 PM

Stable Diffusion XL (SDXL) DreamBooth Training eror (((

John Ceed•11/27/23, 6:08 PM

File "/opt/conda/bin/accelerate", line 8, in <module>
sys.exit(main())
File "/opt/conda/lib/python3.10/site-packages/accelerate/commands/accelerate_cli.py", line 47, in main
args.func(args)
File "/opt/conda/lib/python3.10/site-packages/accelerate/commands/launch.py", line 977, in launch_command
multi_gpu_launcher(args)
File "/opt/conda/lib/python3.10/site-packages/accelerate/commands/launch.py", line 646, in multi_gpu_launcher
distrib_run.run(args)
File "/opt/conda/lib/python3.10/site-packages/torch/distributed/run.py", line 785, in run
elastic_launch(
File "/opt/conda/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in call
return launch_agent(self._config, self._entrypoint, list(args))
File "/opt/conda/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 250, in launch_agent
raise ChildFailedError(
torch.distributed.elastic.multiprocessing.errors.ChildFailedError:
=====================================================
./sdxl_train.py FAILED
-----------------------------------------------------
Failures:
<NO_OTHER_FAILURES>
-----------------------------------------------------
Root Cause (first observed failure):
[0]:
time : 2023-11-27_18:01:49
host : c4565fefbe4d
rank : 1 (local_rank: 1)
exitcode : -6 (pid: 1053)
error_file: <N/A>
traceback : Signal 6 (SIGABRT) received by PID 1053
=====================================================

mikemenders•11/27/23, 8:01 PM

I discover the hidden things of Prodigy. For example, if I don't have 25 pictures, but 21 or 26, I get more exciting results. For example, with 26 images, I had to lower the d_coef from 1.02 to 0.85, and even then Lora was a bit strong.
Even at 21 images it might be too strong if I set it to 4 repetitions instead of 5. So there are still some small things to explore, but the fact is that whether Lora is stronger or weaker, it still gives a nice result, as long as I fine tune its flexibility.

mikemenders•11/27/23, 8:18 PM

Here is a sample from SD 1.5 with 768x768 and without Hires Fix

01427-2719919611-RAW_photo_a_cvillaloboss_woman_in_beautifil_dress_sitting_in_a_chair_in_restaurant_portrait_looking_at_viewer__lora_cvillalo.png

mikemenders•11/27/23, 8:22 PM

I trained model with 512x512 images

Troubleman•11/27/23, 8:25 PM

Does anyone know the formula for number or regularization images? Something like repeats * train image count / batch size == reg images?

John Ceed•11/27/23, 8:25 PM

I can't use more than 13 training photos. error

JJohn Ceed I can't use more than 13 training photos. error

Troubleman•11/27/23, 8:27 PM

Including the error in your message might help

SStefanoid9343 no training on this new model yet right?

Furkan Gözükara SECourses•11/27/23, 10:03 PM

they have script but 0 info

DDR.Siva Thanks to @Dr. Furkan Gözükara , Kaggle (for this wonderful free platform) , Hug...

Furkan Gözükara SECourses•11/27/23, 10:03 PM

good job

AAIGambino Will you have a tutorial on using esrgan models @Dr. Furkan Gözükara ?

Furkan Gözükara SECourses•11/27/23, 10:04 PM

you just put them into folder test

Furkan Gözükara SECourses•11/27/23, 10:04 PM

but i plan to do

NNed Shoaei I am using RunPod's Template

Furkan Gözükara SECourses•11/27/23, 10:04 PM

can you tell me templat ename

JJohn Ceed File "/opt/conda/bin/accelerate", line 8, in <module> sys.exit(main()) Fil...

Furkan Gözükara SECourses•11/27/23, 10:04 PM

this doesnt tell the error

Furkan Gözükara SECourses•11/27/23, 10:04 PM

can you record video and send me from dm

TTroubleman Does anyone know the formula for number or regularization images? Something like...

Furkan Gözükara SECourses•11/27/23, 10:05 PM

repeat * train count

JJohn Ceed I can't use more than 13 training photos. error

Furkan Gözükara SECourses•11/27/23, 10:05 PM

you can use

FFurkan Gözükara SECourses can you tell me templat ename

Ned Shoaei•11/27/23, 11:22 PM

RunPod has two SD templates, one is fast, I am using the other one. it's webui 10.2.1

NNed Shoaei RunPod has two SD templates, one is fast, I am using the other one. it's webui 1...

Furkan Gözükara SECourses•11/27/23, 11:45 PM

ok i have web ui controlnet installer

Furkan Gözükara SECourses•11/27/23, 11:46 PM

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

DR.Siva•11/28/23, 12:33 AM

Just throwing an idea, how will be the outcome of Lora trained with base model used from the output of dreambooth..

DR.Siva•11/28/23, 1:42 AM

using extra weight to lora and ohwx woman only brings out better results for me(my trained lora) , giving atleast 1.1 to lora and ((ohwx woman)/ ohwx woman:2 gives better results

Le•11/28/23, 2:14 AM

Or of this is not even the top option we have now, and there is a better way to make it with even more quality

AIGambino•11/28/23, 2:31 AM

I have two custom models I trained in SDXL, I want to have them both in the same photo, how can I do this?

AIGambino•11/28/23, 2:47 AM

wow what causes this? these colors arent in my training images at all :\

Rae•11/28/23, 3:31 AM

any tool to mass rename images, or have an easy tool to easily adjust filenames with drag / drop filename crumbs?

RRae any tool to mass rename images, or have an easy tool to easily adjust filenames ...

GeekDadKevin•11/28/23, 3:59 AM

I use https://www.advancedrenamer.com/

Advanced Renamer - Free and fast batch rename utility for files and...

Batch rename files and folders in a snap. Perfect for any kind of file renaming including music and photo files. Preview your files before renaming and undo erroneous renaming jobs.

Advanced Renamer - Free and fast batch rename utility for files and...

Batch rename files and folders in a snap. Perfect for any kind of file renaming including music and photo files. Preview your files before renaming and undo erroneous renaming jobs.

Nora•11/28/23, 4:36 AM

What is the most recent video about using dreambooth in google colab without a laptop with gpu

Neuromasters•11/28/23, 9:18 AM

@Dr. Furkan Gözükara You've released a lot of new stuff! Now I just have to find the time to study it all)
I don't understand how classes work, I've been trying to figure it out for how long, I can't get it. Physically, when we train a large model on characters, we roughly divide into a man and a woman. And if on subjects? Then every class is a subject? Car, Chair, Door, bag or sheep, Dragon, whale

And then I haven't seen anyone use the indicators, and here is the Reg_1 folder, how the learning structure should look in it. How to write files in it and what should they be?

Original message was deleted

Furkan Gözükara SECourses•11/28/23, 10:48 AM

looks like that service D-ID and eleven labs

AAIGambino I have two custom models I trained in SDXL, I want to have them both in the same...

Furkan Gözükara SECourses•11/28/23, 10:49 AM

extract lora and do regional prompting

RRae any tool to mass rename images, or have an easy tool to easily adjust filenames ...

Furkan Gözükara SECourses•11/28/23, 10:50 AM

i would use chatgpt to get written easy python script

I have access to 8x 3090 systems and can not get it to work 🙈 Such a shame as I have all this proc

Similar Threads