Software Engineering Courses (SECourses)•3y ago

should I try reduce my training images further? (from 50 to 10?) and uncheck "text_encoder_learner"

should I try reduce my training images further? (from 50 to 10?) and uncheck "text_encoder_learner"? (where do I find this?) feel like I am missing something fundamental...

Hhozaggers should I try reduce my training images further? (from 50 to 10?) and uncheck "t...

Deleted User•12/6/23, 1:30 PM

13 with 1024,1024 working good

Hhozaggers should I try reduce my training images further? (from 50 to 10?) and uncheck "t...

Deleted User•12/6/23, 1:31 PM

As you are using 512,512 maybe more images around 15 to 20 will also work

Bbaloooo 1st. I create the payload with UI 2nd. I work on improving the payload, iterativ...

andrsoe•12/6/23, 1:31 PM

Im really struggling to explain my self, for that i am sorry, but:
-You describe the workflow, super nice!
-But are you saying if you could just use the UI you would hit 100% correct generations, and you are "fighting the results you are getting from UI vs the json payload"?

DDeleted User As you are using 512,512 maybe more images around 15 to 20 will also work

hozaggersOP•12/6/23, 1:32 PM

thanks, is there anyway to increase the timeout or is this a limitation by Kaggle? ideally I want to use all my training data

Hhozaggers thanks, is there anyway to increase the timeout or is this a limitation by Kaggl...

Deleted User•12/6/23, 1:32 PM

It’s bug maybe they will solve the problem in near future

Hhozaggers thanks, is there anyway to increase the timeout or is this a limitation by Kaggl...

Deleted User•12/6/23, 1:33 PM

You are doing 512,512 training

hozaggersOP•12/6/23, 1:34 PM

yes I am doing 512x512

Hhozaggers yes I am doing 512x512

Deleted User•12/6/23, 1:35 PM

Try 25 images

Deleted User•12/6/23, 1:35 PM

Or 20

hozaggersOP•12/6/23, 1:35 PM

yep I'm trying 10, just to confirm reducing images helps the problem

Deleted User•12/6/23, 1:35 PM

When they give you caching time

Deleted User•12/6/23, 1:36 PM

Catching must be finish with in 30 minutes

Deleted User•12/6/23, 1:36 PM

I have did 1024,1024 with 15 images too

Deleted User•12/6/23, 1:37 PM

13 is best for 1024,1024

Deleted User•12/6/23, 1:37 PM

Maybe with 512 more imgaes will also do the job for around 20

hozaggersOP•12/6/23, 1:37 PM

will kaggle premium give more caching time? or is it a bug which will probably appear there too?

Hhozaggers will kaggle premium give more caching time? or is it a bug which will probably a...

Deleted User•12/6/23, 1:38 PM

I have never used premium

Aldo Raine•12/6/23, 3:02 PM

the cache time issue will be resolved if you just start running again right after the 30 min

virginia•12/6/23, 3:03 PM

Furkan

Aldo Raine•12/6/23, 3:03 PM

That is what @Dr. Furkan Gözükara told me and it worked for me:
Furkan Gözükara SECourses — 11/30/2023 12:16 PM
ok easy to fix
click start training again
it will cache remaining files

virginia•12/6/23, 3:03 PM

seeming huggingface changed, i tried to upload

virginia•12/6/23, 3:05 PM

but yousee all models come to this place

AAldo Raine Yes 29GB max

Furkan Gözükara SECourses•12/6/23, 3:29 PM

load it from hugging face

virginia•12/6/23, 3:29 PM

it seems uploading to branch, i do not know how to download it from branch

AAIGambino Whats going on with this yolo script

Furkan Gözükara SECourses•12/6/23, 3:29 PM

i think it is about meta data. you message me about this from patreon?

virginia•12/6/23, 3:30 PM

yes i guess so, let me send it to you from patreon

7740 I have already installed all missing custom nodes from that tab, but still missi...

Furkan Gözükara SECourses•12/6/23, 3:30 PM

sadly i am not that expert of comfyui

virginia•12/6/23, 3:30 PM

no does not matter with comfyui

virginia•12/6/23, 3:30 PM

only matters with uploading to huggingface

Hhozaggers does anyone know what might be causing the training to fail?

Furkan Gözükara SECourses•12/6/23, 3:31 PM

yes this happens when caching takes too long

Furkan Gözükara SECourses•12/6/23, 3:31 PM

reduce number repeat

Furkan Gözükara SECourses•12/6/23, 3:31 PM

or start process again

Furkan Gözükara SECourses•12/6/23, 3:31 PM

it will continue caching

EElisoli on magic animate i get this error and it stuck like this : pipeline_animation.p...

Furkan Gözükara SECourses•12/6/23, 3:31 PM

it is working just wait

virginia•12/6/23, 3:32 PM

i need it to be seen on this page

virginia•12/6/23, 3:33 PM

now all here

virginia•12/6/23, 3:34 PM

in the past it is okay, now changed

virginia•12/6/23, 3:39 PM

you see, checkpoint is always about 7 g, without uploading to huggingface, the training is useless

lucy•12/6/23, 3:44 PM

@Dr. Furkan Gözükara have you had any good results with magic animate yet? all I've tried was quite disappointing haha

lucy•12/6/23, 3:44 PM

I've tried very slow movement too and roughly same stature of characters etc

lucy•12/6/23, 3:44 PM

it's nowhere near as good as the repo implies it is

Llucy @Dr. Furkan Gözükara have you had any good results with magic animate yet? all I...

Furkan Gözükara SECourses•12/6/23, 3:53 PM

same :d

Furkan Gözükara SECourses•12/6/23, 3:53 PM

looks like another fake research paper

lucy•12/6/23, 3:57 PM

maybe animate anything will be slightly better, because it doesnt use this weird area approach and uses actual posenet

DDR.Siva any special argument ? keywords ? etc to be used ? Or ohwx / bbuk woman and man ...

mtnmecca•12/6/23, 4:00 PM

use the special tokens you used in training. For example main prompt could be: ohwx man standing next to bbuk woman on a street corner. In the adetailer prompt (using face yolov8n) use: ohwx man <ohwx man lora> [SEP] bbuk woman <bbuk woman lora>. If it swaps faces then reverse this command. Make sure to set detection K of 2 so it does not swap other faces if there are more than 2. If you had a ckpt trained on both then you wouldn't have to use the <lora> portion. Either way works

FFurkan Gözükara SECourses load it from hugging face

Aldo Raine•12/6/23, 4:07 PM

Isn't using this loading it from huggingface?
stablediffusionapi/juggernaut-xl-v7

Mmtnmecca use the special tokens you used in training. For example main prompt could be: ...

DR.Siva•12/6/23, 4:18 PM

ok will try. its safetensor , not ckpt

.
I still have 1024 pictures of both of us. is it possioble to put both of our photos and train ? If so what would be the regularization pictures ? it should be about couples right ?

DR.Siva•12/6/23, 4:18 PM

@Dr. Furkan Gözükara some guidence would be great.

mtnmecca•12/6/23, 4:19 PM

sorry I meant the same thing: safetensor is a version of a ckpt. i use safetensor on everything

AAldo Raine Isn't using this loading it from huggingface? stablediffusionapi/juggernaut-xl-v...

Furkan Gözükara SECourses•12/6/23, 4:22 PM

yes it is

should I try reduce my training images further? (from 50 to 10?) and uncheck "text_encoder_learner"

Similar Threads