using adam optimizer?

FFurkan Gözükara SECourses using adam optimizer?

A

ashley•6/6/23, 5:35 PM

Yes

F

Furkan Gözükara SECoursesOP•6/6/23, 5:36 PM

also web ui version?

N

Nom•6/6/23, 5:36 PM

@ashleyk I just tried your template's kohya, it doesnt run lora training, did you try it?

FFurkan Gözükara SECourses also web ui version?

A

ashley•6/6/23, 5:36 PM

A1111? v1.3.2.

F

Furkan Gözükara SECoursesOP•6/6/23, 5:37 PM

no kohya ss

F

Furkan Gözükara SECoursesOP•6/6/23, 5:37 PM

https://github.com/bmaltais/kohya_ss

GitHub

GitHub - bmaltais/kohya_ss

Contribute to bmaltais/kohya_ss development by creating an account on GitHub.

NNom @ashleyk I just tried your template's kohya, it doesnt run lora training, did yo...

A

ashley•6/6/23, 5:37 PM

Gradio is broken, you have to check the logs unfortunately

N

Nom•6/6/23, 5:37 PM

you mean check the logs to troubleshoot?

FFurkan Gözükara SECourses no kohya ss

A

ashley•6/6/23, 5:38 PM

I am on v21.7.3

A

ashley•6/6/23, 5:38 PM

commit: 0865e7c342c1cc3cc7ab782e4bbf6c1f65d258f6

NNom you mean check the logs to troubleshoot?

A

ashley•6/6/23, 5:39 PM

To see what its doing, the web interface doesn't update for some reason, seems like a Gradio issue

cd /workspace/kohya_ss
source venv/bin/activate
fuser -k 3010/tcp
./gui.sh --listen 0.0.0.0 --server_port 3010

cd /workspace/kohya_ss
source venv/bin/activate
fuser -k 3010/tcp
./gui.sh --listen 0.0.0.0 --server_port 3010

cd /workspace/kohya_ss
source venv/bin/activate
fuser -k 3010/tcp
./gui.sh --listen 0.0.0.0 --server_port 3010

cd /workspace/kohya_ss
source venv/bin/activate
fuser -k 3010/tcp
./gui.sh --listen 0.0.0.0 --server_port 3010

Then you can see what it is doing

A

ashley•6/6/23, 5:40 PM

Gradio must be the worst Python project in history, it has endless bugs

P

peels•6/6/23, 5:40 PM

안녕하세요 유투브 보다가 왔어요.

P

peels•6/6/23, 5:40 PM

Hello, I came here after watching YouTube.

Ppeels Hello, I came here after watching YouTube.

A

ashley•6/6/23, 5:41 PM

Welcome

P

peels•6/6/23, 5:42 PM

Is there a way to optimize learning with one or a small amount of images?

P

peels•6/6/23, 5:43 PM

I'd like to learn the company logo, so what should I do?

F

frandagostino•6/6/23, 5:43 PM

Perhaps Textual Inversion to learn the concept?

P

peels•6/6/23, 5:45 PM

What is the number of images you need? And I want to apply the logo to my clothes. For example, Chanel logo ramen, Chanel t-shirt, Chanel sleeveless shirt, and Chanel dress. I want to use it in the direction of the logo.

Aashley To see what its doing, the web interface doesn't update for some reason, seems l...

N

Nom•6/6/23, 5:45 PM

heres the error

message.txt2.15KB

NNom heres the error

A

ashley•6/6/23, 5:46 PM

Some kind of bug in Kohya, I also got this and couldn't fix it, I think restarting the pod resolved it.

Ppeels What is the number of images you need? And I want to apply the logo to my clothe...

F

frandagostino•6/6/23, 5:49 PM

Check this guide: https://github.com/d8ahazard/sd_dreambooth_extension/discussions/547/

F

frandagostino•6/6/23, 5:50 PM

sorry not that one

F

frandagostino•6/6/23, 5:50 PM

This one:

F

frandagostino•6/6/23, 5:50 PM

https://civitai.com/models/56570/renata-valliulina?commentId=118383&modal=commentThread

F

frandagostino•6/6/23, 5:51 PM

It says it uses 18 images to train on a model of a person, I dont know how it could work on a logo

F

frandagostino•6/6/23, 5:51 PM

I think you can generate a small dataset with the logo in different situations and test

F

frandagostino•6/6/23, 5:53 PM

Full disclosure, I'm trying to train my first TI model, so mine is all theory

Aashley Some kind of bug in Kohya, I also got this and couldn't fix it, I think restarti...

N

Nom•6/6/23, 5:54 PM

You are right, restart worked but then I encounter this

make buckets
min_bucket_reso and max_bucket_reso are ignored if bucket_no_upscale is set, because bucket reso is defined by image size automatically / bucket_no_upscaleが指定された場合は、bucketの解像度は画像サイズから自動計算されるため、min_bucket_resoとmax_bucket_resoは無視されます
number of images (including repeats) / 各bucketの画像枚数（繰り返し回数を含む）
/workspace/kohya_ss/venv/lib/python3.10/site-packages/numpy/core/fromnumeric.py:3432: RuntimeWarning: Mean of empty slice.
  return _methods._mean(a, axis=axis, dtype=dtype,
/workspace/kohya_ss/venv/lib/python3.10/site-packages/numpy/core/_methods.py:190: RuntimeWarning: invalid value encountered in double_scalars
  ret = ret.dtype.type(ret / rcount)
mean ar error (without repeats): nan
No data found. Please verify arguments (train_data_dir must be the parent of folders with images) / 画像がありません。引数指定を確認してください（train_data_dirには画像があるフォルダではなく、画像があるフォルダの親フォルダを指定する必要があります）

make buckets
min_bucket_reso and max_bucket_reso are ignored if bucket_no_upscale is set, because bucket reso is defined by image size automatically / bucket_no_upscaleが指定された場合は、bucketの解像度は画像サイズから自動計算されるため、min_bucket_resoとmax_bucket_resoは無視されます
number of images (including repeats) / 各bucketの画像枚数（繰り返し回数を含む）
/workspace/kohya_ss/venv/lib/python3.10/site-packages/numpy/core/fromnumeric.py:3432: RuntimeWarning: Mean of empty slice.
  return _methods._mean(a, axis=axis, dtype=dtype,
/workspace/kohya_ss/venv/lib/python3.10/site-packages/numpy/core/_methods.py:190: RuntimeWarning: invalid value encountered in double_scalars
  ret = ret.dtype.type(ret / rcount)
mean ar error (without repeats): nan
No data found. Please verify arguments (train_data_dir must be the parent of folders with images) / 画像がありません。引数指定を確認してください（train_data_dirには画像があるフォルダではなく、画像があるフォルダの親フォルダを指定する必要があります）

make buckets
min_bucket_reso and max_bucket_reso are ignored if bucket_no_upscale is set, because bucket reso is defined by image size automatically / bucket_no_upscaleが指定された場合は、bucketの解像度は画像サイズから自動計算されるため、min_bucket_resoとmax_bucket_resoは無視されます
number of images (including repeats) / 各bucketの画像枚数（繰り返し回数を含む）
/workspace/kohya_ss/venv/lib/python3.10/site-packages/numpy/core/fromnumeric.py:3432: RuntimeWarning: Mean of empty slice.
  return _methods._mean(a, axis=axis, dtype=dtype,
/workspace/kohya_ss/venv/lib/python3.10/site-packages/numpy/core/_methods.py:190: RuntimeWarning: invalid value encountered in double_scalars
  ret = ret.dtype.type(ret / rcount)
mean ar error (without repeats): nan
No data found. Please verify arguments (train_data_dir must be the parent of folders with images) / 画像がありません。引数指定を確認してください（train_data_dirには画像があるフォルダではなく、画像があるフォルダの親フォルダを指定する必要があります）

make buckets
min_bucket_reso and max_bucket_reso are ignored if bucket_no_upscale is set, because bucket reso is defined by image size automatically / bucket_no_upscaleが指定された場合は、bucketの解像度は画像サイズから自動計算されるため、min_bucket_resoとmax_bucket_resoは無視されます
number of images (including repeats) / 各bucketの画像枚数（繰り返し回数を含む）
/workspace/kohya_ss/venv/lib/python3.10/site-packages/numpy/core/fromnumeric.py:3432: RuntimeWarning: Mean of empty slice.
  return _methods._mean(a, axis=axis, dtype=dtype,
/workspace/kohya_ss/venv/lib/python3.10/site-packages/numpy/core/_methods.py:190: RuntimeWarning: invalid value encountered in double_scalars
  ret = ret.dtype.type(ret / rcount)
mean ar error (without repeats): nan
No data found. Please verify arguments (train_data_dir must be the parent of folders with images) / 画像がありません。引数指定を確認してください（train_data_dirには画像があるフォルダではなく、画像があるフォルダの親フォルダを指定する必要があります）

NNom You are right, restart worked but then I encounter this ```make buckets min_buck...

A

ashley•6/6/23, 5:56 PM

Did you go to tools, configure the stuff, click "Prepare training data", and then "Copy into Folders Tab" before clicking "Train model" ?

Ppeels Hello, I came here after watching YouTube.

F

Furkan Gözükara SECoursesOP•6/6/23, 5:59 PM

hello welcome

Aashley Did you go to tools, configure the stuff, click "Prepare training data", and the...

N

Nom•6/6/23, 5:59 PM

that did it

Ppeels Is there a way to optimize learning with one or a small amount of images?

F

Furkan Gözükara SECoursesOP•6/6/23, 5:59 PM

how many. and why small number?

NNom that did it 🙂

A

ashley•6/6/23, 6:00 PM

Did you also disable xformers?

Aashley Did you also disable xformers?

N

Nom•6/6/23, 6:00 PM

no

do I have to? I need to do inference after

NNom no 😦 do I have to? I need to do inference after

A

ashley•6/6/23, 6:01 PM

Inference is in A1111, you need to disable xformers for training to get better quality

N

Nom•6/6/23, 6:01 PM

ok how do I do that?

NNom ok how do I do that?

A

ashley•6/6/23, 6:02 PM

Under "Training parameters" expand "Advanced Configuration" and UNCHECK "Use xformers"

A

ashley•6/6/23, 6:04 PM

It also defaults to "cosine" as "LR scheduler" but I seem to get better results using "constant_with_warmup"

F

frandagostino•6/6/23, 6:11 PM

Hi @Dr. Furkan Gözükara, how are you? What do you recommend for doing some Textual Inversion training experiments?
Colab (I'm paying 10 USD plan) or Runpod?
Locally I've a a Apple M2 8-cores 16GB, which I think is super slow for training.

Ffrandagostino Hi @Dr. Furkan Gözükara, how are you? What do you recommend for doing some Textu...

F

Furkan Gözükara SECoursesOP•6/6/23, 6:13 PM

do you have any particular reason for textual inversion

F

frandagostino•6/6/23, 6:15 PM

No, just that I've seen great models in Civitai, low size, and wanted to do some exploration

N

Nom•6/6/23, 6:21 PM

@Dr. Furkan Gözükara do you have any advise on which training method to use if I'm looking to get a sketch style like this:

hand-drawn-cat-outline-illustration_23-2149299744.jpg.jpg

E

Eggo•6/6/23, 6:34 PM

hey!! for anyone doing lora training with 1111 dreambooth extension, how are you getting txt2img to recognize it? for me now, it wont pickup on any Lora unless i use the "sd-webui-additional-networks" extension. although i believe the previous method with "a1111-sd-webui-locon" extension was more convenient (red icon at the top under generate)? (side note, when i train with dreambooth extension, it saves the lora as .pt files not safetensors. it only saves as safetensors if i check the option "generate lora weights for extra networks" , but then i cant use those in the prompt (i receive error)

L

Le•6/6/23, 6:43 PM

dreambooth is installed but it is not visible as a tab.

Original message was deleted

Q

qwerty_qwer•6/6/23, 6:44 PM

reload ui

L

Le•6/6/23, 6:44 PM

I did. Closed terminals, changed port. Nothing.

L

Le•6/6/23, 6:48 PM

does anyone know what to do here?

L

Le•6/6/23, 6:50 PM

and to be honest, I have dreambooth installed on my webui in my pc, and I dont see it in the tab either

L

Le•6/6/23, 6:51 PM

is it a python problem?

Similar Threads

Similar Threads

Similar Threads