How to install Stable Diffusion WebUI Colab Alternative (free)

Рет қаралды 51,414

Күн бұрын

Hello, everyone!
What’s happened to Stable Diffusion?
Where to find a colab notebook without disconnects?
How to use Stable Diffusion for free?
How to bypass the Stable Diffusion WebUI block?
What to use instead of Google Colab and Kaggle?
All these questions have become exceptionally relevant in the last weeks.
In this video, I will show you the most up-to-date method for using Automatiс1111 for free. Since Google Colab has blocked almost all possible ways to run Stable Diffusion (Automatic1111, ComfyUI, SwarmUI, etc), and Kaggle is following suit (my notebook with Automatic1111 still works), I suggest using Sagemaker Studio Lab, a free and even more convenient alternative! Plus, you'll have a daily usage quota of 4 hours without the need to reload models every time.
Please watch the video to the end and follow the instructions precisely to avoid any errors during setup and while using the notebook.
Let's go!
____________________________
Timestamps:
0:00 Intro
0:07 Quota info
1:27 How not to break everything
____________________________
🔥 Patreon with Ultimate notebook (LoRA, ControlNets, VAE) for Sagemaker: / marat_ai
🔗 Base Notebooks: drive.google.com/file/d/1Jek7...

Пікірлер: 193

@marat_ai 10 ай бұрын

IMPORTANT! If you encounter a 'No such file or directory' error, it's likely that you imported the notebook into the wrong folder. You should import the notebook into the 'home/studio-lab-user/' directory, as demonstrated in the video. After first start you can safely delete the 'sagemaker-studiolab-notebooks' folder. To start fresh and reinstall everything, go to the 'maintenance' tab in the notebook, delete all virtual environments and files, then restart the session and reinstall everything once again. A big thank you to @simunaqv for helping resolve this issue!

@marat_ai 9 ай бұрын

For those who are still experiencing problems with this method, you can try Stable Diffusion in Colab! The video is available here: kzfaq.info/get/bejne/otOgdNKhv5qVe2g.html

@soccerlover5048 9 ай бұрын

Hi. There is python 3.9.16 environment by default. Can you tell pls how to update python to 3.10.6 at least to run your notebook on it?

@buttmun7806 9 ай бұрын

It's working now, thanks for your response!

@kiraishagaming 9 ай бұрын

The studio lab won't let me to use GPU compute type

@CraftBlack 6 ай бұрын

Thanks!

@ronit043 9 ай бұрын

I've been waiting for someone to make a tutorial on this topic for months. Thank you very much!!!

@user-vz6lw6no3v 10 ай бұрын

Wow, this video saved me!

@TheDeerGeek 10 ай бұрын

Hey man, this is really cool, but I have a request Would it be possible that you can also add an Lora and VAE option to the notebook, and maybe and embeddings option

@megaaziib 9 ай бұрын

can u please make tutorial how to train lora sdxl with this? i want to train lora sdxl but my gpu is only 12gb vram, it still not enough for training lora sdxl.

@agnosticatheist4093 10 ай бұрын

Finally ❤

@thepayktis3356 9 ай бұрын

RuntimeError: "LayerNormKernelImpl" not implemented for 'Half'

@AmitPandey-bq6yl 6 ай бұрын

do we need to install a1111 first??

@vbtaro-englishchannel 9 ай бұрын

When I accessed SageMaker Studio Lab, I could only see 403 ERROR. There seem to be too much access. I don't know what to do.

@homenti2644 10 ай бұрын

Do you know how to download the resulting models from checkpoint manager

@japheth2316 9 ай бұрын

What do I have to do if I want to install stable defusion LX?

@RelaxLittleSouls 9 ай бұрын

quando eu vou fazer o cadrasto diz erro na pagina, la pedi empresa mas nao tenho empresa

@maiabuljeta5749 7 ай бұрын

I absolutely loved this wonderful video. Thank you so much. How do I xformers?

@shitpost_xxx 10 ай бұрын

what model you used on your thumbnail?

@johanjanjanitra 3 ай бұрын

Hi, 7:05 where can i download your ultimate version notebook for download lora, etc? Thanks.

@banglarclassroom3646 9 ай бұрын

Sir I am getting this: ModuleNotFoundError: No module named 'tqdm'. What should I do now

@sena.gcolab 4 ай бұрын

hi, how to update python to 3.10? is it possible? sagemaker still use python 3.9, which is make some extensions are not working :/

@user-jq6uf4by7m 9 ай бұрын

thanks ❤

10 ай бұрын

Is it possible to use trained models contained in dirve, putting the path to it in step two?

@tanjirou_yuki3mran135 9 ай бұрын

do you know how ? cuz 'Im suffering here

@AndreAndre-gw3rs 10 ай бұрын

thanks

@iiiComments 6 ай бұрын

Thanks, it still works, but I installed the reactor extension and it does not recognize it, I get an error

@Nekotico 8 ай бұрын

FileNotFoundError error, when i try to download a model, any clue?

@lucky-zu4me 20 күн бұрын

helpful!!!

@Nekotico 8 ай бұрын

u are a hero, btw , any discord community server?

@Angetyde 10 ай бұрын

hi..😊but what if I want to add deforum?

@muhammadhafid8466 10 ай бұрын

Can anyone help me? Everytime when i'm request account, it keep saying "Something went wrong. Please try again later."

@DecHoward4869 10 ай бұрын

Thanks, i got my amz account second. I will try to install it by following your steps in video later. Hopefully, everything will be running smoothly, can i add or train Lora by using this way?

@marat_ai 10 ай бұрын

yes, sure

@MohamedElGhazi-ek6vp 10 ай бұрын

thanks how much gpu ram they give you?

@Wilson-gj1fj 8 ай бұрын

Hey thanks for the video ! Do you know how can I install CivitiAI models in the notebook ? (because I have error with models that didn't come from huggingface) Thanks in advanced

@hydragons 2 ай бұрын

Go to your account settings and scroll down to the API Key. Make an API key. Copy it and save it somewhere safe. add that API key as your token at the end of the URL like ?token=Your-key-here

@henridusasca8077 4 ай бұрын

Hello, does it work for fooocus?

@user-pw3wc3pf1j 10 ай бұрын

Hello! I need help! In Step 3 "ImportError: libgthread-2.0.so.0: cannot open shared object file: No such file or directory'. How to fix it?

@marat_ai 10 ай бұрын

Hi, it seems that you used the previous version of the notebook. I have already fixed that, and the link is updated. Please nuke everything (command in maintenance), restart the runtime, and reinstall it once again.

@QHawk7 10 ай бұрын

*Great , Thank you man* , I have multiple google colab notebooks , I'm new to all this, how can I import them into kaggle or another service like sagemaker studio lab, or paperspace or else, and run them there, do I have to change certain code? what about files like mp3 or mp4 or pics that I used to use with my notebooks , either for video processing or picture enhancement or audio transcription , etc, how can I upload them to those alternatives? colab is not giving me access lately because of you know , "limitations" :/

@marat_ai 10 ай бұрын

Glad to help! In order to import Colab notebooks, you should modify the code because Kaggle and SageMaker environments work a bit differently. It might be easy or quite problematic, depending on the notebook and libraries used.

@QHawk7 10 ай бұрын

Yeah, True , I tried importing them , some worked, some I had to change some code, or add some code for them to work , but most of them don't work at all, I don't have much experience in coding ..but I think there would an AI app to convert google colabs into kaggle notebooks , until that , wish if developers start publish kaggle notebooks instead, since google is limiting too much these days and refusing to host some GPU intensive apps like midjourney alternatives, stable diffusion , large LLM models , they even banned a lot of accounts forever ...And they own kaggle too btw -_- // Thanks for you quick answer man. @@marat_ai

@bgmipubg2941 10 ай бұрын

There is no runtime available right now. (Problem on use GPU

@TikThoks-ct7df 9 ай бұрын

my step 2.1 of downloading model is not working it's showing error

@fernandodiaz8231 10 ай бұрын

Thanks for this contribution. It is a great example of collaboration and creativity. Before trying it, I want to ask you if, once the 4 hours of GPU or 8 hours of CPU are consumed, what can be done. Is a new account created?

@ineedmoneyto 10 ай бұрын

No it will reset every day

@marat_ai 10 ай бұрын

Thanks! As for your question, why not?I haven't had that before, usually, 4 hours are enough for me.

@IRDOnlineChannel 8 ай бұрын

I cant get OTP when start the project

@dragibus66 4 ай бұрын

it's work ! thanks but i don't know why it's take several times to install ans i lost my model somewhere ( edit even if i change the model i get always yours ! ) ! should i run it again to install it or i can download later on the interface ? anyway thanks it's work on my laptop who is currently poor for running stable without my external gpu when i am outside home

@blynkpham4369 7 ай бұрын

can you please do this with facechain?

@GenRevibes 7 ай бұрын

How to do this with fooocus

@andreabuzzolan9807 10 ай бұрын

Is it possible to use models loras ecc saved in google drive?

@marat_ai 10 ай бұрын

Technically, yes, but it would be easy to do this manually

@ivansmirnov2819 7 ай бұрын

I got this error now in your notebook: TypeError: __init__() got an unexpected keyword argument 'socket_options' Do you know how to solve it?

@Nostra189 6 ай бұрын

I got the same error

@RealPotato_01 10 ай бұрын

Hey, theres a bug going on that i cant get the phone verification code even after putting it right and trying differt phone number, do u know how to fix this??? Its been days for me already

@marat_ai 10 ай бұрын

I think it's because of your mobile carrier. I see the following options: use a different number from another country, preferably the US, by asking friends or kind people on Reddit. Alternatively, you can use number rental services, paid or free, but free ones are usually banned. So, it's better to pay half a dollar for a number

@user-qf6pr3rs4l 8 ай бұрын

i got a socket options error, pls help

@nunu9849 10 ай бұрын

better to go for gpu or cpu with notebook ?

@WisdomCart 5 ай бұрын

GPU (T4) is the better choice in this case.

@lifemarketing9876 10 ай бұрын

I signed up 2 days ago and I haven't heard back from them yet. How long does it take to get accepted?

@marat_ai 10 ай бұрын

I don't know, try a different email, preferably gmail, and consider using a VPN

@ivansmirnov2819 7 ай бұрын

GPU runtime almost always unavailable now in sagemaker

@user-hu3yj5gn9l 9 ай бұрын

Thanks for the share 🙌 If possible can you make a similar guide for RVC ( AI Cover songs ) on Sagemaker ??

@marat_ai 9 ай бұрын

Yes, that's possible, but definitely not in the next 2 weeks

@ILSDI667 9 ай бұрын

@@marat_aiwhy tho

@mharzmhason1787 3 ай бұрын

can't login in sagemaker studio lab anymore.

@joshua1846 5 ай бұрын

does that mean you can use it for 3 hours a day?

@santiagoaguilera6341 10 ай бұрын

Thanks man. Just a question. How can I load a Lora that I already have downloaded in my computer?

@marat_ai 10 ай бұрын

You are welcome! I would recommend creating a Hugging Face account, creating rep with your models and downloading them from hf. It's much more convenient and reliable.

@santiagoaguilera6341 10 ай бұрын

@@marat_ai Alright, I’ll do that. Thanks a lot

@Storieminime 10 ай бұрын

Hi, thanks for your guides. I'm not a programming expert, I wonder: Is it possible to use Kohya_ss to train a lora with your up-to-date method ?

@marat_ai 10 ай бұрын

sure

@nakul9282 7 ай бұрын

In step 3 this error showing - RuntimeError: Torch is not able to use GPU; add --skip-torch-cuda-test to COMMANDLINE_ARGS variable to disable this check. I am using CPU because GPU is not available.

@user-vw9lr1rk4l 6 ай бұрын

Without gpu it won't work

@oliviertorres8001 6 ай бұрын

Hello. Is it possible to import LORA with the regular process of Automatic1111? Or do we need your Patreon Notebook config file ? Thx

@marat_ai 6 ай бұрын

Hi! You can do it by yourself, but I've added that to the Patreon version. Just today, I've updated the notebook for Kaggle and SageMaker to the latest Automatic.

@oliviertorres8001 6 ай бұрын

@@marat_ai I made a mistake and instead of subscribing to SageMaker Studio Lab, I subscribed to SageMaker Studio (via AWS account). 2 mouths free with no limitation on storage (100 Go max). No pb to launch the JupyterLab, but your script doesn't work because the root name of the path changed between these two environments. Is there a chance I can fix that by my own ?

@ngokhongbmt 9 ай бұрын

guide me to do with facefusion

@nunu9849 10 ай бұрын

i cant download a model...its says "FileNotFoundError: [Errno 2] No such file or directory"

@buttmun7806 10 ай бұрын

same here :/

@greatjensen 10 ай бұрын

Me as well

@marat_ai 10 ай бұрын

The issue is resolved, please check the pinned comment

@kumar0758 9 ай бұрын

Bro how we can run this mobile

@sayansarkarz5414 10 ай бұрын

Can I not use links from Civitai to download the models?

@marat_ai 10 ай бұрын

Yes, sure

@sayansarkarz5414 10 ай бұрын

It's not working. I tried to use links of Realistic Vision and Magicmix but when I launched Stable Diffusion, it shows a single random model named "v1-5-pruned-emaonly.safetensors [6ce0161689]" which also generated horrible results. It's weird.@@marat_ai

@BB-gs4cb 9 ай бұрын

Hello, how can I download the automatic_segamaker base mentioned in 2:42?

@marat_ai 9 ай бұрын

Check the description

@BB-gs4cb 9 ай бұрын

@@marat_ai I have downloaded the file at the bottom of your description, (Link at "Base Notebooks:"). But the name of file is "Base_SD_SageMaker notebook03.ipynb", not the one in your video "automatic Sagemaker Base". And later it does not work. Would you please give me more instruction? Thank You!

@marat_ai 9 ай бұрын

it's the same notebook, probably recently updated, and it works. You can specify your problem, but most likely, you are not alone, and your issue has been resolved. Just check the comments.

@BB-gs4cb 9 ай бұрын

@@marat_ai Thank you! It works!

@digitalpabs 10 ай бұрын

is this working with sdxl?

@marat_ai 10 ай бұрын

yes

@epicgameryash9795 9 ай бұрын

Is it still working??

@marat_ai 9 ай бұрын

sure

@user-ps2wd3vo5d 10 ай бұрын

That's great! It's much more convenient to use than Google. However, I have a problem bothering me. I have generated a large number of images, and I'm wondering how to download them quickly and easily. When I use the Chrome browser, I can only download them 10 by 10, and it fails if there are too many.

@marat_ai 10 ай бұрын

You are welcome! As for you question, I understand you pain, just create new cell and add the code bellow, then call the function # Import the 'zipfile' module to work with ZIP archives. import zipfile # Define a function called 'create_zip_archive' that takes two parameters: # - 'source_dir': The directory whose contents will be zipped. # - 'output_path': The path where the resulting ZIP archive will be saved. def create_zip_archive(source_dir, output_path): with zipfile.ZipFile(output_path, 'w', zipfile.ZIP_DEFLATED) as zipf: for root, dirs, files in os.walk(source_dir): for file in files: file_path = os.path.join(root, file) zipf.write(file_path, os.path.relpath(file_path, source_dir)) # Define the source directory to be zipped. source_dir = '/home/studio-lab-user/stable-diffusion-webui/outputs' # Define the output path for the resulting ZIP archive. output_path = '/home/studio-lab-user/outputs.zip' # Call the 'create_zip_archive' function to create the ZIP archive. create_zip_archive(source_dir, output_path)

@user-ps2wd3vo5d 10 ай бұрын

After I performed the operation, this block of code appeared. Did I do something wrong? NameError Traceback (most recent call last) /tmp/ipykernel_91/3186569932 .py in () 21 22 # Call the 'create_zip_archive' function to create the ZIP archive. ---> 23 create_zip_archive(source_dir, output_path) /tmp/ipykernel_91/3186569932 .py in create_zip_archive(source_dir, output_path) 9 def create_zip_archive(source_dir, output_path): 10 with zipfile.ZipFile(output_path, 'w', zipfile.ZIP_DEFLATED) as zipf: ---> 11 for root, dirs, files in os.walk(source_dir): 12 for file in files: 13 file_path = os.path.join(root, file) NameError: name 'os' is not defined

@marat_ai 10 ай бұрын

@@user-ps2wd3vo5d add import os

@denisalirezai4853 9 ай бұрын

There is no GPU available. Tried it some days now. Everytime: "There is no runtime available right now. Please change the compute type or try again later".... running stable diff with CPU would be nice xD

@epicgameryash9795 9 ай бұрын

You are right. Its everytime there isn't available GPU problem

@embers89 7 ай бұрын

hello! can you update the notebook please?

@CRIMELAB357 7 ай бұрын

I don't think dreambooth works on collab again

@kenttom 9 ай бұрын

never get to use GPU . always have to wait .

@11_amankumarmall14 Ай бұрын

Hey marat bro can you also make a notebook for lightning ai new one same as sagemaker but more than that

@marat_ai Ай бұрын

Hi! Yes, I knew about Lightning and have already done some tests. It's a really good way to go. I've been quite busy making videos and replying to comments, but I'm going to make a new one soon!

@11_amankumarmall14 Ай бұрын

@@marat_ai thanks

@torodinson5260 10 ай бұрын

cant download model (No such file or directory)

@marat_ai 10 ай бұрын

Did you use your link or the one I already specified (realistic vision)? Most likely, you used the wrong link. The link should be for downloading the model, not just a model's webpage.

@torodinson5260 9 ай бұрын

First I tried to download your models, it didn't work. then I changed the link but still the other model doesn't download either. @@marat_ai

@catdecode 10 ай бұрын

bro u forgot to switch to GPU when run SD ?

@marat_ai 10 ай бұрын

It's possible, there were many video clips during editing, so I might have mixed up the fragments. But of course, for SD you should use GPU, and as I mentioned in the video, since the notebook is already prepared

@thekingmultiversal 5 ай бұрын

This is no longer the best option as NGROK has set a limit of 1GB, which can be completely used in a few minutes after generating some high-resolution images. After that, the person will have to wait a month, which is worse than waiting 24 hours.

@marat_ai 5 ай бұрын

You are free to use any tunnel you want, and I think that having several ngrok accounts is also possible.

@sena.gcolab 4 ай бұрын

yeah you're right, just using it for several times, and ngrok stopped working. do you have any alternate tunnel recommendation?

@SixtyScarySeconds 21 күн бұрын

Ngrok is notorious for disconnection.

@alisyr45 10 ай бұрын

Where is the free and final notebook that requires a $5 subscription

@marat_ai 10 ай бұрын

Sorry, I just fixed that. The notebook is now available.

@supereditor236 10 ай бұрын

Give the free link also bro pls

@marat_ai 10 ай бұрын

The notebook on my patreon page is already free, just look carefully

@supereditor236 10 ай бұрын

@@marat_aithank you bro ❤

@dbmmf1770 3 ай бұрын

very nice how you leave everything out with credit card. 😂

@simunaqv 10 ай бұрын

I was getting error with the targetr_dir value set to "/home/studio-lab-user/stable-diffusion-webui/models/Stable-diffusion/". I fixed it by changing it to "/home/studio-lab-user/sagemaker-studiolab-notebooks/stable-diffusion-webui/models/Stable-diffusion/".

@marat_ai 10 ай бұрын

Damn, I got it! You had this error because you, and other guys who have the same problem here, imported the notebook in the wrong folder, the standard one 'sagemaker-studiolab-notebooks,' which I didn't use in the video. This is why it works for some of you and doesn't for others. Just import the notebook to the 'sagemaker-studiolab-notebooks' folder. Would be even better if you just delete everything and keep the 'studio-lab-user' folder clean before adding my notebook, what I actually did in the video

@dragibus66 3 ай бұрын

no answer !

@ProjektRaven Ай бұрын

6:30 "heavily restricted with available storage" it give me 25gb, while google drive only 15gb 🤣

@marat_ai Ай бұрын

It's not as much because all the environment stuff already takes up about 10 -15 GB! But 15 gigs on Colab is already something, like 64 GB in a base iPhone.

@roughedge-machineworks 9 ай бұрын

ehm... just install it locally and dont bother with the crappy online options?

@marat_ai 9 ай бұрын

Hmm, there are a bunch of people who just don't have a powerful GPU, so this video is for them...

@jerchongkong5387 9 ай бұрын

dude, not everyone lives in first world countries with access to low priced GPUs.

@roughedge-machineworks 9 ай бұрын

@@jerchongkong5387 can't say i do either. But wasting money on cloud resources when you can just saved up on for one is kinda stupid.

@blazz 10 ай бұрын

When I press "Start Runtime" it asks for my phone number then verification code. But i never get the verification code.

@marat_ai 10 ай бұрын

try to spam, I had the same issue and got sms from GoPro for some reason

@RealPotato_01 10 ай бұрын

@@marat_aii still havent got the code do u know how to fix???

@erickhc 10 ай бұрын

I keep getting "[Errno 2] No such file or directory: '/home/studio-lab-user/stable-diffusion-webui'" when downloading models and trying to run the third cell.

@marat_ai 10 ай бұрын

If you run all cells properly as shown in the video, everything should work. Try reinstalling everything once again

@thinker80 10 ай бұрын

@@marat_ai there is some error in the code

@erickhc 10 ай бұрын

@@marat_ai First of all, thanks for replying, amazing job.

@marat_ai 10 ай бұрын

@@erickhc Thanks for the response. You've shown that it works! Because there are many people who claim that it doesn't work, I tested it myself several times, and it also worked for me. I faced a similar issue when I was creating this notebook. That's why I emphasized multiple times that everything should be done carefully. Sagemaker is indeed quite strange.

@Magentochile 9 ай бұрын

Si les sale este problema 👉🏼 "NotImplementedError: No operator found for `memory_efficient_attention_forward` with inputs: query : shape=(1, 4096, 1, 512) (torch.float32) key : shape=(1, 4096, 1, 512) (torch.float32)..." cambiar "--xformers" por "--opt-sub-quad-attention" en el paso 3 y listo! working 🤗

@tiggersaldy6237 6 ай бұрын

@Magentochile - did you resolve the "NotImplementedError"?

@White_Soup 10 ай бұрын

Hey man, this is really cool, but I have a request Would it be possible that you can also add an Lora and VAE option to the notebook, and maybe and embeddings option