Home Creators Posts Import Register Favorites Logout

Downloads

Content

22 July UPDATE: Just released a 1-click installer for runpod, like all runpod installers, just drag and drop it inside the workspace on the left side of the UI then click on the "Terminal" icon on the right side on the UI. Copy and paste these two lines then press enter:

chmod +x INSTANTSTYLE_AUTO_INSTALL-RUNPOD.sh

./INSTANTSTYLE_AUTO_INSTALL-RUNPOD.sh

Wait for everything to be installed and then click on the public URL. By default it installs the juggernautxl8 model from (stablediffusionapi/juggernautxl8), if you want to use another model, make sure you have enough spaces on the gpu you rented and then press ctrl C to stop the webui, go inside the InstantStyle folder, then click on the "Terminal" icon and then type:python app.py --share --model_path --model_path stablediffusionapi/juggernautxl8where you replace "stablediffusionapi/juggernautxl8" by another model from huggingface in diffuser format (I suggest just looking what models are available from the user stablediffusionapi.Hey everyone! I've created a 1-click installer for the AMAZING InstantStyle WebUI that allows you to generate new images in the same style as a single reference style image or to apply a style to another image WITHOUT TRAINING! You Absolutely need to try this out!

Watch the video about InstantStyle: https://youtu.be/yzKjQLE6ABc

The installer of course automates the entire install process!

Here's how it works:

1. Download the installer (INSTANTSTYLE_AUTO_INSTALL.bat) and the launcher (LAUNCHER-InstantStyle.bat )

2. Create a new folder on your computer and place the installer inside

3. Launch the installer

4.???

5. Profit  😎

You can then put the launcher inside the newly created "InstantStyle" folder and double click on it each time you want to launch InstantStyle!😉

As always, supporting me on Patreon allows me to keep creating helpful resources like this for the community. Thank you for your support - now go have some fun😉!

Files

Comments

Aitrepreneur

Still need to try it out myself, I have another video planned for this week but after that yeah I'll take a look

Matt Elder

Installed with no errors, but when first trying to generate an image, I get an error message saying "\InstantStyle\InstantStyle\ip_adapter\attention_processor.py:542: UserWarning: 1Torch was not compiled with flash attention. (Triggered internally at ..\aten\src\ATen\native\transformers\cuda\sdp_utils.cpp:263.)". The generation appears to be running but incredibly slow. Probably 1/10 the speed of generating an image in Forge.

Bob O

File "F:\InstantStyle\env\lib\site-packages\torch\nn\modules\module.py", line 1150, in convert return t.to(device, dtype if t.is_floating_point() or t.is_complex() else None, non_blocking) torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 26.00 MiB. GPU 0 has a total capacity of 11.99 GiB of which 0 bytes is free. Of the allocated memory 10.85 GiB is allocated by PyTorch, and 389.68 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True to avoid fragmentation. See documentation for Memory Management (https://pytorch.org/docs/stable/notes/cuda.html#environment-variables)

Oscar Guerrero

I installed it following the instructions. After installation I launched it and I got this next error, after which it the window closes and nothing happens: A matching Triton is not available, some optimizations will not be enabled Traceback (most recent call last): File "E:\InstantStyle\InstantStyle\env\lib\site-packages\xformers\__init__.py", line 55, in _is_triton_available from xformers.triton.softmax import softmax as triton_softmax # noqa File "E:\InstantStyle\InstantStyle\env\lib\site-packages\xformers\triton\softmax.py", line 11, in import triton ModuleNotFoundError: No module named 'triton'

Parnation

Same issues Matt and Oscar are having. Installed fine(ish?) with a 1Torch error, then relaunching gives the Triton error. Took almost 7 minutes to make the first image when it was able to, with a 4070 ti Edit: Upgraded pip, then ran 'pip install triton' to get rid of the first problem.

Aitrepreneur

It does require more vram to use unfortunately, so depending on your GPU it might take some time to generate

Aitrepreneur

This means you don't have enough vram to run the operation, you can try enabling system memory fallback but the image generation will take a long time

Aitrepreneur

This error is normal because triton is not available on windows

Aitrepreneur

triton error is normal because triton is not available on windows

Mathias R.

Is there any way to run this with 10GB VRAM? It's painfully slow for me. Maybe a google collab setup?

Zoltán István Bíró

Please help! I don't understand how to download the files from huggingface? (URL?) In which directory should I put the models? After that, how do I choose which model I want to use? Do I have to rewrite the path every time to use a different model, which is shown at 6:00 in your video? Thanks for helping!

Aitrepreneur

You don't have to do anything, just dl the installer and launcher and just double click on the installer, everything will be done automatically

Aitrepreneur

google colab will not work, not enough free vram for that unfortunately

Oleg Kuvaev

Git LFS initialized. Cloning into 'InstantStyle'... remote: Enumerating objects: 47, done. remote: Counting objects: 100% (44/44), done. remote: Compressing objects: 100% (44/44), done. remote: Total 47 (delta 5), reused 0 (delta 0), pack-reused 3 Unpacking objects: 100% (47/47), 1.38 MiB | 3.88 MiB/s, done. After this- nothing happening. Not even freeze or error.... just nothing and blinking cursor forever

Aitrepreneur

it's normal, as I said in the video, there is a not of big files to dl, it takes a while depending on your connection

Zoltán István Bíró

Sorry, but that wasn't my question. I copy here, now in bullet points, what I asked: 1. I don't understand how to download the files from huggingface? (URL?) 2. In which directory should I put the models? 3. After that, how do I choose which model I want to use? 4. Do I have to rewrite the path every time to use a different model, which is shown at 6:00 in your video? Thanks for helping!

Aitrepreneur

ah right, well as I said in the video, you need to edit the launcher file, and replace the stabilityai/stable-diffusion-xl-base-1.0 with the name of a sdxl model found on huggingface, you can use the models from this user: https://huggingface.co/stablediffusionapi just chose a model you wanna use (in my video I used this model for example https://huggingface.co/stablediffusionapi/vxpanimaponyv_xl) then you click on the copy url icon next to the name and you replace the stabilityai/stable-diffusion-xl-base-1.0 inside the launcher with (in this case) stablediffusionapi/vxpanimaponyv_xl then you save the file and relaunch it. It will then automatically download the model (for the first time) and launch the webui with this new model. Next time you want to use another model, you can either save the url of the model inside the launcher as a comment or copy and paste the launcher with different model names already put inside. So you might have a basesdxl_LAUNCHER-InstantStyle.bat file a anime_LAUNCHER-InstantStyle.bat file etc, so that you don't need to edit the launcher each time. Also the model is downloaded only the first time, next time it's loads automatically

Matt Elder

That might have been something you should have mentioned in the video. Go look at the comments on YouTube. Lots of people are having all kinds of issues with this. I have to ask, what kind of video card are you using when recording? One person in the YouTube comments had a fairly high end recent card and ONE IMAGE from this took them 20+ minutes to generate.

Aitrepreneur

true that's something I definitely don't take into account since for me one image generates in like 15s but indeed I have a 4090

Matt Elder

The person who said it was going to take them 20 minutes had something like a 3070 Ti, so still a relatively recent card. Perhaps something else was wrong on their system. All I know is that this thing just about froze up on my dual Xeon, 10c/20t (per cpu) with 32GB of RAM and a Quadro M6000 with 12GB. This computer easily runs Forge with decent speed even on relatively large images, but was going to take over an hour to generate a single image with this script.

Oleg Kuvaev

Yes u'r right. Installation is fine. Just needed patience But its also so damn slow. One image like 15 minutes on NVIDIA GeForce RTX 4060 Ti 8 GB

Anthony Marc

I can't get it to launch. I waited through all of the installation, but when I try to use the launcher it opens the command prompt, runs through a couple of things, and then goes away. The last thing it tells me is "OSError: Error no file named pytorch_model.bin, tf_model.h5, model.ckpt.index or flax_model.msgpack found in directory dxl_models/image_encoder" Is that something that is hanging up the launch? Shouldn't I expect it to open a browser tab, or am I doing something wrong? Thanks!

Anthony Marc

I had that happen to. I closed the window and launched the installer again. Then it went further and showed all of the files it was downloading. It takes quite a bit of time. I'm on a pretty smoking fast connection on a pretty smoking fast computer and it still took more than an hour. Normally for these things I just let it run and go to bed and play with it in the morning, so I don't know how long it took in the end.

Ron Harris-White II

Is there a way to change the dimensions of the image to something other than 1024x1024?

Patrice.R

It seems that I do not have the triton model installed despite completing my install: A matching Triton is not available, some optimizations will not be enabled Traceback (most recent call last): File "C:\Users\Duggy\Downloads\InstantStyle - 1 click set up\InstantStyle\env\lib\site-packages\xformers\__init__.py", line 55, in _is_triton_available from xformers.triton.softmax import softmax as triton_softmax # noqa File "C:\Users\Duggy\Downloads\InstantStyle - 1 click set up\InstantStyle\env\lib\site-packages\xformers\triton\softmax.py", line 11, in import triton ModuleNotFoundError: No module named 'triton'

Aitrepreneur

when you input the image in the second field, it will take that ratio for the generation, otherwise you need to modify the app.py file but this will then use even more vram for generation

Aitrepreneur

that means that not everything was downloaded, relaunch the installer again in a new folder and be patient

Aitrepreneur

oof that seems a bit too long tbh, really depends on your connection though, depends on what you call fast I suppose

Aitrepreneur

yes that's normal, it's because triton is not available on windows, normal error

Ron Harris-White II

I have a fairly beefy card, so I am not as concerned with VRAM. Thanks!

Van

I'm using a 4090 with 24gb vRAM and my vRAM instantly caps out and lags my computer and I wind up having to cancel the program after waiting several minutes.

Patrice.R

but I can't get it to launch. I have downloaded everything. I relaunch the launcher several times and nothing opens. Can you please help????

Absalon Prieto

getting this when trying to generate anything. ^^^^^^^^^^^^^ File "D:\AIArt\InstantStyle\InstantStyle\env\Lib\site-packages\torch\nn\functional.py", line 2546, in layer_norm return torch.layer_norm(input, normalized_shape, weight, bias, eps, torch.backends.cudnn.enabled) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ RuntimeError: "LayerNormKernelImpl" not implemented for 'Half'

Aitrepreneur

what do you have inside the command prompt window? What else beside the normal triton error?

Aitrepreneur

that's weird, make sure that you don't have some other heavy duty software running in the background as well, with your gpu it should work absolutely fine too

Patrice.R

Here's the information showing on the command prompt window. Following this the command window closes: A matching Triton is not available, some optimizations will not be enabled Traceback (most recent call last): File "C:\Users\Patrice\Downloads\InstantStyle\InstantStyle\env\lib\site-packages\xformers\__init__.py", line 55, in _is_triton_available from xformers.triton.softmax import softmax as triton_softmax # noqa File "C:\Users\Patrice\Downloads\InstantStyle\InstantStyle\env\lib\site-packages\xformers\triton\softmax.py", line 11, in import triton ModuleNotFoundError: No module named 'triton' Loading pipeline components...: 100%|████████████████████████████████████████████████████| 7/7 [00:09<00:00, 1.39s/it] Traceback (most recent call last): File "C:\Users\Patrice\Downloads\InstantStyle\InstantStyle\app.py", line 65, in ip_model = IPAdapterXL(pipe, image_encoder_path, ip_ckpt, device, target_blocks=["up_blocks.0.attentions.1"]) File "C:\Users\Patrice\Downloads\InstantStyle\InstantStyle\ip_adapter\ip_adapter.py", line 78, in __init__ self.image_encoder = CLIPVisionModelWithProjection.from_pretrained(self.image_encoder_path).to( File "C:\Users\Patrice\Downloads\InstantStyle\InstantStyle\env\lib\site-packages\transformers\modeling_utils.py", line 3260, in from_pretrained raise EnvironmentError( OSError: Error no file named pytorch_model.bin, tf_model.h5, model.ckpt.index or flax_model.msgpack found in directory sdxl_models/image_encoder.

Aitrepreneur

Looks like it's getting cut on patreon chat, better dm me instead but I think that not everything was dl correctly, try relaunching the installer again in a new folder and be patient for everything to install, some files take time to dl make sure you got no antivirus/firewall/vpn running before relaunching, it's always the same issue anyway..

Nico S

How do I install into another drive than the default C drive? I've installed into a drive with little space, and my drive has run out of space. Now I have to delete or move those files taking up the space. What files do I have to look for to move or delete?

Patrice.R

it works thanks! The firewall and anti-virus were the causes for the bad install, duh!;) Where do I find the generated images? when I download it it , I have a txt.file and not a PNG

huiminh ding

do you have this in comfyui gui?

Jay

I have the same error and I have a 4090

Rich Sekmistrz

I have a similar issue to one above - I have disabled my windows AV and all Windows 11 AntiVirus and Threat protection, as well as my Firewall - I get this error. I have tried 2x and waited til it ended with no rushing or relaunching. Any ideas? I will try again for a 3rd time but it seems I am just stuck: Traceback (most recent call last): File "E:\AI\Style Copy\InstantStyle\app.py", line 65, in ip_model = IPAdapterXL(pipe, image_encoder_path, ip_ckpt, device, target_blocks=["up_blocks.0.attentions.1"]) File "E:\AI\Style Copy\InstantStyle\ip_adapter\ip_adapter.py", line 78, in __init__ self.image_encoder = CLIPVisionModelWithProjection.from_pretrained(self.image_encoder_path).to( File "E:\AI\Style Copy\InstantStyle\env\lib\site-packages\transformers\modeling_utils.py", line 3260, in from_pretrained raise EnvironmentError( OSError: Error no file named pytorch_model.bin, tf_model.h5, model.ckpt.index or flax_model.msgpack found in directory sdlx_models/image_encoder.

Nico S

How large free space do I need on the drive for the download?

Nico S

Thank you!

PatreonUser

Is there an easy way to convert the 1-click install to a runpod template/install?

Minh Thien Nguyen

how can I change the model it uses? I tried to follow your instruction in the Launch.bat. My code look like this python app.py --inbrowser --model_path stablediffusionapi/majicmix-fantasy The launcher did load a lot of things but then shutdown without any errors Thank you

Aitrepreneur

because that's a 1.5 model, not a sdxl model, choose another one

Whatever

Hello, I used your auto installer for InstantStyle and After everything is installed I get this error: torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 20.00 MiB. GPU 0 has a total capacity of 2.00 GiB of which 0 bytes is free. Of the allocated memory 8.96 GiB is allocated by PyTorch, and 284.05 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True to avoid fragmentation. I know how to force PyTorch to work on CPU only in A1111, for example, but in your installation of InstantStyle I cannot find any configuration/environment variables/webui file I can modify. How can I force InstantStyle to work with the CPU? Hope you can help, and thank you in advance!

GrindBird

I have the same issue, let me know if you figure it out!

Aitrepreneur

hi, this error means you don't have enough Vram to do the operation, unfortunately instantstyle is really unoptimized and require a lot of vram to run at a decent speed, it will never work on a cpu unfortunately

Kuzo

is there any way to use my local model checkpoints without using the hugging face api?

Aitrepreneur

for now no, they need to be in diffuser format

Breaker

I personally can't get the 1 Click installer to work, as it closes automatically after a few seconds :(

Nico S

What are the system requirements for using InstantStyle? How much memory and what GPU is required?

Aitrepreneur

could a multitude of reasons, first open a command prompt window and drag and drop the installer file inside the window, then press enter, this should avoid the window from closing and it will at least give us an error code we can use to troubleshoot the error, make also sure that you have disabled any antivirus/firewall from running and blocking the file/connection. This could also be a bad initial python install, you need to uninstall your current python installation and reinstall it correctly. Go to the add and remove programs, search for python and uninstall both the current python version and the python install program. Once this is done, go here and download this installer: https://www.python.org/ftp/python/3.10.6/python-3.10.6-amd64.exe Run it and check the “Add python 3.10 to Path” checkbox and continue with the installation.

Bert van der meer

winget error over here, not found. [WARN] Winget is not installed on this system. [INFO] Installing Winget... % Total % Received % Xferd Average Speed Time Time Time Current Dload Upload Total Spent Left Speed 100 9 100 9 0 0 41 0 --:--:-- --:--:-- --:--:-- 42 [INFO] Winget installed successfully. Python already installed. Git already installed. [INFO] Installing Microsoft.VCRedist.2015+.x64... 'winget' is not recognized as an internal or external command, operable program or batch file. [INFO] Installing Microsoft.VCRedist.2015+.x86... 'winget' is not recognized as an internal or external command, operable program or batch file. [INFO] Installing vs_BuildTools... % Total % Received % Xferd Average Speed Time Time Time Current Dload Upload Total Spent Left Speed 0 0 0 0 0 0 0 0 --:--:-- --:--:-- --:--:-- 0 100 3908k 100 3908k 0 0 4518k 0 --:--:-- --:--:-- --:--:-- 4518k Git LFS initialized. fatal: destination path 'InstantStyle' already exists and is not an empty directory. Looking in indexes: https://download.pytorch.org/whl/cu118 Collecting torch Downloading https://download.pytorch.org/whl/cu118/torch-2.3.0%2Bcu118-cp310-cp310-win_amd64.whl (2673.0 MB) |████████████████████████████████| 2668.3 MB 226 kB/s eta 0:00:22

Aitrepreneur

redo the install just delete the InstantStyle folder and try again

Bill G.

Hi.. Okay.. So I followed your directions with 2 of your 1 click installs.. This one.. Stand alone ID and Instant Style.. neither one will work. CMD ran the programing fully and then just shut down after install... moved bat files into the folders as directed, ran both bat files, CMD ran, shut down automatically. Hmmmm.. Now what? What the hell can I be doing wrong? HELP. Thanks.

Aitrepreneur

could a multitude of reasons, first open a command prompt window and drag and drop the installer file inside the window, then press enter, this should avoid the window from closing and it will at least give us an error code we can use to troubleshoot the error, make also sure that you have disabled any antivirus/firewall from running and blocking the file/connection. This could also be a bad initial python install, you need to uninstall any of your current python installations and reinstall it correctly. Go to the add and remove programs, search for python and uninstall both the current python version and the python install program. Once this is done, go here and download this installer: https://www.python.org/ftp/python/3.10.11/python-3.10.11-amd64.exe Run it and check the “Add python 3.10 to Path” checkbox and continue with the installation. You can check that the right python version is installed by opening a new command prompt window and typing: python --version and it should give you the 3.10.11 version Then just relaunch the 1-click installer in a new folder and try again.

Bill G.

Okay new issue, Finally Got Installed as I told you about; everything working fine all day Autolaunch.bat working great, then next day, went back into file, tried to open the launch auto bat and all that happened was a FLASH on the SCREEN of the CMD as it seemed to be trying to load, (for Spit Second) and closes) disappears from screen and nothing happens. reinstalled completely and same issue,, installed fine, started up. I closed program and , put auto bat into file folder, but this time suto.bat attempted to run and flash on the screen of CMD open and closes instantly, and nothing happened. the failure this time happened right after install, Any ideas what can be going on? It worked fine one minute, wont load next day, nothing changed.

Aitrepreneur

Hmm and you are using the right launcher.bat file right? It could a multitude of reasons, but first open a command prompt window and drag and drop the launcher file inside the window, then press enter, this should avoid the window from closing and it will at least give us an error code we can use to troubleshoot the error, make also sure that you have disabled any antivirus/firewall from running and blocking the file/connection.

notsofst

Hi, can you share the specs needed for this app? I'm receiving an out of memory error from CUDA: torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 20.00 MiB. GPU I'm on an EC2 p2.xlarge which has a single NVIDIA K80 GPU with 12 GB vRAM. The line in app.py that is failing is: ip_model = IPAdapterXL(pipe, image_encoder_path, ip_ckpt, device, target_blocks=["up_blocks.0.attentions.1"]) Using nvidia-smi, it does look like it's maxing out the memory. Is there a tweak to keep it down, or what size card would you recommend?

Aitrepreneur

I don't know what driver you got but normally if you dl the latest nvidia drivers you should now have access to system memory fallback that wiil be enabled by default, allowing you to offload some of that required gpu power to ram, so it will work but it will be slow with a 12gb of vram GPU, the more vram you got the faster it will be

winwin

Would you pls tell me how to run the InstantSytle on runpod?

Aitrepreneur

I haven't tried it yet but I might do a quick installer if I have the time

notsofst

Just following up, I was able to get an image to generate with ~21 GB of video memory on a Amazon G5 EC2 instance.

Aitrepreneur

I will try to make a runpod template so that it's easier to run for everyone

winwin

Many thx👍

Aitrepreneur

hi, just uploaded the runpod installer, check the patreon post for the updated info on how to use it :)

winwin

I appreciate your help!how can download the result as a photo file? I can only download the result as a text file now.

Aitrepreneur

I mean just right click on the image and save image as?

winwin

yes, right click can save the file as a small file size webp (100kb), can I get a higher resolution picture? If I click the download icon, It is saved as a txt file

Peter Rossa

curl: (35) schannel: next InitializeSecurityContext failed: Unknown error (0x80092012) - The revocation function was unable to check revocation for the certificate. [ERROR] Download failed. Please restart the installer

Aitrepreneur

I mean it saves the current image with its current resolution, if you want an higher res, you can just upscale it inside auto1111 for example

Aitrepreneur

Looks like a connection issue, make sure you don't have any antivirus/firewall/vpn running before launching the installer

winwin

All right ,thank you!

kasukus

dosen't work for me also. im using a asus RTX3060TI 8GB VRAM

Soda Sullivan

i had an issue were it stopped. but stable diffusion had not loaded. i simply ran the installer again and it worked fine

Kythe Druid

when i use the onclick windows keeps giving me the following: "The package is either corrupted or invalid."

Kythe Druid

Microsoft Windows [Version 10.0.22631.5039] (c) Microsoft Corporation. All rights reserved. A:\AI_programs>A:\AI_programs\INSTANTSTYLE_AUTO_INSTALL.bat [WARN] Winget is not installed on this system. [INFO] Installing Winget... % Total % Received % Xferd Average Speed Time Time Time Current Dload Upload Total Spent Left Speed 100 9 100 9 0 0 15 0 --:--:-- --:--:-- --:--:-- 15 [INFO] Winget installed successfully. Python already installed. Git already installed. [INFO] Installing Microsoft.VCRedist.2015+.x64... 'winget' is not recognized as an internal or external command, operable program or batch file. [INFO] Installing Microsoft.VCRedist.2015+.x86... 'winget' is not recognized as an internal or external command, operable program or batch file. [INFO] Installing vs_BuildTools... % Total % Received % Xferd Average Speed Time Time Time Current Dload Upload Total Spent Left Speed 0 0 0 0 0 0 0 0 --:--:-- --:--:-- --:--:-- 0 100 4349k 100 4349k 0 0 2632k 0 0:00:01 0:00:01 --:--:-- 9068k Git LFS initialized. Cloning into 'InstantStyle'... remote: Enumerating objects: 47, done. remote: Counting objects: 100% (47/47), done. remote: Compressing objects: 100% (41/41), done. remote: Total 47 (delta 5), reused 47 (delta 5), pack-reused 0 (from 0) Unpacking objects: 100% (47/47), 1.38 MiB | 7.39 MiB/s, done. Filtering content: 100% (3/3), 7.52 GiB | 31.92 MiB/s, done. Looking in indexes: https://download.pytorch.org/whl/cu118 Collecting torch Using cached https://download.pytorch.org/whl/cu118/torch-2.6.0%2Bcu118-cp310-cp310-win_amd64.whl (2728.9 MB) Collecting torchvision Using cached https://download.pytorch.org/whl/cu118/torchvision-0.21.0%2Bcu118-cp310-cp310-win_amd64.whl (5.3 MB) Collecting torchaudio Using cached https://download.pytorch.org/whl/cu118/torchaudio-2.6.0%2Bcu118-cp310-cp310-win_amd64.whl (4.0 MB) Collecting xformers Using cached https://download.pytorch.org/whl/cu118/xformers-0.0.27.post2%2Bcu118-cp310-cp310-win_amd64.whl (10.8 MB) Collecting filelock Using cached https://download.pytorch.org/whl/filelock-3.13.1-py3-none-any.whl (11 kB) Collecting jinja2 Using cached https://download.pytorch.org/whl/Jinja2-3.1.4-py3-none-any.whl (133 kB) Collecting sympy==1.13.1 Using cached https://download.pytorch.org/whl/sympy-1.13.1-py3-none-any.whl (6.2 MB) Collecting typing-extensions>=4.10.0 Using cached https://download.pytorch.org/whl/typing_extensions-4.12.2-py3-none-any.whl (37 kB) Collecting networkx Using cached https://download.pytorch.org/whl/networkx-3.3-py3-none-any.whl (1.7 MB) Collecting fsspec Using cached https://download.pytorch.org/whl/fsspec-2024.6.1-py3-none-any.whl (177 kB) Collecting mpmath<1.4,>=1.1.0 Using cached https://download.pytorch.org/whl/mpmath-1.3.0-py3-none-any.whl (536 kB) Collecting numpy Using cached https://download.pytorch.org/whl/numpy-2.1.2-cp310-cp310-win_amd64.whl (12.9 MB) Collecting pillow!=8.3.*,>=5.3.0 Using cached https://download.pytorch.org/whl/pillow-11.0.0-cp310-cp310-win_amd64.whl (2.6 MB) Collecting xformers Using cached https://download.pytorch.org/whl/cu118/xformers-0.0.27.post1%2Bcu118-cp310-cp310-win_amd64.whl (10.8 MB) Using cached https://download.pytorch.org/whl/cu118/xformers-0.0.27%2Bcu118-cp310-cp310-win_amd64.whl (73.9 MB) Using cached https://download.pytorch.org/whl/cu118/xformers-0.0.26.post1%2Bcu118-cp310-cp310-win_amd64.whl (102.2 MB) Using cached https://download.pytorch.org/whl/cu118/xformers-0.0.26%2Bcu118-cp310-cp310-win_amd64.whl (102.2 MB) Using cached https://download.pytorch.org/whl/cu118/xformers-0.0.25.post1%2Bcu118-cp310-cp310-win_amd64.whl (102.1 MB) Using cached https://download.pytorch.org/whl/cu118/xformers-0.0.25%2Bcu118-cp310-cp310-win_amd64.whl (102.1 MB) Using cached https://download.pytorch.org/whl/cu118/xformers-0.0.24%2Bcu118-cp310-cp310-win_amd64.whl (102.1 MB) Using cached https://download.pytorch.org/whl/cu118/xformers-0.0.23.post1%2Bcu118-cp310-cp310-win_amd64.whl (97.8 MB) Using cached https://download.pytorch.org/whl/cu118/xformers-0.0.23%2Bcu118-cp310-cp310-win_amd64.whl (97.7 MB) Using cached https://download.pytorch.org/whl/cu118/xformers-0.0.22.post7%2Bcu118-cp310-cp310-win_amd64.whl (97.5 MB) Using cached https://download.pytorch.org/whl/cu118/xformers-0.0.22.post4%2Bcu118-cp310-cp310-win_amd64.whl (97.5 MB) Using cached https://download.pytorch.org/whl/cu118/xformers-0.0.22.post3%2Bcu118-cp310-cp310-win_amd64.whl (97.5 MB) Using cached https://download.pytorch.org/whl/cu118/xformers-0.0.22.post2%2Bcu118-cp310-cp310-win_amd64.whl (97.5 MB) INFO: pip is looking at multiple versions of torchaudio to determine which version is compatible with other requirements. This could take a while. Collecting torchaudio Using cached https://download.pytorch.org/whl/cu118/torchaudio-2.5.1%2Bcu118-cp310-cp310-win_amd64.whl (4.0 MB) Using cached https://download.pytorch.org/whl/cu118/torchaudio-2.5.0%2Bcu118-cp310-cp310-win_amd64.whl (4.0 MB) Using cached https://download.pytorch.org/whl/cu118/torchaudio-2.4.1%2Bcu118-cp310-cp310-win_amd64.whl (4.0 MB) Using cached https://download.pytorch.org/whl/cu118/torchaudio-2.4.0%2Bcu118-cp310-cp310-win_amd64.whl (4.0 MB) Using cached https://download.pytorch.org/whl/cu118/torchaudio-2.3.1%2Bcu118-cp310-cp310-win_amd64.whl (4.0 MB) Using cached https://download.pytorch.org/whl/cu118/torchaudio-2.3.0%2Bcu118-cp310-cp310-win_amd64.whl (4.0 MB) Using cached https://download.pytorch.org/whl/cu118/torchaudio-2.2.2%2Bcu118-cp310-cp310-win_amd64.whl (4.0 MB) Using cached https://download.pytorch.org/whl/cu118/torchaudio-2.2.1%2Bcu118-cp310-cp310-win_amd64.whl (4.0 MB) Using cached https://download.pytorch.org/whl/cu118/torchaudio-2.2.0%2Bcu118-cp310-cp310-win_amd64.whl (4.0 MB) Using cached https://download.pytorch.org/whl/cu118/torchaudio-2.1.2%2Bcu118-cp310-cp310-win_amd64.whl (3.9 MB) Using cached https://download.pytorch.org/whl/cu118/torchaudio-2.1.1%2Bcu118-cp310-cp310-win_amd64.whl (3.9 MB) Using cached https://download.pytorch.org/whl/cu118/torchaudio-2.1.0%2Bcu118-cp310-cp310-win_amd64.whl (3.9 MB) Using cached https://download.pytorch.org/whl/cu118/torchaudio-2.0.2%2Bcu118-cp310-cp310-win_amd64.whl (2.5 MB) Using cached https://download.pytorch.org/whl/cu118/torchaudio-2.0.1%2Bcu118-cp310-cp310-win_amd64.whl (2.5 MB) Using cached https://download.pytorch.org/whl/cu118/torchaudio-2.0.0%2Bcu118-cp310-cp310-win_amd64.whl (2.5 MB) INFO: pip is looking at multiple versions of torchvision to determine which version is compatible with other requirements. This could take a while. Collecting torchvision Using cached https://download.pytorch.org/whl/cu118/torchvision-0.20.1%2Bcu118-cp310-cp310-win_amd64.whl (5.3 MB) Collecting torch Using cached https://download.pytorch.org/whl/cu118/torch-2.5.1%2Bcu118-cp310-cp310-win_amd64.whl (2700.1 MB) INFO: pip is looking at multiple versions of torch to determine which version is compatible with other requirements. This could take a while. Collecting torchvision Using cached https://download.pytorch.org/whl/cu118/torchvision-0.20.0%2Bcu118-cp310-cp310-win_amd64.whl (5.3 MB) Collecting torch Using cached https://download.pytorch.org/whl/cu118/torch-2.5.0%2Bcu118-cp310-cp310-win_amd64.whl (2700.2 MB) Collecting torchvision Using cached https://download.pytorch.org/whl/cu118/torchvision-0.19.1%2Bcu118-cp310-cp310-win_amd64.whl (5.0 MB) Collecting torch Using cached https://download.pytorch.org/whl/cu118/torch-2.4.1%2Bcu118-cp310-cp310-win_amd64.whl (2695.4 MB) Collecting torchvision Using cached https://download.pytorch.org/whl/cu118/torchvision-0.19.0%2Bcu118-cp310-cp310-win_amd64.whl (5.0 MB) Collecting torch Using cached https://download.pytorch.org/whl/cu118/torch-2.4.0%2Bcu118-cp310-cp310-win_amd64.whl (2692.5 MB) Collecting MarkupSafe>=2.0 Using cached https://download.pytorch.org/whl/MarkupSafe-2.1.5-cp310-cp310-win_amd64.whl (17 kB) Installing collected packages: mpmath, typing-extensions, sympy, pillow, numpy, networkx, MarkupSafe, fsspec, filelock, jinja2, torch, xformers, torchvision, torchaudio Successfully installed MarkupSafe-2.1.5 filelock-3.13.1 fsspec-2024.6.1 jinja2-3.1.4 mpmath-1.3.0 networkx-3.3 numpy-2.1.2 pillow-11.0.0 sympy-1.13.1 torch-2.4.0+cu118 torchaudio-2.4.0+cu118 torchvision-0.19.0+cu118 typing-extensions-4.12.2 xformers-0.0.27.post2+cu118 [notice] A new release of pip available: 22.2.1 -> 25.0.1 [notice] To update, run: python.exe -m pip install --upgrade pip Collecting diffusers>=0.25.1 Using cached diffusers-0.32.2-py3-none-any.whl (3.2 MB) Collecting transformers>=4.37.1 Using cached transformers-4.49.0-py3-none-any.whl (10.0 MB) Collecting accelerate Using cached accelerate-1.5.2-py3-none-any.whl (345 kB) Collecting safetensors Using cached safetensors-0.5.3-cp38-abi3-win_amd64.whl (308 kB) Collecting einops Using cached einops-0.8.1-py3-none-any.whl (64 kB) Collecting spaces>=0.19.4 Using cached spaces-0.32.0-py3-none-any.whl (29 kB) Collecting omegaconf Using cached omegaconf-2.3.0-py3-none-any.whl (79 kB) Collecting peft Using cached peft-0.14.0-py3-none-any.whl (374 kB) Collecting huggingface-hub>=0.20.2 Using cached huggingface_hub-0.29.3-py3-none-any.whl (468 kB) Collecting opencv-python Using cached opencv_python-4.11.0.86-cp37-abi3-win_amd64.whl (39.5 MB) Collecting gradio Using cached gradio-5.21.0-py3-none-any.whl (46.2 MB) Collecting controlnet_aux Using cached controlnet_aux-0.0.9-py3-none-any.whl (282 kB) Collecting gdown Using cached gdown-5.2.0-py3-none-any.whl (18 kB) Collecting importlib-metadata Using cached importlib_metadata-8.6.1-py3-none-any.whl (26 kB) Collecting requests Using cached requests-2.32.3-py3-none-any.whl (64 kB) Requirement already satisfied: Pillow in a:\ai_programs\instantstyle\env\lib\site-packages (from diffusers>=0.25.1->-r requirements.txt (line 1)) (11.0.0) Collecting regex!=2019.12.17 Using cached regex-2024.11.6-cp310-cp310-win_amd64.whl (274 kB) Requirement already satisfied: filelock in a:\ai_programs\instantstyle\env\lib\site-packages (from diffusers>=0.25.1->-r requirements.txt (line 1)) (3.13.1) Requirement already satisfied: numpy in a:\ai_programs\instantstyle\env\lib\site-packages (from diffusers>=0.25.1->-r requirements.txt (line 1)) (2.1.2) Collecting pyyaml>=5.1 Using cached PyYAML-6.0.2-cp310-cp310-win_amd64.whl (161 kB) Collecting packaging>=20.0 Using cached packaging-24.2-py3-none-any.whl (65 kB) Collecting tqdm>=4.27 Using cached tqdm-4.67.1-py3-none-any.whl (78 kB) Collecting tokenizers<0.22,>=0.21 Using cached tokenizers-0.21.1-cp39-abi3-win_amd64.whl (2.4 MB) Collecting psutil Using cached psutil-7.0.0-cp37-abi3-win_amd64.whl (244 kB) Requirement already satisfied: torch>=2.0.0 in a:\ai_programs\instantstyle\env\lib\site-packages (from accelerate->-r requirements.txt (line 5)) (2.4.0+cu118) Collecting pydantic<3,>=1 Using cached pydantic-2.10.6-py3-none-any.whl (431 kB) Collecting httpx>=0.20 Using cached httpx-0.28.1-py3-none-any.whl (73 kB) Collecting psutil Using cached psutil-5.9.8-cp37-abi3-win_amd64.whl (255 kB) Requirement already satisfied: typing-extensions<5,>=4 in a:\ai_programs\instantstyle\env\lib\site-packages (from spaces>=0.19.4->-r requirements.txt (line 8)) (4.12.2) Collecting antlr4-python3-runtime==4.9.* Using cached antlr4_python3_runtime-4.9.3-py3-none-any.whl Requirement already satisfied: fsspec>=2023.5.0 in a:\ai_programs\instantstyle\env\lib\site-packages (from huggingface-hub>=0.20.2->-r requirements.txt (line 11)) (2024.6.1) Collecting fastapi<1.0,>=0.115.2 Using cached fastapi-0.115.11-py3-none-any.whl (94 kB) Collecting pandas<3.0,>=1.0 Using cached pandas-2.2.3-cp310-cp310-win_amd64.whl (11.6 MB) Collecting pydub Using cached pydub-0.25.1-py2.py3-none-any.whl (32 kB) Collecting orjson~=3.0 Using cached orjson-3.10.15-cp310-cp310-win_amd64.whl (133 kB) Collecting ruff>=0.9.3 Using cached ruff-0.11.0-py3-none-win_amd64.whl (11.4 MB) Requirement already satisfied: jinja2<4.0 in a:\ai_programs\instantstyle\env\lib\site-packages (from gradio->-r requirements.txt (line 13)) (3.1.4) Collecting uvicorn>=0.14.0 Using cached uvicorn-0.34.0-py3-none-any.whl (62 kB) Collecting anyio<5.0,>=3.0 Using cached anyio-4.9.0-py3-none-any.whl (100 kB) Collecting tomlkit<0.14.0,>=0.12.0 Using cached tomlkit-0.13.2-py3-none-any.whl (37 kB) Collecting gradio-client==1.7.2 Using cached gradio_client-1.7.2-py3-none-any.whl (322 kB) Collecting starlette<1.0,>=0.40.0 Using cached starlette-0.46.1-py3-none-any.whl (71 kB) Collecting aiofiles<24.0,>=22.0 Using cached aiofiles-23.2.1-py3-none-any.whl (15 kB) Collecting python-multipart>=0.0.18 Using cached python_multipart-0.0.20-py3-none-any.whl (24 kB) Collecting ffmpy Using cached ffmpy-0.5.0-py3-none-any.whl (6.0 kB) Collecting safehttpx<0.2.0,>=0.1.6 Using cached safehttpx-0.1.6-py3-none-any.whl (8.7 kB) Collecting semantic-version~=2.0 Using cached semantic_version-2.10.0-py2.py3-none-any.whl (15 kB) Collecting typer<1.0,>=0.12 Using cached typer-0.15.2-py3-none-any.whl (45 kB) Requirement already satisfied: markupsafe~=2.0 in a:\ai_programs\instantstyle\env\lib\site-packages (from gradio->-r requirements.txt (line 13)) (2.1.5) Collecting groovy~=0.1 Using cached groovy-0.1.2-py3-none-any.whl (14 kB) Collecting websockets<16.0,>=10.0 Using cached websockets-15.0.1-cp310-cp310-win_amd64.whl (176 kB) Requirement already satisfied: torchvision in a:\ai_programs\instantstyle\env\lib\site-packages (from controlnet_aux->-r requirements.txt (line 14)) (0.19.0+cu118) Collecting scikit-image Using cached scikit_image-0.25.2-cp310-cp310-win_amd64.whl (12.8 MB) Collecting timm<=0.6.7 Using cached timm-0.6.7-py3-none-any.whl (509 kB) Collecting opencv-python-headless Using cached opencv_python_headless-4.11.0.86-cp37-abi3-win_amd64.whl (39.4 MB) Collecting scipy Using cached scipy-1.15.2-cp310-cp310-win_amd64.whl (41.2 MB) Collecting beautifulsoup4 Using cached beautifulsoup4-4.13.3-py3-none-any.whl (186 kB) Collecting exceptiongroup>=1.0.2 Using cached exceptiongroup-1.2.2-py3-none-any.whl (16 kB) Collecting idna>=2.8 Using cached idna-3.10-py3-none-any.whl (70 kB) Collecting sniffio>=1.1 Using cached sniffio-1.3.1-py3-none-any.whl (10 kB) Collecting certifi Using cached certifi-2025.1.31-py3-none-any.whl (166 kB) Collecting httpcore==1.* Using cached httpcore-1.0.7-py3-none-any.whl (78 kB) Collecting h11<0.15,>=0.13 Using cached h11-0.14.0-py3-none-any.whl (58 kB) Collecting pytz>=2020.1 Using cached pytz-2025.1-py2.py3-none-any.whl (507 kB) Collecting python-dateutil>=2.8.2 Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl (229 kB) Collecting tzdata>=2022.7 Using cached tzdata-2025.1-py2.py3-none-any.whl (346 kB) Collecting pydantic-core==2.27.2 Using cached pydantic_core-2.27.2-cp310-cp310-win_amd64.whl (2.0 MB) Collecting annotated-types>=0.6.0 Using cached annotated_types-0.7.0-py3-none-any.whl (13 kB) Collecting charset-normalizer<4,>=2 Using cached charset_normalizer-3.4.1-cp310-cp310-win_amd64.whl (102 kB) Collecting urllib3<3,>=1.21.1 Using cached urllib3-2.3.0-py3-none-any.whl (128 kB) Requirement already satisfied: sympy in a:\ai_programs\instantstyle\env\lib\site-packages (from torch>=2.0.0->accelerate->-r requirements.txt (line 5)) (1.13.1) Requirement already satisfied: networkx in a:\ai_programs\instantstyle\env\lib\site-packages (from torch>=2.0.0->accelerate->-r requirements.txt (line 5)) (3.3) Collecting colorama Using cached colorama-0.4.6-py2.py3-none-any.whl (25 kB) Collecting click>=8.0.0 Using cached click-8.1.8-py3-none-any.whl (98 kB) Collecting rich>=10.11.0 Using cached rich-13.9.4-py3-none-any.whl (242 kB) Collecting shellingham>=1.3.0 Using cached shellingham-1.5.4-py2.py3-none-any.whl (9.8 kB) Collecting soupsieve>1.2 Using cached soupsieve-2.6-py3-none-any.whl (36 kB) Collecting zipp>=3.20 Using cached zipp-3.21.0-py3-none-any.whl (9.6 kB) Collecting PySocks!=1.5.7,>=1.5.6 Using cached PySocks-1.7.1-py3-none-any.whl (16 kB) Collecting imageio!=2.35.0,>=2.33 Using cached imageio-2.37.0-py3-none-any.whl (315 kB) Collecting lazy-loader>=0.4 Using cached lazy_loader-0.4-py3-none-any.whl (12 kB) Collecting tifffile>=2022.8.12 Using cached tifffile-2025.3.13-py3-none-any.whl (226 kB) Collecting six>=1.5 Using cached six-1.17.0-py2.py3-none-any.whl (11 kB) Collecting markdown-it-py>=2.2.0 Using cached markdown_it_py-3.0.0-py3-none-any.whl (87 kB) Collecting pygments<3.0.0,>=2.13.0 Using cached pygments-2.19.1-py3-none-any.whl (1.2 MB) Requirement already satisfied: mpmath<1.4,>=1.1.0 in a:\ai_programs\instantstyle\env\lib\site-packages (from sympy->torch>=2.0.0->accelerate->-r requirements.txt (line 5)) (1.3.0) Collecting mdurl~=0.1 Using cached mdurl-0.1.2-py3-none-any.whl (10.0 kB) Installing collected packages: pytz, pydub, antlr4-python3-runtime, zipp, websockets, urllib3, tzdata, tomlkit, tifffile, soupsieve, sniffio, six, shellingham, semantic-version, scipy, safetensors, ruff, regex, pyyaml, python-multipart, PySocks, pygments, pydantic-core, psutil, packaging, orjson, opencv-python-headless, opencv-python, mdurl, imageio, idna, h11, groovy, ffmpy, exceptiongroup, einops, colorama, charset-normalizer, certifi, annotated-types, aiofiles, tqdm, requests, python-dateutil, pydantic, omegaconf, markdown-it-py, lazy-loader, importlib-metadata, httpcore, click, beautifulsoup4, anyio, uvicorn, starlette, scikit-image, rich, pandas, huggingface-hub, httpx, typer, tokenizers, timm, safehttpx, gradio-client, gdown, fastapi, diffusers, accelerate, transformers, gradio, controlnet_aux, spaces, peft Successfully installed PySocks-1.7.1 accelerate-1.5.2 aiofiles-23.2.1 annotated-types-0.7.0 antlr4-python3-runtime-4.9.3 anyio-4.9.0 beautifulsoup4-4.13.3 certifi-2025.1.31 charset-normalizer-3.4.1 click-8.1.8 colorama-0.4.6 controlnet_aux-0.0.9 diffusers-0.32.2 einops-0.8.1 exceptiongroup-1.2.2 fastapi-0.115.11 ffmpy-0.5.0 gdown-5.2.0 gradio-5.21.0 gradio-client-1.7.2 groovy-0.1.2 h11-0.14.0 httpcore-1.0.7 httpx-0.28.1 huggingface-hub-0.29.3 idna-3.10 imageio-2.37.0 importlib-metadata-8.6.1 lazy-loader-0.4 markdown-it-py-3.0.0 mdurl-0.1.2 omegaconf-2.3.0 opencv-python-4.11.0.86 opencv-python-headless-4.11.0.86 orjson-3.10.15 packaging-24.2 pandas-2.2.3 peft-0.14.0 psutil-5.9.8 pydantic-2.10.6 pydantic-core-2.27.2 pydub-0.25.1 pygments-2.19.1 python-dateutil-2.9.0.post0 python-multipart-0.0.20 pytz-2025.1 pyyaml-6.0.2 regex-2024.11.6 requests-2.32.3 rich-13.9.4 ruff-0.11.0 safehttpx-0.1.6 safetensors-0.5.3 scikit-image-0.25.2 scipy-1.15.2 semantic-version-2.10.0 shellingham-1.5.4 six-1.17.0 sniffio-1.3.1 soupsieve-2.6 spaces-0.32.0 starlette-0.46.1 tifffile-2025.3.13 timm-0.6.7 tokenizers-0.21.1 tomlkit-0.13.2 tqdm-4.67.1 transformers-4.49.0 typer-0.15.2 tzdata-2025.1 urllib3-2.3.0 uvicorn-0.34.0 websockets-15.0.1 zipp-3.21.0 [notice] A new release of pip available: 22.2.1 -> 25.0.1 [notice] To update, run: python.exe -m pip install --upgrade pip A module that was compiled using NumPy 1.x cannot be run in NumPy 2.1.2 as it may crash. To support both 1.x and 2.x versions of NumPy, modules must be compiled with NumPy 2.0. Some module may need to rebuild instead e.g. with 'pybind11>=2.12'. If you are a user of the module, the easiest solution will be to downgrade to 'numpy<2' or try to upgrade the affected module. We expect that some modules will need time to support NumPy 2. Traceback (most recent call last): File "A:\AI_programs\InstantStyle\app.py", line 7, in import torch File "A:\AI_programs\InstantStyle\env\lib\site-packages\torch\__init__.py", line 2120, in from torch._higher_order_ops import cond File "A:\AI_programs\InstantStyle\env\lib\site-packages\torch\_higher_order_ops\__init__.py", line 1, in from .cond import cond File "A:\AI_programs\InstantStyle\env\lib\site-packages\torch\_higher_order_ops\cond.py", line 5, in import torch._subclasses.functional_tensor File "A:\AI_programs\InstantStyle\env\lib\site-packages\torch\_subclasses\functional_tensor.py", line 42, in class FunctionalTensor(torch.Tensor): File "A:\AI_programs\InstantStyle\env\lib\site-packages\torch\_subclasses\functional_tensor.py", line 258, in FunctionalTensor cpu = _conversion_method_template(device=torch.device("cpu")) A:\AI_programs\InstantStyle\env\lib\site-packages\torch\_subclasses\functional_tensor.py:258: UserWarning: Failed to initialize NumPy: _ARRAY_API not found (Triggered internally at C:\actions-runner\_work\pytorch\pytorch\builder\windows\pytorch\torch\csrc\utils\tensor_numpy.cpp:84.) cpu = _conversion_method_template(device=torch.device("cpu")) A matching Triton is not available, some optimizations will not be enabled Traceback (most recent call last): File "A:\AI_programs\InstantStyle\env\lib\site-packages\xformers\__init__.py", line 57, in _is_triton_available import triton # noqa ModuleNotFoundError: No module named 'triton' Loading pipeline components...: 43%|██████████████████████▎ | 3/7 [00:13<00:18, 4.65s/it] Traceback (most recent call last): File "A:\AI_programs\InstantStyle\app.py", line 54, in pipe = StableDiffusionXLControlNetPipeline.from_pretrained( File "A:\AI_programs\InstantStyle\env\lib\site-packages\huggingface_hub\utils\_validators.py", line 114, in _inner_fn return fn(*args, **kwargs) File "A:\AI_programs\InstantStyle\env\lib\site-packages\diffusers\pipelines\pipeline_utils.py", line 924, in from_pretrained loaded_sub_model = load_sub_model( File "A:\AI_programs\InstantStyle\env\lib\site-packages\diffusers\pipelines\pipeline_loading_utils.py", line 725, in load_sub_model loaded_sub_model = load_method(os.path.join(cached_folder, name), **loading_kwargs) File "A:\AI_programs\InstantStyle\env\lib\site-packages\huggingface_hub\utils\_validators.py", line 114, in _inner_fn return fn(*args, **kwargs) File "A:\AI_programs\InstantStyle\env\lib\site-packages\diffusers\schedulers\scheduling_utils.py", line 157, in from_pretrained return cls.from_config(config, return_unused_kwargs=return_unused_kwargs, **kwargs) File "A:\AI_programs\InstantStyle\env\lib\site-packages\diffusers\configuration_utils.py", line 260, in from_config model = cls(**init_dict) File "A:\AI_programs\InstantStyle\env\lib\site-packages\diffusers\configuration_utils.py", line 665, in inner_init init(self, *args, **init_kwargs) File "A:\AI_programs\InstantStyle\env\lib\site-packages\diffusers\schedulers\scheduling_euler_discrete.py", line 239, in __init__ timesteps = torch.from_numpy(timesteps).to(dtype=torch.float32) RuntimeError: Numpy is not available A:\AI_programs>