Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

please upload windows install protable version #19

Open
rakib91221 opened this issue Jan 29, 2025 · 20 comments
Open

please upload windows install protable version #19

rakib91221 opened this issue Jan 29, 2025 · 20 comments
Labels
enhancement New feature or request help wanted Extra attention is needed

Comments

@rakib91221
Copy link

please upload windows install protable version or one click install version..thanks
@sdbds sdbds mentioned this issue Jan 29, 2025
Closed
@VigibotDev
Copy link

Amazing project. I try it on Windows 1 clic before putting a GPU inside my Debian server but I get :


uv 0.5.25 (9c07c3fc5 2025-01-28)
uv installed|UV模块已安装.
Page de codes active : 65001
UV cache directory already exists, skipping disk space check
Windows .venv
Installing main requirements
Resolved 2 packages in 376ms
Audited 2 packages in 0.02ms
Resolved 98 packages in 1.71s
  x Failed to build `antlr4-python3-runtime==4.9.3`
  |-> The build backend returned an error
  `-> Call to `setuptools.build_meta:__legacy__.build_wheel` failed (exit code: 1)

      [stderr]
      Traceback (most recent call last):
        File "<string>", line 8, in <module>
        File "D:\A\IA\YuE-for-windows-main\.venv\lib\site-packages\setuptools\__init__.py", line 8, in <module>
          import _distutils_hack.override  # noqa: F401
        File "D:\A\IA\YuE-for-windows-main\.venv\lib\site-packages\_distutils_hack\override.py", line 1, in <module>
          __import__('_distutils_hack').do_override()
        File "D:\A\IA\YuE-for-windows-main\.venv\lib\site-packages\_distutils_hack\__init__.py", line 70, in
      do_override
          ensure_local_distutils()
        File "D:\A\IA\YuE-for-windows-main\.venv\lib\site-packages\_distutils_hack\__init__.py", line 56, in
      ensure_local_distutils
          core = importlib.import_module('distutils.core')
        File "C:\Users\Pascal\AppData\Roaming\uv\python\cpython-3.10.16-windows-x86_64-none\lib\importlib\__init__.py",
      line 126, in import_module
          return _bootstrap._gcd_import(name[level:], package, level)
        File "D:\A\IA\YuE-for-windows-main\.venv\lib\site-packages\setuptools\_distutils\core.py", line 14, in
      <module>
          from .config import PyPIRCCommand
        File "D:\A\IA\YuE-for-windows-main\.venv\lib\site-packages\setuptools\_distutils\config.py", line 7, in
      <module>
          import email.message
        File "C:\Users\Pascal\AppData\Roaming\uv\python\cpython-3.10.16-windows-x86_64-none\lib\email\message.py",
      line 15, in <module>
          from email import utils
        File "C:\Users\Pascal\AppData\Roaming\uv\python\cpython-3.10.16-windows-x86_64-none\lib\email\utils.py", line
      29, in <module>
          import socket
        File "C:\Users\Pascal\AppData\Roaming\uv\python\cpython-3.10.16-windows-x86_64-none\lib\socket.py", line 51,
      in <module>
          import _socket
      ImportError: Module use of python39.dll conflicts with this version of Python.

      hint: This usually indicates a problem with the package or the build environment.
Install main requirements failed
Install failed|安装失败。

@pointave
Copy link

pointave commented Jan 30, 2025 •
edited
Loading

I think since it uses flash-attention you need WSL. Its a lot slower than they said, took like 20 minutes for 1 minute of music and lyrics while using 18gb of vram.

@sdbds
Copy link

sdbds commented Jan 30, 2025

Amazing project. I try it on Windows 1 clic before putting a GPU inside my Debian server but I get :令人惊叹的项目。我在将 GPU 放入我的 Debian 服务器之前，先在 Windows 上试了一下，但得到：


uv 0.5.25 (9c07c3fc5 2025-01-28)
uv installed|UV模块已安装.
Page de codes active : 65001
UV cache directory already exists, skipping disk space check
Windows .venv
Installing main requirements
Resolved 2 packages in 376ms
Audited 2 packages in 0.02ms
Resolved 98 packages in 1.71s
  x Failed to build `antlr4-python3-runtime==4.9.3`
  |-> The build backend returned an error
  `-> Call to `setuptools.build_meta:__legacy__.build_wheel` failed (exit code: 1)

      [stderr]
      Traceback (most recent call last):
        File "<string>", line 8, in <module>
        File "D:\A\IA\YuE-for-windows-main\.venv\lib\site-packages\setuptools\__init__.py", line 8, in <module>
          import _distutils_hack.override  # noqa: F401
        File "D:\A\IA\YuE-for-windows-main\.venv\lib\site-packages\_distutils_hack\override.py", line 1, in <module>
          __import__('_distutils_hack').do_override()
        File "D:\A\IA\YuE-for-windows-main\.venv\lib\site-packages\_distutils_hack\__init__.py", line 70, in
      do_override
          ensure_local_distutils()
        File "D:\A\IA\YuE-for-windows-main\.venv\lib\site-packages\_distutils_hack\__init__.py", line 56, in
      ensure_local_distutils
          core = importlib.import_module('distutils.core')
        File "C:\Users\Pascal\AppData\Roaming\uv\python\cpython-3.10.16-windows-x86_64-none\lib\importlib\__init__.py",
      line 126, in import_module
          return _bootstrap._gcd_import(name[level:], package, level)
        File "D:\A\IA\YuE-for-windows-main\.venv\lib\site-packages\setuptools\_distutils\core.py", line 14, in
      <module>
          from .config import PyPIRCCommand
        File "D:\A\IA\YuE-for-windows-main\.venv\lib\site-packages\setuptools\_distutils\config.py", line 7, in
      <module>
          import email.message
        File "C:\Users\Pascal\AppData\Roaming\uv\python\cpython-3.10.16-windows-x86_64-none\lib\email\message.py",
      line 15, in <module>
          from email import utils
        File "C:\Users\Pascal\AppData\Roaming\uv\python\cpython-3.10.16-windows-x86_64-none\lib\email\utils.py", line
      29, in <module>
          import socket
        File "C:\Users\Pascal\AppData\Roaming\uv\python\cpython-3.10.16-windows-x86_64-none\lib\socket.py", line 51,
      in <module>
          import _socket
      ImportError: Module use of python39.dll conflicts with this version of Python.

      hint: This usually indicates a problem with the package or the build environment.
Install main requirements failed
Install failed|安装失败。

It appears that your Python module is 3.9, but this script uses Python 3.10.
If you have created a venv in advance, please delete it and then run the script again.

@sdbds
Copy link

sdbds commented Jan 31, 2025

Image

I just made a simple UI, which has been successfully tested on Windows. Welcome to test.

@rakib91221
Copy link
Author

rakib91221 commented Jan 31, 2025 •
edited
Loading

Image

I just made a simple UI, which has been successfully tested on Windows. Welcome to test.

thanks..where windows installer link? and this ui work rtx3060-12GB gpu

@SoftologyPro
Copy link

SoftologyPro commented Jan 31, 2025 •
edited
Loading

I now include YuE with Visions of Chaos along with hundreds of other AI systems if anyone wants to have an eaiser install on Windows.
https://softology.pro/voc.htm

Image

@DarkAlchy
Copy link

Too bad it requires 3.10.x as I just moved off of that about two weeks ago for a few of my programs.

@SoftologyPro
Copy link

Too bad it requires 3.10.x as I just moved off of that about two weeks ago for a few of my programs.

If you mean Visions of Chaos you can point it to a 3.10 directory so it uses 3.10 while your other programs use 3.12

@DarkAlchy
Copy link

yeah, I will just wait until it is updated. If it were in a venv pip install environment, I would do it. Still, want a way to get this multimodal to work on windows by itself for now.

@sdbds
Copy link

sdbds commented Feb 2, 2025

Too bad it requires 3.10.x as I just moved off of that about two weeks ago for a few of my programs.

UV will automatically install version 3.10x, you do not need to download Python, it will install it automatically

@DarkAlchy
Copy link

To activate the ML section it said I had to do it, and since I didn't create a venv for it, I uninstalled it since I have had other Python programs zap my system-wide environment (once bitten twice shy). I did manage to get the MM working anyway.

@SoftologyPro
Copy link

To activate the ML section it said I had to do it, and since I didn't create a venv for it, I uninstalled it since I have had other Python programs zap my system-wide environment (once bitten twice shy). I did manage to get the MM working anyway.

Understood. I have had other installs kill my main python env too without asking. That is why every mode in Visions of Chaos creates their own virtual envs to run under. You basically have to do that anyway when dealing with so many different systems that all need their own unique versions of Python packages. If you ever want to try it again join my Discord and I can give you some tips about keeping the version of Python and CUDA VoC uses separate from your defaults (all you have to do is install the VoC required versions, not add them to your path, and then point to the install folders in VoC, it handles the rest).

@DarkAlchy
Copy link

Will do. My hope is YuE gets a much-needed optimization because it needs it. Took about 20m for 90s of music. Music was of poor quality, but a 64kbps mp3 didn't help that. I am so glad this is finally hitting local, but not only do we need it optimized so it doesn't take lifetimes, but we need a way to train it so it can learn different styles. For instance, even UDIO struggles with my main style even if I upload it an example. Now if I could train it then it would no longer struggle.

@sdbds
Copy link

sdbds commented Feb 2, 2025

https://huggingface.co/onnx-community/YuE-s2-1B-general-ONNX

stage2 can use ONNX for faster

@DarkAlchy
Copy link

q4 too.

@Universeal13
Copy link

Image

I just made a simple UI, which has been successfully tested on Windows. Welcome to test.

Where can we find it for a test?

@SoftologyPro
Copy link

https://huggingface.co/onnx-community/YuE-s2-1B-general-ONNX

stage2 can use ONNX for faster

How do we use this model?
Changing the
--stage2_model m-a-p/YuE-s2-1B-general
argument to
--stage2_model onnx-community/YuE-s2-1B-general-ONNX
does not work and gives the error
OSError: onnx-community/YuE-s2-1B-general-ONNX does not appear to have a file named pytorch_model.bin, model.safetensors, tf_model.h5, model.ckpt or flax_model.msgpack.

@DarkAlchy
Copy link

When I last used Onnx I had to compile it as it is compiled. As I was told Onnx cannot be universal it must be compiled on the machine it is to be ran on from source files. I do forget what the source format was, but was super easy to compile, but I see no files we can compile from.

When I last used Onnx it was for video upscalers that I found the time to initialize was longer than the savings and ditched Onnx. That would not be the case for this since it takes so long for Stage2.

@SoftologyPro
Copy link

When I last used Onnx I had to compile it as it is compiled. As I was told Onnx cannot be universal it must be compiled on the machine it is to be ran on from source files. I do forget what the source format was, but was super easy to compile, but I see no files we can compile from.

When I last used Onnx it was for video upscalers that I found the time to initialize was longer than the savings and ditched Onnx. That would not be the case for this since it takes so long for Stage2.

Ok. I thought your comment was that you had used the onyx model for a speedup in stage 2.

@DarkAlchy
Copy link

No, I was letting you know something is amiss because we need to compile that into Onnx as Onnx is machine dependant, and unless something changed my Onnx only works for machines that match my specs exactly. Compiling them is really fast too on a 4090.

I think the one linking the Onnx is just trying to let us know it can be done not that those can be used. Besides, I know in ComfyUI it takes a special node to load Onnx models, and I expect that to be a universal case that the code will need to be updated to take advantage of it.

Oh, another thing I really think will help is Sage Attention as it blew away SDPA, and FA2, in Linux. In Windows I don't see much of an advantage, but with this who knows. Needs to be coded for that as well.

What I am basically saying is that this is ripe to be ravished to optimize and I bet we see 10x speed up, at least. Maybe 2m for a complete song on a 4090 instead of 20-30m.

@a43992899 a43992899 added enhancement New feature or request help wanted Extra attention is needed labels Feb 5, 2025
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees
No one assigned
Labels
enhancement New feature or request help wanted Extra attention is needed
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
8 participants
@VigibotDev @sdbds @DarkAlchy @a43992899 @SoftologyPro @pointave @rakib91221 @Universeal13