Missing requirements-nvidia.txt?

[Karlmeister]

Hey Yuki, how are you doing? I'm trying to install and setup Yuna (kinda a rough path for an ol' ruster developer like me). I could address some issues that I had installing CPU dependencies, but now I'm trying NVIDIA GPU dependencies and got this error:

========== Installation... ==========
1. CPU
2. NVIDIA GPU
3. AMD GPU
4. Metal
5. Go back
> 2
Installing NVIDIA dependencies...
Requirement already satisfied: llama-cpp-python in c:\users\karlmeister\appdata\local\programs\python\python312\lib\site-packages (0.3.1)
Requirement already satisfied: typing-extensions>=4.5.0 in c:\users\karlmeister\appdata\local\programs\python\python312\lib\site-packages (from llama-cpp-python) (4.12.2)
Requirement already satisfied: numpy>=1.20.0 in c:\users\karlmeister\appdata\local\programs\python\python312\lib\site-packages (from llama-cpp-python) (1.26.4)
Requirement already satisfied: diskcache>=5.6.1 in c:\users\karlmeister\appdata\local\programs\python\python312\lib\site-packages (from llama-cpp-python) (5.6.3)
Requirement already satisfied: jinja2>=2.11.3 in c:\users\karlmeister\appdata\local\programs\python\python312\lib\site-packages (from llama-cpp-python) (3.1.4)
Requirement already satisfied: MarkupSafe>=2.0 in c:\users\karlmeister\appdata\local\programs\python\python312\lib\site-packages (from jinja2>=2.11.3->llama-cpp-python) (3.0.2)
ERROR: Could not open requirements file: [Errno 2] No such file or directory: 'requirements-nvidia.txt'

Checking the code in index.sh, for CPU dependencies installer is relying in existing file requirements.txt but analog file 'requirements-nvidia.txt' for GPU dependencies doesn't exist in the root folder of the project.

I'm missing something or this file is from another project/repo?

Thank you very much!

PD: To new Windows users (I guess the same happens in Linux and Mac), don't use Python 3.13 for now... you'll get a "deprecated warning" over a python library while installing CPU dependencies and won't be able to complete installation.

[cloner174]

Hi!

I hope you are doing well.

Please check the following path:
lib/utils/docker/

You will find the file requirements-nvidia.txt there. I recommend installing it manually.

[yukiarimo]

I see. Gonna combine all stuff soon. But you can also install it with a Docker, if you want, should work :)

[Karlmeister]

Thank you, bruh! I copied the file to the root folder and could install all the GPU Nvidia dependencies!
Anyway, I'm stuck right now in the frontend with Yuna.... i send her messages but she doesn't answer...
BTW all of this was with a "direct install" over my Windows 11. After that, I followed your advice and tried to install it with Docker. I followed the instructions in \lib\utils\docker\readme.md, but now I'm stuck trying to start the container... looks like an error ocurrs when it tries to install/load the torchaudio library :(

2024-11-03 22:40:22 Container image Copyright (c) 2016-2023, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
2024-11-03 22:40:22 
2024-11-03 22:40:22 This container image and its contents are governed by the NVIDIA Deep Learning Container License.
2024-11-03 22:40:22 By pulling and using the container, you accept the terms and conditions of this license:
2024-11-03 22:40:22 https://developer.nvidia.com/ngc/nvidia-deep-learning-container-license
2024-11-03 22:40:21 Traceback (most recent call last):
2024-11-03 22:40:21   File "/app/index.py", line 5, in <module>
2024-11-03 22:40:21     from lib.router import handle_history_request, handle_image_request, handle_message_request, handle_audio_request, services, handle_search_request, handle_textfile_request
2024-11-03 22:40:21   File "/app/lib/router.py", line 7, in <module>
2024-11-03 22:40:21     from lib.vision import capture_image, create_image
2024-11-03 22:40:21   File "/app/lib/vision.py", line 4, in <module>
2024-11-03 22:40:21     from lib.audio import speak_text
2024-11-03 22:40:21   File "/app/lib/audio.py", line 5, in <module>
2024-11-03 22:40:21     import torchaudio
2024-11-03 22:40:21   File "/usr/local/lib/python3.10/dist-packages/torchaudio/__init__.py", line 2, in <module>
2024-11-03 22:40:21     from . import _extension  # noqa  # usort: skip
2024-11-03 22:40:21   File "/usr/local/lib/python3.10/dist-packages/torchaudio/_extension/__init__.py", line 38, in <module>
2024-11-03 22:40:21     _load_lib("libtorchaudio")
2024-11-03 22:40:21   File "/usr/local/lib/python3.10/dist-packages/torchaudio/_extension/utils.py", line 60, in _load_lib
2024-11-03 22:40:21     torch.ops.load_library(path)
2024-11-03 22:40:21   File "/usr/local/lib/python3.10/dist-packages/torch/_ops.py", line 1032, in load_library
2024-11-03 22:40:21     ctypes.CDLL(path)
2024-11-03 22:40:21   File "/usr/lib/python3.10/ctypes/__init__.py", line 374, in __init__
2024-11-03 22:40:21     self._handle = _dlopen(self._name, mode)
2024-11-03 22:40:21 OSError: /usr/local/lib/python3.10/dist-packages/torchaudio/lib/libtorchaudio.so: undefined symbol: _ZNK3c105Error4whatEv

I'll try to figure out what is happening in both scenarios. In the meantime, any help will be appreciated!

[yukiarimo]

Good luck