-
Notifications
You must be signed in to change notification settings - Fork 1.1k
[Windows] [Windows] "Failed building wheel for llama-cpp-python #54
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.
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
Comments
Hey @kerbi yes that's correct you need to essentially build it on the system that you're going to be running it on in order for the compiler to detect processor specific optimizations (SSE, AVX, AVX2, etc). These optimizations can have a significant impact on performance. For debugging this cmake issue, can you try to seperately build the base llama.cpp library following the windows instructions here https://github.com/ggerganov/llama.cpp#usage If you discover the correct msvc version / magic incantation to build this for windows from source I'll be happy to add it to the docs. @MillionthOdin16 you're on a windows system do you have any suggestions on what I could add to the docs for setting up the right compiler on windows? |
I am having the exact same problem, "ERROR: Could not build wheels for llama-cpp-python which use PEP 517 and cannot be installed directly", do you know how to fix it specifically? |
The same problem. |
I have this issue in the ubuntu 18.04. Building wheels for collected packages: llama-cpp-python, peft × Building wheel for llama-cpp-python (pyproject.toml) did not run successfully.
note: This error originates from a subprocess, and is likely not a problem with pip. |
This is likely because you're building inside a virtual environment and HyperV has disabled Intel AVX extensions. See #244 |
The same problem while installing PrivateGPT on windows 10. :/
|
@riyanswat The error message says that it doesn't find any instance of Visual Studio (not to be confused with Visual Studio Code!). Can you try to install Visual Studio with C++ Build Tools and try it again? |
Please open a new ticket for this issue. Please also use the tags: < details > < / details > If you're going to paste pages of logs. |
Uh oh!
There was an error while loading. Please reload this page.
Edit : For now i've installed the wheel from "https://github.com/Loufe/llama-cpp-python/blob/main/wheels/llama_cpp_python-0.1.26-cp310-cp310-win_amd64.whl". The installation of the wheel works. So everything is fine for me. Got things working also in WSL with no issue.
I would still be happy to build the wheel myself, first as a learning experience, to understand what I did wrong, and secondly, because if I understood well from "#40", it might lead to better performance If I compile it myself ? "The issue is that the binaries will likely not be built with the correct optimizations for the users particular CPU which will likely result in much worse performance than the user expects."
Though maybe I did not understood correctly and it doesn't matter.
I leave the issue in case it might be useful to someone, or in case someone wants to try to help me build the wheel for fun.
Hi !
I've been trying to install this package for a while, but I can't get it working on windows.
When I run "pip install llama-cpp-python", I get the following errors :
(short version, i'll put the full output at the end of the message)
It seems that it is trying to find a C compiler and then build the wheel for the library (as far as I understand it).
At some point it seems to find one :
-- Trying 'Visual Studio 16 2019 x64 v142' generator - success
But then it seems to fail :
And then give me a very long output full of directory paths to explain to me why it failed.
Close to the end I can see an error that might be relevant :
But I don't really know what to make of it.
I would really love to understand how to make it work on windows, but I lack knowledge on building wheels.
I've tried :
- Upgrading pip and setup tools
- I installed Visual studio AND build tools for C++ (therefore I have cmake on my computer, but I don't know if it's even used when trying to build the wheel considering the previous output...)
- Find 'CMake's output for more information.', but I have no idea where to find it and google didn't helped me on that one.
- Downloading the repo and trying to build it using cmake, but maybe I did it wrong :
-Tried doing the same thing with cygwin after installing it
I've tried since yesterday to make it work but I can't figure it out. Is there someone that could help me get this working on my windows machine ?
Thank you very much in advance.
Additional info :
I'm trying to install it in a conda environment named 'textgen", but not sure it is relevant.
Full error output :
The text was updated successfully, but these errors were encountered: