pip install fairseq2
Collecting fairseq2
Obtaining dependency information for fairseq2 from https://files.pythonhosted.org/packages/cd/27/46c14e28e8cb0aa602660ce64d4547a37f460d382e4fcf94f2a53d47e5b0/fairseq2-0.1.0-py3-none-any.whl.metadata
Using cached fairseq2-0.1.0-py3-none-any.whl.metadata (1.2 kB)
INFO: pip is looking at multiple versions of fairseq2 to determine which version is compatible with other requirements. This could take a while.
ERROR: Could not find a version that satisfies the requirement fairseq2n==0.1.0 (from fairseq2) (from versions: none)
ERROR: No matching distribution found for fairseq2n==0.1.0
Hey folks, I am working on v0.1.1 right now that will fix some of the issues you have been experiencing. Two side notes though:
I expect to finish v0.1.1 by tomorrow (08/29), and will keep you posted!
Attempting to build from source also fails with a
tbb-devel
requirement failure:I’ve gotten through a few hurdles on Windows, but it’ll take me a few days to really get back to it:
I’ll bundle it up at some point when I’ve gotten it deployed.
For other Mac users, I created two script to build fairseq2 on the Mac. Tested on a M2 chipset. https://github.com/chipcode-nl/mac-fairseq2
Hey everyone, yesterday I released v0.1.1 that tries to address some of the issues you have been experiencing. I also significantly revised the installation instructions in README. We also have a pre-built package for Apple-silicon computers now. Please try the installation in a fresh environment by following the instructions in README and let me know how it goes. Thank you!
@rgryta we have couple places in our code base where we rely on fairseq2n. For instance we have a PyTorch native implementation of sentencepiece written in C++ and most of our models use that implementation for their tokenizers. Having said that, depending on users’ feedback, I will check whether I can refactor it even further and make it an optional dependency.
@huangwei2913 this dude made a repo, I just followed the steps in the build bash file and it worked. Thanks for @chipcode-nl !
Hello, Same problem for me on :
ERROR: Could not find a version that satisfies the requirement fairseq2n==0.1.0 (from fairseq2) (from versions: none) ERROR: No matching distribution found for fairseq2n==0.1.0 (from fairseq2)
using macOS. both installing from pip as well as
returns an error for ERROR: Could not find a version that satisfies the requirement fairseq2n==0.1.0 (from fairseq2) (from versions: none) ERROR: No matching distribution found for fairseq2n==0.1.0
Thank you @silentlit for your assistance. A friend recommended that I try miniforge instead of virtualenvwrapper. This suggestion worked well, and I successfully installed fairseq2.
I find that, it might because python recognize macos version error. Activating your python env, import platform, and run platform.platform(). If the print is not equal to your macos version, this error would happen. Reboot and use the Terminal to create python env might fit this problem. It work for me.
windows 10 enterprise, python 3.10.5, the fairseq2 is not installed in my PC. If I install this package on Ubuntu, is it possible?
Tested this option on Mac M2 and Mac Intel both compiled.
For someone use Linux, I just download it by upgrade pip:
That work for me
Python 3.11.4 Apple M1
Same. Apple Silicon, M1 Python 3.11
Following the Install from Source instructions:
pip install .
ERROR: Could not find a version that satisfies the requirement fairseq2n==0.1.0 (from fairseq2) (from versions: none)
ERROR: No matching distribution found for fairseq2n==0.1.0
Same problem on
Darwin Antons-MacBook-Pro.local 22.6.0 Darwin Kernel Version 22.6.0: Wed Jul 5 22:22:05 PDT 2023; root:xnu-8796.141.3~6/RELEASE_ARM64_T6000 arm64
MacBook Pro m1 pro; macOS 13.5
pip 23.0.1 from /opt/homebrew/lib/python3.11/site-packages/pip (python 3.11)
Python 3.11.4
Same problem here - can’t install fairseq2 on Win, Python 3.10, torch 2.0.1. Hope this will be solved soon.
Not complaining, by any means. You guys have been doing tremendously good work. I guess a temporary workaround for people without Linux would be to install a dummy fairseq2n module before installing fairseq2 wheel (from what I saw those are for any os) and then uninstall the former.