pip3 install fairseq2 - doesn't work

[twerff]

I'm trying to install seamlessm4q on my Macbook M2. Installation results in an error related to fairseq2.
So, I tried to manually install fairseq using % pip3 install fairseq2. It gives the following error:

Collecting fairseq2
  Downloading fairseq2-0.2.1-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.
  Downloading fairseq2-0.2.0-py3-none-any.whl.metadata (1.2 kB)
  Downloading fairseq2-0.1.1-py3-none-any.whl.metadata (1.2 kB)
  Downloading fairseq2-0.1.0-py3-none-any.whl.metadata (1.2 kB)
ERROR: Cannot install fairseq2==0.1.0, fairseq2==0.1.1, fairseq2==0.2.0 and fairseq2==0.2.1 because these package versions have conflicting dependencies.

The conflict is caused by:
    fairseq2 0.2.1 depends on fairseq2n==0.2.1
    fairseq2 0.2.0 depends on fairseq2n==0.2.0
    fairseq2 0.1.1 depends on fairseq2n==0.1.1
    fairseq2 0.1.0 depends on fairseq2n==0.1.0

To fix this you could try to:
1. loosen the range of package versions you've specified
2. remove package versions to allow pip attempt to solve the dependency conflict

ERROR: ResolutionImpossible: for help visit https://pip.pypa.io/en/latest/topics/dependency-resolution/#dealing-with-dependency-conflicts

Can anyone help me out?

[MountainGG]

same problem on ubuntu22.04(wsl2)

[AjayArvind2207]

Same issue on Ubuntu-24.04(wsl2). Tried installing from source too but no luck. Any advice would be helpful!

[AjayArvind2207]

@MountainGG I managed to fix mine, turns out I was on python 3.12 and fairseq2 doesnt seem to be supported on it yet, so I just downgraded my virtual env to python3.10 and I was able to install it. Hope this helps!

[yinde0]

I have downgraded mine to python 3.90 and I am still getting the same error.

[silentlit]

Check the python env:
import platform
platform.platform() or platform.processor()
If you got 'macOS-10.16-x86_64-i386-64bit' or 'i386', that is the env matter.

You could try to create a new env like:
CONDA_SUBDIR=osx-arm64 conda create -n ENV_NAME python=3.XX
and you would create a arm env. Try to check the env import platform;platform.platform();platform.processor(). If the env correct, you will got 'macOS-14.5-arm64-arm-64bit' and 'arm'.

Reference:
MacOS M1 system is detected as ARM by Python package even though I'm using Rosetta