Can't install prodigy Apple M1 using conda environment

Hello, I have a M1 Macbook and I'm using python inside a conda environment. But I'm still having issues installing prodigy. I've installed open-cv and tensor flow in this environment using wheel. I'm not sure how to fix this.

conda activate python_conda

(python_conda) veronica@Veronicas-Air Downloads % pip install ./prodigy*.whl

ERROR: prodigy-1.10.8-cp36.cp37.cp38-cp36m.cp37m.cp38-macosx_10_14_x86_64.whl is not a supported wheel on this platform.

The current prodigy wheels only support x86_64, so it's only possible to use them in emulation model on an M1. If you install conda for x86_64, I think it should work, e.g. miniforge for OS X x86_64: GitHub - conda-forge/miniforge: A conda-forge distribution. wait, miniforge has python 3.9 by default, so an better recommendation would be miniconda with python 3.8: Miniconda — Conda documentation

You should also make sure you're using python 3.6-3.8 and not python 3.9, which also isn't supported by this wheel. You can check the version and architecture for python in your conda env with:

python --version
python -c "import platform; print(platform.machine())"

We are planning to include both python 3.9 and OS X arm64 wheels for the upcoming v1.11 release that will switch to spaCy v3. I'm not sure whether python 3.9 or arm64 wheels are planned for v1.10 at this point, but we do have the basic infrastructure to support this now.

1 Like

The plan is to introduce the new wheels for Python versions and architectures for v1.11, alongside the simpler PyPi download server :slightly_smiling_face:

(The v1.10.x wheels are currently built as combined wheels for all Python versions, which would get pretty messy if we wanted to add more versions and architectures – but we also don't want to change the distribution format between patch releases.)

1 Like

@adriane @ines Thank you I had to install miniconda, miniforge doesn't seem to work for this.