search

GT4SD / gt4sd-core

GT4SD, an open-source library to accelerate hypothesis generation in the scientific discovery process.
https://gt4sd.github.io/gt4sd-core/
MIT License
336 stars 74 forks source link

Can't install the package with conda or from source #230

Closed bransom960 closed 1 year ago

bransom960 commented 1 year ago

Following the instructions in the documentation for the installation, the installation doesn't finish. When I try to build the package from source, the installation isn't able to function in a notebook.

I tried 3 different ways of installing

Using the conda installation in the documentation, after: conda env create -f conda.yml

I get this error: ResolvePackageNotFound:

Even though I have this package installed

when I run: python setup.py install

I get: pkg_resources.extern.packaging.requirements.InvalidRequirement: . suffix can only be used with == or != operators torch (>=1.9.)

when I run: python setup.py build pip install gt4sd

It seems to make it through the installation, but then I get this error in the example jupyter notebooks: File ~/anaconda3/lib/python3.10/site-packages/gt4sd/training_pipelines/moses/organ/core.py:32 29 from dataclasses import dataclass, field 30 from typing import Any, Dict ---> 32 from ..guacamol_baselines.moses_baselines.organ_train import main 33 from moses.script_utils import MetricsReward 35 from ...core import TrainingPipelineArguments ModuleNotFoundError: No module named 'gt4sd.training_pipelines.moses.guacamol_baselines'

And I have tried to begin to go down the rabbit hole of updating various packages and installing them, but after adjusting a few scripts I stopped.

System (please complete the following information):

drugilsberg commented 1 year ago

Hi @bransom960, unfortunately support for M1 is not yet there given issues with some does. We are tracking the issue here: https://github.com/GT4SD/gt4sd-core/issues/200.

The recommended solution is to use conda to install gt4sd in a x86 environment as described here: https://towardsdatascience.com/python-conda-environments-for-both-arm64-and-x86-64-on-m1-apple-silicon-147b943ffa55?gi=f8e4e3b2ece3.

Closing since it is a duplicate.