I have to use @pypi as I want to use a private package: https://github.com/mycompany/earth_shattering_repo.git So I tried: @pypi(packages={ “pyodbc”:“5.1.0", “git+https://github.com/mycompany/earth_shattering_repo.git” : “”, “pandas”: “2.2", “pyarrow”: “>=10.0.1,<10.1.0", “numpy”: “1.26.4", “pandas”: “2.2", “scikit-learn”: “1.3.2", “lightgbm”: “4.5.0", “pyarrow”: “>=10.0.1,<10.1.0", “s3fs”: “2024.3.1", “xgboost”: “2.0.3" }) but get: Could not find a binary distribution for lightgbm==4.5.0 for the platform linux-64 when trying to run in our AWS cloud environment. I understand that this could potentially be resolved using: https://github.com/Netflix/metaflow-nflx-extensions/tree/fix/git-pkg-conda-lock meaning I could get lightgbm etc. via @conda_base at the same time (thanks @dry-beach-38304). I am kind of reluctant to use this for now. Sorry for being so stupid but is there a way to determine the binary versions of lightgbm for linux-64 and also osx-arm64? Thanks!

Not sure what you mean by “determine the binary version” for both platforms. A solution though would be to have your docker include the library directly. Why are you reluctant to use the extension though?