In this repo, i am trying to create enviroments for reproducible data science that fit my needs
Most development happens on MacOS aarch64-darwin, the actual runs usually happen on x86_64-linux.
- Working CUDA environment with pytorch (and transformers) (whereas cuda on linux and CPU version on Mac)
- R environment with MASS, dplyr and box (at minimum)
- spacy package: spacy on nix only works on MacOS until 3.7.4
- ibis-framework: is randomly broken because of its dependencies
- dm-tree
- wandb