pkgs/development/python-modules/cleanlab/default.nix

   1 {
   2   lib,
   3   buildPythonPackage,
   4   fetchFromGitHub,
   5
   6   # build-system
   7   setuptools,
   8
   9   # dependencies
  10   numpy,
  11   scikit-learn,
  12   termcolor,
  13   tqdm,
  14   pandas,
  15
  16   # test dependencies
  17   cleanvision,
  18   datasets,
  19   fasttext,
  20   hypothesis,
  21   keras,
  22   matplotlib,
  23   pytestCheckHook,
  24   pytest-lazy-fixture,
  25   skorch,
  26   tensorflow,
  27   torch,
  28   torchvision,
  29   wget,
  30 }:
  31
  32 buildPythonPackage rec {
  33   pname = "cleanlab";
  34   version = "2.7.0";
  35   pyproject = true;
  36
  37   src = fetchFromGitHub {
  38     owner = "cleanlab";
  39     repo = "cleanlab";
  40     rev = "refs/tags/v${version}";
  41     hash = "sha256-0kCEIHNOXIkdwDH5zCVWnR/W79ppc/1PFsJ/a4goGzk=";
  42   };
  43
  44   build-system = [ setuptools ];
  45
  46   dependencies = [
  47     numpy
  48     scikit-learn
  49     termcolor
  50     tqdm
  51     pandas
  52   ];
  53
  54   # This is ONLY turned off when we have testing enabled.
  55   # The reason we do this is because of duplicate packages in the enclosure
  56   # when using the packages in nativeCheckInputs.
  57   # Affected packages: grpcio protobuf tensorboard tensorboard-plugin-profile
  58   catchConflicts = (!doCheck);
  59   doCheck = true;
  60
  61   nativeCheckInputs = [
  62     cleanvision
  63     datasets
  64     fasttext
  65     hypothesis
  66     keras
  67     matplotlib
  68     pytestCheckHook
  69     pytest-lazy-fixture
  70     skorch
  71     tensorflow
  72     torch
  73     torchvision
  74     wget
  75   ];
  76
  77   disabledTests = [
  78     # Requires the datasets we prevent from downloading
  79     "test_create_imagelab"
  80   ];
  81
  82   disabledTestPaths = [
  83     # Requires internet
  84     "tests/test_dataset.py"
  85     # Requires the datasets we just prevented from downloading
  86     "tests/datalab/test_cleanvision_integration.py"
  87     # Fails because of issues with the keras derivation
  88     "tests/test_frameworks.py"
  89   ];
  90
  91   meta = {
  92     description = "Standard data-centric AI package for data quality and machine learning with messy, real-world data and labels";
  93     homepage = "https://github.com/cleanlab/cleanlab";
  94     changelog = "https://github.com/cleanlab/cleanlab/releases/tag/v${version}";
  95     license = lib.licenses.agpl3Only;
  96     maintainers = with lib.maintainers; [ happysalada ];
  97   };
  98 }