Skip to content

TypeError in Cython code on numba==0.63.1 and causalml==0.15.4 #859

@CloseChoice

Description

@CloseChoice

Describe the bug
When importing causalml.dataset on the 0.15.4 release we are experiencing the following failure:

TypeError: C variable sklearn.utils._random.DEFAULT_SEED has wrong signature (expected __pyx_t_7sklearn_5utils_9_typedefs_uint32_t, got __pyx_t_7sklearn_5utils_9_typedefs_uint32_t const )

To Reproduce
Steps to reproduce the behavior:

# uv pip install numba==0.63.1
# uv pip install causalml==0.15.4   # order matters here!
import causalml.dataset

Expected behavior
Should pass

Environment (please complete the following information):

  • OS: Ubuntu, MacOS
  • Python Version: 3.13, 3.14
  • Versions of Major Dependencies: pandas==2.3.3, scikit-learn==1.8.0, cython==3.2.2

Additional context
We experience this in the SHAP pipelines: https://github.com/shap/shap/actions/runs/20113311832/job/57716416017#step:11:15.
We could also upgrade to the latest release of causalml if that supports scipy>=1.16 (a requirement of ours), since the bug seems to be fixed on causalml==0.15.5

Metadata

Metadata

Assignees

No one assigned

    Labels

    bugSomething isn't working

    Type

    No type
    No fields configured for issues without a type.

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions