Skip to content

Tutorials/Accelerated Python: Install cuDF and cuML in the main environment#154

Open
brycelelbach wants to merge 1 commit intomainfrom
fix/remove-rapids-venv
Open

Tutorials/Accelerated Python: Install cuDF and cuML in the main environment#154
brycelelbach wants to merge 1 commit intomainfrom
fix/remove-rapids-venv

Conversation

@brycelelbach
Copy link
Collaborator

Summary

  • Removes the separate RAPIDS Python venv and its Jupyter kernel registration from the accelerated-python Dockerfile, reducing container size by eliminating a duplicate Python environment with ipykernel/pytest.
  • Installs cudf-cu13==26.2.1 and cuml-cu13==26.2.0 directly in the main environment via requirements.txt, now that the numba-cuda compatibility issue is resolved.
  • Switches all 5 notebooks that used the Python 3 (RAPIDS 25.10) kernel to the standard python3 kernel, and runs RAPIDS tests with regular pytest instead of the venv pytest.
  • Removes RAPIDS kernelspec handling from test-notebook-format.py so it no longer accepts or mentions the RAPIDS kernel as an alternative.

@brycelelbach brycelelbach force-pushed the fix/remove-rapids-venv branch from e5c625d to 93e79e6 Compare March 11, 2026 16:42
@brycelelbach brycelelbach enabled auto-merge (rebase) March 11, 2026 18:56
…onment.

cuDF and cuML no longer have a compatibility issue with numba-cuda, so
they can be installed directly in the main Python environment instead
of an isolated RAPIDS venv. This removes the separate venv, its Jupyter
kernel registration, and the duplicate ipykernel/pytest installs,
reducing container size. Notebooks that used the RAPIDS kernel now use
the standard python3 kernel, and the format checker no longer accepts
or mentions the RAPIDS kernelspec.
@brycelelbach brycelelbach force-pushed the fix/remove-rapids-venv branch from 93e79e6 to 54480fb Compare March 11, 2026 21:09
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant