[feat] bump graphlearn to 1.3.8, drop numpy<2 pin#557
Closed
tiankongdeguiji wants to merge 2 commits into
Closed
Conversation
…post1) graphlearn 1.3.7 and faiss-gpu-cu12 1.11.0 were the deps pinning numpy<2. - graphlearn 1.3.7 -> 1.3.8 (declares no numpy bound). - faiss-gpu-cu12 1.11.0 -> 1.14.1.post1: 1.11.0 requires numpy<2; faiss became numpy-2 ready at 1.13.2 (numpy>=2,<3). Source moves from the OSS-mirrored wheel to PyPI (matching how faiss-cpu is already sourced); its nvidia-cuda-runtime/cublas constraints are unchanged (>=12.1), so the CUDA stack is unaffected. Require numpy>2 first to exercise the numpy 2.x path through CI before relaxing the pin entirely. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01NPSTfsBJfgwD1Y6afhyVzg
tiankongdeguiji
force-pushed
the
feat/graphlearn-1.3.8-numpy2
branch
from
8417db8 to
6250267
Compare
…faiss via [gpu] - tzrec/datasets/sampler.py: numpy 2.0 removed np.string_; use np.bytes_ (same 'S' dtype, np.char.decode unchanged) when parsing graphlearn string node attrs. Fixes AttributeError crashing the negative/hard-negative/TDM sampler DataLoader workers under numpy 2. - .github/workflows/buildtest_ci.yml: install the built wheel's [gpu] extra so faiss-gpu-cu12 upgrades to the numpy-2 build. Plain `pip install <wheel>` only pulls install_requires (numpy>2, no faiss), bumping numpy to 2.x while leaving the image's numpy-1.x faiss 1.11.0 -> `import tzrec` crashed with "numpy.core.multiarray failed to import". Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01NPSTfsBJfgwD1Y6afhyVzg
Collaborator
Author
|
Superseded by #551 (the 1.3.0 bump), which already includes these changes: numpy 2, faiss upgraded to a numpy-2 build (1.14.3+cu12), graphlearn 1.3.8, |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What
graphlearn1.3.7 → 1.3.8 (new wheels underthird_party/graphlearn/).numpy<2pin.Why
graphlearn 1.3.7 was the only dependency forcing
numpy<2. graphlearn 1.3.8adds numpy>=2 support, so the pin is no longer needed.
Verification
Staged so CI proves numpy 2.x compatibility before the pin is relaxed:
numpy>2to force the numpy 2.x path through the full CImatrix (GPU / CPU / build / codestyle / pytyping).
numpy.🤖 Generated with Claude Code