Small fix - NumPy compat and LightGCN eval mode

This is a small bugfix / cleanup / compat fix for LensKit 2025.6.

What's Changed

• Remove tests that break with NumPy 2.4 ArrayLike changes by @mdekstrand in #977
• Put LightGCN in eval mode after training (2025) by @mdekstrand in #975
• Drop failing x86 wheel builds (2025) by @mdekstrand in #979

Full Changelog: v2025.6.2...v2025.6.3

NDCG bugfix with missing data

Fixes a bug where NDCG would divide by zero when there are no test items for the user.

What's Changed

• Resolves #956: Fix NDCG ZeroDivisionError by @samiravaez in #957
• Add E-R diagram to illustrate the data model by @mdekstrand in #962

Full Changelog: v2025.6.1...v2025.6.2

Bug + missing feature update for 2025.6

This is a quick bug-fix and missing-feature release for 2025.6.

6th 2025 Improvement

Another feature release for 2025, with some bug fixes as well.

See the release notes for user-friendly details.

What's Changed

• Update PyArrow to 22.x by @mdekstrand in #914
• Build Conda packages on MacOS Intel by @mdekstrand in #913
• Update locked dependencies and support Python 3.14 by @mdekstrand in #915
• CI build and test coverage updates by @mdekstrand in #919
• FA*IR: bug fixes and test updates by @samiravaez in #918
• ItemKNN: return neighbor counts + disable scoring parallelism by @mdekstrand in #922
• Add rank_biased_precision function that takes explicit weights by @sushobhan2024 in #923
• Add unified object representation function and more str/repr implementations by @mdekstrand in #924
• De-flake the stochastic tests by @mdekstrand in #925
• Add Bash-based CLI tests by @mdekstrand in #926
• Add concat and merge operations to ItemList by @mdekstrand in #927
• Reduce test burden for PyTorch-optimized components by @mdekstrand in #928
• Add initial support for session-based recommendation by @mdekstrand in #929
• Add cat_matrix to AttributeSet by @sushobhan2024 in #912
• FAIR: select protected items when they are ranked before unprotected items by @samiravaez in #939
• Add additional support and tests for generating session recommendations by @mdekstrand in #945
• Add co_occurrence method to relationships by @mdekstrand in #950

Full Changelog: v2025.5.0...v2025.6.0

Fifth improvement release

Some small renames and updates, along with entropy functions. Finalized a couple weeks ago but not released.

What's Changed

• Rename rank metric k arguments to n by @mdekstrand in #902
• Rename MetricAccumulator to MeasurementCollector by @mdekstrand in #906
• Add entropy and rank biased entropy by @sushobhan2024 in #890
• Add Rust test coverage support and clean up test CI by @mdekstrand in #909

Full Changelog: v2025.4.0...v2025.5.0

Fourth 2025 Improvement Release

These improvements are relatively minor, but are new features, so I'm bumping to 2025.4.

What's Changed

• Add additional Arrow CSR manipulation, including matrix transpose by @mdekstrand in #893
• Use pipelines to test scorers by @mdekstrand in #895
• Add PipelineProfiler to support collecting performance profiles on pipeline runs by @mdekstrand in #897
• Add Torch random_generator option and use in FlexMF by @mdekstrand in #898
• Update user guide to UsesTrainer + provide default finalize implementation by @mdekstrand in #899
• Improvements to stochastic tests by @mdekstrand in #900

Full Changelog: v2025.3.2...v2025.4.0

2025.3 Bugfix

This has a few small bugfixes, along with some experimental feature updates, for the 2025.3 series.

What's Changed

• Fix Jupyter progress bar rate limit by @mdekstrand in #876
• Simplify and refactor component tests by @mdekstrand in #879
• Update conda build + increase flaky retries by @mdekstrand in #882
• Update Dataset and ItemList repr functions by @mdekstrand in #877
• Fall back to normal model when Dynamo compilation fails by @mdekstrand in #883
• Exclude missing values from sparse matrix attribute representations by @mdekstrand in #884
• FAIR reranker: minor fixes and refinement by @samiravaez in #885
• Fix exponential embedding sizes and tuning settings by @mdekstrand in #886
• Add reranking behavior metrics: rbo and lip by @sushobhan2024 in #856

Full Changelog: v2025.3.1...v2025.3.2

Corrected 2025 Second Update

Corrected version of 2025.3.0 to fix publishing and versioning errors.

2025 Second Update

Second major feature update + bugfix release for the 2025 series! Many changes, see the milestone and release notes for more.

Merged PRs

• Simplify test workfow by @mdekstrand in #666
• Add StochasticTopNSampler to be correct version of "softmax" sampler by @mdekstrand in #667
• Add flexible Torch matrix factorizer by @mdekstrand in #668
• Export configuration classes from lenskit.basic by @mdekstrand in #672
• Improve logging in Ray cluster setups by @mdekstrand in #673
• Limit test users to only have users who had item in the train data. by @sushobhan2024 in #676
• Add train and recommend commands to the CLI by @mdekstrand in #677
• Improve NumPy -> Torch conversions for read-only data by @mdekstrand in #680
• Reduce excess matrix manipulations in item-KNN by @mdekstrand in #678
• Improve progress bar update API by @mdekstrand in #681
• Update FlexMF parameters to better defaults by @mdekstrand in #682
• Run Doctor under coverage by @mdekstrand in #683
• Use uv for development environments instead of pixi by @mdekstrand in #700
• Add lenskit.state module and move the ParameterContainer interface by @mdekstrand in #702
• Create and use separate ModelTrainer objects by @mdekstrand in #701
• Support training checkpointing in ALS by @mdekstrand in #704
• Add CPU & GPU groups by @mdekstrand in #706
• Improve logging and reduce sorting in dataset relationships by @mdekstrand in #707
• Improve debug logging and add xopen to CLI by @mdekstrand in #708
• Introduce WARP loss to FlexMF by @mdekstrand in #709
• Build with setuptools instead of hatch by @mdekstrand in #710
• Add better compression support to CLI by @mdekstrand in #712
• Update pre-commit hook versions by @mdekstrand in #713
• Add Rust extension infrastructure and accelerate ItemKNN by @mdekstrand in #715
• Replace just with invoke for development tasks by @mdekstrand in #716
• Add versions to dataset schemas by @mdekstrand in #721
• Add SparseRow extension type for CSRs in Arrow by @mdekstrand in #722
• Use Rust to accelerate negative sampling by @mdekstrand in #723
• Re-add progress bars and signal checking to Item KNN builder by @mdekstrand in #724
• Add ndarray dep and use it for sampling by @mdekstrand in #725
• Use Maturin to simplify build setup by @mdekstrand in #726
• Reduce base structure memory consumption by @mdekstrand in #728
• Add Rust acceleration to User KNN by @mdekstrand in #730
• Rewrite ALS solvers in Rust by @mdekstrand in #731
• Rewrite FunkSVD in Rust by @mdekstrand in #732
• Build with ABI3 and maturin-action by @mdekstrand in #733
• Add SPEC-1 (Lazy Loading) and incorporate version numbers into docs by @mdekstrand in #736
• Implement is_sorted for matrix tables in Rust by @mdekstrand in #737
• Correctly detect no-genre movies in MovieLens by @mdekstrand in #738
• Adding Mean Average Precision by @albus-droid in #739
• Add tests to avoid overflow in softmax stochastic sampling by @mdekstrand in #742
• Add configurable weighting models by @mdekstrand in #744
• Parallelize item-item scoring by @mdekstrand in #745
• Support comma-separated lists for uniform nested parallelism configuration by @mdekstrand in #746
• Test and fix crash on Linux ARM by @mdekstrand in #748
• Update ALS solve to zero rows for no-info users/items by @mdekstrand in #749
• Add importers for UCSD Amazon data by @mdekstrand in #751
• refactor FlexMF with less string compares by @mdekstrand in #752
• Add sklearn NMF as a scorer by @FroggoLight in #741
• Speed up ItemList isin checks and evaluation metrics by @mdekstrand in #753
• Speed up ItemList indexing by @mdekstrand in #754
• Add Gini coefficient metrics by @mdekstrand in #755
• Speed up weight computations by @mdekstrand in #756
• Speed up weighted metrics by @mdekstrand in #757
• Add ItemList.top_n and speed up ItemList operations by @mdekstrand in #758
• Add subprocess progress reporting by @mdekstrand in #759
• Add a rate limiter and use ray.wait to manage Ray concurrency by @mdekstrand in #763
• Switch back to Just + docopt by @mdekstrand in #764
• Add lenskit data subset command, improve ItemListCollection conversion, and support random seeds in the CLI by @mdekstrand in #773
• Force Python UTF-8 mode for license hader update in pre-commit by @mdekstrand in #774
• Refactored batch accuracy tests into standard suite by @FroggoLight in #776
• Use pytest-retry to retry the flaky stochastic ranker test by @mdekstrand in #783
• Add component input hooks to the pipeline by @mdekstrand in #782
• Add additional tests to verify training item filtering by @mdekstrand in #788
• Use 64-bit row pointers in matrix relationship sets by @mdekstrand in #789
• Improve dataset documentation by @mdekstrand in #790
• Increaes minimum PyArrow vesion by @mdekstrand in #791
• Add LensKit settings and power querying by @mdekstrand in #794
• Revise documentation to be main site by @mdekstrand in #798
• Add devcontainer and update development dependencies by @mdekstrand in #799
• Automatically search up for lenskit.toml in CLI by @mdekstrand in #802
• Fix syntax in testsetup for item-list docs by @mdekstrand in #805
• added binarize_ratings function in DatasetBuilder by @vibhor-5 in #803
• Fix locate_configuration_root and Conda tests by @mdekstrand in #808
• Drop PyTorch max and update locked dependencies by @mdekstrand in #807
• Work around Ray extension type serialization errors by @mdekstrand in #809
• Use Rust-based hash table instead of Pandas index to implement Vocabulary by @mdekstrand in #811
• Refactor indirect hashtables to support reuse by @mdekstrand in #813
• Use uvx --isolated for compat tests by @mdekstrand in #816
• Use mise tasks instead of just by @mdekstrand in #814
• Implement a CoordinateTable structure for fast duplicate checking by @mdekstrand in #815
• Improve recommend CLI for batch inference, output, profiling by @mdekstrand in #817
• Fix coordinate table breakage when adding interactions after filtering or removing by @mdekstrand in #820
• Simplify PyPI CUDA setup by @mdekstrand in #824
• Implement usable pipeline configuration files and CLI support by @mdekstrand in #831
• Implement faster negative sampling and use CoordinateTable by @mdekstrand in #832
• More FlexMF performance improvements by @mdekstrand in #840
• Add parallel config to lenskit doctor by @mdekstrand in #842
• Add support for repeated interactions by @FroggoLight in #812
• Make list metric DF have a standard Index when there is one key field by @mdekstrand in #845
• Add torch-geometric and LightGCN by @mdekstrand in #822
• Unify list and decomposed metric interfac...

Release fix for 2025.3.0a3

Small fix to the release process for 2025.3.0a3.