Changelog¶

Development Version¶

Version 0.4.10¶

(February 2026)

FIX device mismatch when Y_in_cpu=True with GPU backend in RidgeCV random search.
ENH add iteration rate and estimated time to completion to progress bar.
ENH change ETA format to hh:mm:ss for better readability.

Version 0.4.9¶

(January 2026)

FIX out-of-bounds error when convergence occurs at last iteration.

Version 0.4.8¶

(August 2025)

MNT maintenance release.

Version 0.4.7¶

(August 2025)

FIX estimators to pass scikit-learn checks with scikit-learn >= 1.6.
MNT run tests on macOS.

Version 0.4.6¶

(June 2024)

FIX ComplexWarning for numpy >= 1.26.
MNT use codecov token.

Version 0.4.5¶

(June 2024)

FIX update ~himalaya.kernel_ridge.ColumnKernelizer for scikit-learn versions >= 1.5

Version 0.4.4¶

(March 2024)

FIX cupy boolean dtype

Version 0.4.3¶

(March 2024)

FIX update ~himalaya.kernel_ridge.ColumnKernelizer for scikit-learn versions > 1.4

Version 0.4.2¶

(February 2023)

ENH add better error message when torch.linalg.eigh fails.
ENH add solve_kernel_ridge_cv_svd() solver. It can be used with:class:~himalaya.kernel_ridge.KernelRidgeCV (solver="svd").

Version 0.4.1¶

(February 2023)

FIX avoid error in MultipleKernelRidgeCV with solver_params(return_alphas=True).
ENH add fit_intercept in MultipleKernelRidgeCV.
FIX torch 1.13.1 requires tensor masks to be on the same device as tensors.

Version 0.4.0¶

(June 2022)

DOC explain how to implement a winner-take-all model.
FIX comply with most recent scikit-learn’s check_estimator.
FIX avoid an indexing error in the hypergradient solver, when early stopping after different numbers of iterations for different batches.

Version 0.3.6¶

(April 2022)

DOC improve documentation website, add estimator flowchart.
TST improve test robustness.
ENH add batching over targets in predict_weighted_kernel_ridge().
ENH add solver="auto" in KernelRidge, which switches solver based on the presence of a separate alpha per target.

Version 0.3.5¶

(February 2022)

MNT speed up examples on CPU, to build the doc faster on github actions.
ENH add batching over targets in Ridge, KernelRidge, and WeightedKernelRidge.
ENH add warnings to guide the user between using Ridge or KernelRidge.
ENH add user-friendly errors when the number of samples is inconsistent.
ENH raise ValueError if the indices in cross-validation exceed number of samples.

Version 0.3.4¶

(November 2021)

FIX Ridge with n_samples < n_targets.
FIX update of alphas when local_alpha=False in MultipleKernelRidgeCV.
EXA refactor examples with new generate_multikernel_dataset() function.
MNT add github actions for running tests, building and publishing the doc, and publishing to PyPI.

Version 0.3.3¶

(November 2021)

FIX KernelRidge with n_samples < n_targets.
FIX random search with single alpha in MultipleKernelRidgeCV.

Version 0.3.2¶

(November 2021)

ENH add r2_score_split_svd() scoring function.
ENH add correlation_score_split() scoring function.
ENH add split parameter to the score method in WeightedKernelRidge, MultipleKernelRidgeCV, and GroupRidgeCV.
ENH add force_cpu parameter in all estimators.
FIX remove deprecation warnings for cupy v9.
DOC mention that pytorch 1.9+ is preferred.

Version 0.3.1¶

(September 2021)

MNT Rename BandedRidgeCV into GroupRidgeCV (both names are available).
ENH improve robustness to noise in the cross-validation scores.
ENH start the random search with equal weights in MultipleKernelRidgeCV and GroupRidgeCV.
FIX remove deprecation warnings with pytorch 1.8.
TST improve test coverage.

Version 0.3.0¶

(April 2021)

ENH add fit_intercept parameter in Ridge, RidgeCV, and BandedRidgeCV.
ENH add fit_intercept parameter in KernelRidge, KernelRidgeCV, solve_multiple_kernel_ridge_gradient_descent(), and solve_multiple_kernel_ridge_random_search().
ENH add KernelCenterer.
ENH allow change of backend midscript.
ENH Add option to return selected alpha values in solve_multiple_kernel_ridge_random_search().

Version 0.2.0¶

(December 2020)

Version 0.1.0¶

(March 2020)