JohannesBuchner / askcarl / 15486794598

Builds Branch Commit Type Ran Committer Via Coverage
15486794598 main optimize fast path of no upper limits; add refine_weights to LightGMM fast path push JohannesBuchner github
93.7
15484404374 main Bump version: 1.2.1 → 2.0.0 push JohannesBuchner github
95.13
15484404372 refs/tags/v2.0.0 Bump version: 1.2.1 → 2.0.0 push JohannesBuchner github
95.13
15484259899 main add LightGMM2: a K-folded version of LightGMM push JohannesBuchner github
95.13
15464111730 main add custom re-implementation of scikit-learn's kmeans with max_iter=1 push JohannesBuchner github
94.32
15458715466 main avoid call to matrix_rank, because that also does svg; correct transpose of precisions_cholesky push JohannesBuchner github
96.89
15454787204 main avoid call to matrix_rank, because that also does svg; correct transpose of precisions_cholesky push JohannesBuchner github
96.89
15450188067 main add test data push JohannesBuchner github
96.02
15449997508 main bug fix, and simplify code push JohannesBuchner github
95.75
15449845080 main fix mvn_logpdf -- it was actually fine but cov_to_prec_cholesky gave a transposed result push JohannesBuchner github
95.85
15436951998 main remove fast-track, it does less & serialized. Handle simple case without copying. push JohannesBuchner github
95.94
15434542796 main avoid scipy's multivariate_normal if possible push JohannesBuchner github
95.95
15434201419 main remove seemingly unused function push JohannesBuchner github
96.06
15433916708 main forgot to add tests for the utils push JohannesBuchner github
95.61
15427442868 main add test for refine_weights and fast no-KMeans mode push JohannesBuchner github
94.78