8000 LogisticRegression convert to float64 (for SAG solver) by massich · Pull Request #13243 · scikit-learn/scikit-learn · GitHub

LogisticRegression convert to float64 (for SAG solver)#13243

Merged

GaelVaroquaux merged 31 commits intoscikit-learn:masterfrom

massich:henley_is_8769_minus_merged

Feb 27, 2019

Commits on Jul 8, 2018

Remove unused code

Joan Massich
authored and
NelleV
committed
Squash all the PR 9040 commits

Imbert Arthur
authored and
NelleV
committed
FIX tests
NelleV
committed
TST more numerically stable test_sgd.test_tol_parameter

ogrisel
authored and
NelleV
committed
Added benchmarks to compare SAGA 32b and 64b
NelleV
committed
Fixing gael's comments
NelleV
committed
fix
NelleV
committed

Commits on Jul 19, 2018

Address lesteve comments
glemaitre
committed

Commits on Feb 25, 2019

Merge branch 'master' into henley_is_8769_minus_merged
Joan Massich
committed
fix merging
Joan Massich
committed
avoid using assert_equal
Joan Massich
committed
use all_close
Joan Massich
committed
use explicit ArrayDataset64 and CSRDataset64
Joan Massich
committed
fix: remove unused import
Joan Massich
committed
Use parametrized to cover ArrayDaset-CSRDataset-32-64 matrix
Joan Massich
committed
for consistency use 32 first then 64 + add 64 suffix to variables
Joan Massich
committed
it would be cool if this worked !!!
Joan Massich
committed
more verbose version
Joan Massich
committed

Commits on Feb 26, 2019

revert SGD changes as much as possible.
Joan Massich
committed
Add solvers back to bench_saga
Joan Massich
committed
make 64 explicit in the naming
Joan Massich
committed
remove checking native python type + add comparison between 32 64
Joan Massich
committed
Add whatsnew with everyone with commits
Joan Massich
committed
simplify a bit the testing
Joan Massich
committed
simplify the parametrize
Joan Massich
committed
update whatsnew
Joan Massich
committed

Commits on Feb 27, 2019

fix pep8
Joan Massich
committed

0