Consistently use `suggest_float` by crcrpar · Pull Request #2344 · optuna/optuna

crcrpar · 2021-02-14T07:10:09Z

In the pursuit of the comprehensive use of suggest_float instead of suggest_( | log | discrete)uniform.

codecov-io · 2021-02-14T07:23:37Z

Codecov Report

Merging #2344 (dd903ea) into master (d25c7ee) will decrease coverage by 0.05%.
The diff coverage is n/a.

@@            Coverage Diff             @@
##           master    #2344      +/-   ##
==========================================
- Coverage   91.14%   91.08%   -0.06%     
==========================================
  Files         126      126              
  Lines       10556    10556              
==========================================
- Hits         9621     9615       -6     
- Misses        935      941       +6

Impacted Files	Coverage Δ
optuna/distributions.py	`97.01% <ø> (ø)`
optuna/exceptions.py	`100.00% <ø> (ø)`
optuna/integration/cma.py	`99.57% <ø> (ø)`
optuna/integration/mlflow.py	`100.00% <ø> (ø)`
optuna/integration/skopt.py	`96.66% <ø> (ø)`
optuna/logging.py	`98.33% <ø> (ø)`
optuna/multi_objective/samplers/_random.py	`100.00% <ø> (ø)`
optuna/pruners/_hyperband.py	`98.80% <ø> (ø)`
optuna/pruners/_median.py	`100.00% <ø> (ø)`
optuna/pruners/_nop.py	`100.00% <ø> (ø)`
... and 18 more

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update d25c7ee...dd903ea. Read the comment docs.

…/suggest-float

HideakiImamura · 2021-02-18T02:50:34Z

@himkt Could you review this PR if you have time? Please feel free to remove the assignment if you are busy.

HideakiImamura

Thanks for the code-fix. I have a minor comment. PTAL.

tests/integration_tests/test_skopt.py

Co-authored-by: Hideaki Imamura <38826298+HideakiImamura@users.noreply.github.com>

HideakiImamura

LGTM.

himkt · 2021-02-18T14:42:36Z

Sorry but I want to clarify what consistently use means. 🙏
Is the scope of this PR is to replace suggest_{log,discrete_}uniform (a) with suggest_float (b) where (a) and (b) both appear in the same script, right? (I'm asking because I found there were still suggest_{log,discrete_}uniform in Optuna)

Details

> git grep "suggest_discrete_uniform"
optuna/distributions.py:    This object is instantiated by :func:`~optuna.trial.Trial.suggest_discrete_uniform`, and passed
optuna/integration/chainermn.py:    def suggest_discrete_uniform(self, name: str, low: float, high: float, q: float) -> float:
optuna/integration/chainermn.py:            return self.delegate.suggest_discrete_uniform(name, low, high, q)
optuna/multi_objective/trial.py:    def suggest_discrete_uniform(self, name: str, low: float, high: float, q: float) -> float:
optuna/multi_objective/trial.py:        Please refer to the documentation of :func:`optuna.trial.Trial.suggest_discrete_uniform`
optuna/multi_objective/trial.py:        return self._trial.suggest_discrete_uniform(name, low, high, q)
optuna/trial/_base.py:    def suggest_discrete_uniform(self, name: str, low: float, high: float, q: float) -> float:
optuna/trial/_fixed.py:    def suggest_discrete_uniform(self, name: str, low: float, high: float, q: float) -> float:
optuna/trial/_frozen.py:    def suggest_discrete_uniform(self, name: str, low: float, high: float, q: float) -> float:
optuna/trial/_trial.py:        :func:`~optuna.trial.Trial.suggest_discrete_uniform`.
optuna/trial/_trial.py:            :func:`~optuna.trial.Trial.suggest_discrete_uniform`.
optuna/trial/_trial.py:                    this method falls back to :func:`~optuna.trial.Trial.suggest_discrete_uniform`.
optuna/trial/_trial.py:                return self.suggest_discrete_uniform(name, low, high, step)
optuna/trial/_trial.py:    def suggest_discrete_uniform(self, name: str, low: float, high: float, q: float) -> float:
optuna/trial/_trial.py:                    subsample = trial.suggest_discrete_uniform("subsample", 0.1, 1.0, 0.1)
tests/integration_tests/test_chainermn.py:    def test_suggest_discrete_uniform(storage_mode: str, comm: CommunicatorBase) -> None:
tests/integration_tests/test_chainermn.py:                x1 = mn_trial.suggest_discrete_uniform("x", low, high, q)
tests/integration_tests/test_chainermn.py:                x2 = mn_trial.suggest_discrete_uniform("x", low, high, q)
tests/integration_tests/test_sampler.py:        p3 = trial.suggest_discrete_uniform("p3", 0, 9, 3)
tests/integration_tests/test_sampler.py:        p3 = trial.suggest_discrete_uniform("p3", 0, 9, 3)
tests/integration_tests/test_sampler.py:        p3 = trial.suggest_discrete_uniform("p3", 0, 9, 3)
tests/integration_tests/test_sampler.py:    p6 = trial.suggest_discrete_uniform("p6", 10, 20, 2)
tests/integration_tests/test_sampler.py:    p7 = trial.suggest_discrete_uniform("p7", 0.1, 1.0, 0.1)
tests/integration_tests/test_sampler.py:    p8 = trial.suggest_discrete_uniform("p8", 2.2, 2.2, 0.5)
tests/samplers_tests/test_grid.py:        d = trial.suggest_discrete_uniform("d", -5, 5, 1)
tests/samplers_tests/tpe_tests/test_sampler.py:        t.suggest_discrete_uniform("c", 1.0, 100.0, 3.0)
tests/trial_tests/test_fixed.py:def test_suggest_discrete_uniform() -> None:
tests/trial_tests/test_fixed.py:    assert trial.suggest_discrete_uniform("x", 0.0, 1.0, 0.1) == 0.9
tests/trial_tests/test_fixed.py:        trial.suggest_discrete_uniform("y", 0.0, 1.0, 0.1)
tests/trial_tests/test_frozen.py:        c = trial.suggest_discrete_uniform("c", 0.0, 10.0, 1.0)
tests/trial_tests/test_frozen.py:def test_suggest_discrete_uniform() -> None:
tests/trial_tests/test_frozen.py:    assert trial.suggest_discrete_uniform("x", 0.0, 1.0, 0.1) == 0.9
tests/trial_tests/test_frozen.py:        trial.suggest_discrete_uniform("y", 0.0, 1.0, 0.1)
tests/trial_tests/test_trial.py:    x6 = trial.suggest_discrete_uniform("x3", 10, 20, 1.0)
tests/trial_tests/test_trial.py:def test_check_distribution_suggest_discrete_uniform(
tests/trial_tests/test_trial.py:        trial.suggest_discrete_uniform("x", 10, 20, 2)
tests/trial_tests/test_trial.py:        trial.suggest_discrete_uniform("x", 10, 20, 2)
tests/trial_tests/test_trial.py:        trial.suggest_discrete_uniform("x", 10, 22, 2)
tests/trial_tests/test_trial.py:def test_suggest_discrete_uniform(storage_init_func: Callable[[], storages.BaseStorage]) -> None:
tests/trial_tests/test_trial.py:        assert trial.suggest_discrete_uniform("c", 1.0, 1.0, 1.0) == 1.0  # Suggesting a param.
tests/trial_tests/test_trial.py:            trial.suggest_discrete_uniform("c", 1.0, 1.0, 1.0) == 1.0
tests/trial_tests/test_trial.py:def test_suggest_discrete_uniform_range(
tests/trial_tests/test_trial.py:            x = trial.suggest_discrete_uniform(
tests/trial_tests/test_trial.py:            x = trial.suggest_discrete_uniform(
2021-02-18 23:37:41 [~/work/github.com/himkt/optuna] @ pr
> git grep "trial.suggest_loguniform"
optuna/multi_objective/trial.py:        return self._trial.suggest_loguniform(name, low, high)
optuna/trial/_trial.py:                    c = trial.suggest_loguniform("c", 1e-5, 1e2)
tests/importance_tests/test_mean_decrease_impurity.py:    x2 = trial.suggest_loguniform("x2", 0.1, 3)
tests/importance_tests/test_mean_decrease_impurity.py:    x3 = trial.suggest_loguniform("x3", 2, 4)
tests/integration_tests/test_chainermn.py:        y = trial.suggest_loguniform("y", 20, 30)
tests/integration_tests/test_chainermn.py:                    mn_trial.suggest_loguniform("x1", low1, high1)
tests/integration_tests/test_chainermn.py:                x4 = mn_trial.suggest_loguniform("x2", low2, high2)
tests/integration_tests/test_chainermn.py:                    mn_trial.suggest_loguniform("x", low, high)
tests/integration_tests/test_chainermn.py:                x1 = mn_trial.suggest_loguniform("x", low, high)
tests/integration_tests/test_chainermn.py:                x2 = mn_trial.suggest_loguniform("x", low, high)
tests/integration_tests/test_mlflow.py:    y = trial.suggest_loguniform("y", 20, 30)
tests/integration_tests/test_mlflow.py:    y = trial.suggest_loguniform("y", 20, 30)
tests/integration_tests/test_sampler.py:        p1 = trial.suggest_loguniform("p1", 1, 10)
tests/integration_tests/test_sampler.py:        p1 = trial.suggest_loguniform("p1", 1, 10)
tests/integration_tests/test_sampler.py:        p1 = trial.suggest_loguniform("p1", 50, 100)  # The range has been changed
tests/integration_tests/test_sampler.py:    p2 = trial.suggest_loguniform("p2", 0.0001, 0.3)
tests/integration_tests/test_sampler.py:    p3 = trial.suggest_loguniform("p3", 1.1, 1.1)
tests/integration_tests/test_tensorboard.py:    y = trial.suggest_loguniform("y", 20.0, 30.0)
tests/samplers_tests/test_grid.py:        e = trial.suggest_loguniform("e", 0.0001, 1)
tests/samplers_tests/test_samplers.py:        assert trial.suggest_loguniform("d", 1, 100) == unknown_param_value
tests/study_tests/test_study.py:    y = trial.suggest_loguniform("y", 20, 30)
tests/test_dashboard.py:        y = trial.suggest_loguniform("y", 10, 20)
tests/trial_tests/test_fixed.py:    assert trial.suggest_loguniform("x", 0.1, 1.0) == 0.99
tests/trial_tests/test_fixed.py:        trial.suggest_loguniform("y", 0.0, 1.0)
tests/trial_tests/test_frozen.py:        b = trial.suggest_loguniform("b", 0.1, 10.0)
tests/trial_tests/test_frozen.py:    assert trial.suggest_loguniform("x", 0.1, 1.0) == 0.99
tests/trial_tests/test_frozen.py:        trial.suggest_loguniform("y", 0.0, 1.0)
tests/trial_tests/test_trial.py:    x4 = trial.suggest_loguniform("x2", 1e-5, 1e-3)
tests/trial_tests/test_trial.py:        trial.suggest_loguniform("x", 10, 20)
tests/trial_tests/test_trial.py:        trial.suggest_loguniform("x", 10, 20)
tests/trial_tests/test_trial.py:        trial.suggest_loguniform("x", 10, 30)
tests/trial_tests/test_trial.py:        assert trial.suggest_loguniform("b", 1.0, 1.0) == 1.0  # Suggesting a param.
tests/trial_tests/test_trial.py:        assert trial.suggest_loguniform("b", 1.0, 1.0) == 1.0  # Suggesting the same param.
tutorial/10_key_features/003_efficient_optimization_algorithms.py:    alpha = trial.suggest_loguniform("alpha", 1e-5, 1e-1)
tutorial/20_recipes/003_attributes.py:    svc_c = trial.suggest_loguniform("svc_c", 1e-10, 1e10)

crcrpar · 2021-02-19T00:32:15Z

Basically I want to replace *uniform with *float as possible entirely except for the implementations of *uniform methods.

Some calls of `trials.suggest_(float, loguniform, discreteuniform) are left there deliberately if the caller functions names include uniform. Otherwise, I overlooked them. Let me check them again.

…to code-fix/suggest-float

crcrpar · 2021-02-20T14:52:59Z

I'd like this pull request to get "squash & merge" as this pull request has a bunch of meaningless commits.

himkt · 2021-02-20T16:19:00Z

I'm sorry if I misunderstood something but I'm wondering we can also update tutorials.

https://github.com/crcrpar/optuna/blob/code-fix/suggest-float/tutorial/20_recipes/005_user_defined_sampler.py#L20
https://github.com/crcrpar/optuna/blob/code-fix/suggest-float/tutorial/20_recipes/005_user_defined_sampler.py#L109-L110

It also needs to update an error massage, as well.
(e.g. Only suggest_floatwithlog=False is supported.)

https://github.com/crcrpar/optuna/blob/code-fix/suggest-float/tutorial/20_recipes/005_user_defined_sampler.py#L78

himkt · 2021-02-20T16:19:48Z

I'd like this pull request to get "squash & merge" as this pull request has a bunch of meaningless commits.

Gotcha, I'll merge this PR with squashing!

himkt

Thank you for the patience. LGTM.

This includes using `suggest_float` instead of `suggest_{log | discrete}uniform`as described in optuna#2344

github-actions bot added the optuna.visualization Related to the `optuna.visualization` submodule. This is automatically labeled by github-actions. label Feb 14, 2021

HideakiImamura added the code-fix Change that does not change the behavior, such as code refactoring. label Feb 15, 2021

toshihikoyanase mentioned this pull request Feb 17, 2021

Fix improper stopping with the combination of GridSampler and HyperbandPruner #2353

Merged

Consistently use suggest_float

7909955

crcrpar force-pushed the code-fix/suggest-float branch from 355e987 to 7909955 Compare February 17, 2021 12:04

Merge branch 'master' of ssh://github.com/optuna/optuna into code-fix…

8d4d695

…/suggest-float

crcrpar marked this pull request as ready for review February 17, 2021 23:18

HideakiImamura assigned himkt and HideakiImamura Feb 18, 2021

HideakiImamura reviewed Feb 18, 2021

View reviewed changes

tests/integration_tests/test_skopt.py Outdated Show resolved Hide resolved

Update tests/integration_tests/test_skopt.py

c0f5643

Co-authored-by: Hideaki Imamura <38826298+HideakiImamura@users.noreply.github.com>

HideakiImamura approved these changes Feb 18, 2021

View reviewed changes

crcrpar added 2 commits February 20, 2021 19:41

replace more

1d09980

Merge branch 'master' into code-fix/suggest-float

49935d9

crcrpar requested review from HideakiImamura and himkt February 20, 2021 10:45

crcrpar added 5 commits February 20, 2021 19:50

black

8d9a247

Merge branch 'code-fix/suggest-float' of github.com:crcrpar/optuna in…

93a1b13

…to code-fix/suggest-float

flake8

419b248

fix: size -> step

dfd16ea

fix

dd903ea

replace suggest_uniform with suggest_float

e6f0978

himkt approved these changes Feb 21, 2021

View reviewed changes

himkt merged commit 1eaccb5 into optuna:master Feb 21, 2021

toshihikoyanase added this to the v2.6.0 milestone Mar 4, 2021

toshihikoyanase mentioned this pull request Aug 24, 2021

Support multi-objective optimization in MLflowCallback #2863

Merged

3 tasks

xadrianzetx added a commit to xadrianzetx/optuna that referenced this pull request Aug 24, 2021

Make test objective function unified with rest of the codebase

82b573a

This includes using `suggest_float` instead of `suggest_{log | discrete}uniform`as described in optuna#2344

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Consistently use `suggest_float`#2344

Consistently use `suggest_float`#2344
himkt merged 11 commits intooptuna:masterfrom
crcrpar:code-fix/suggest-float

crcrpar commented Feb 14, 2021 •

edited

Loading

Uh oh!

codecov-io commented Feb 14, 2021 •

edited

Loading

Uh oh!

HideakiImamura commented Feb 18, 2021

Uh oh!

HideakiImamura left a comment

Uh oh!

Uh oh!

HideakiImamura left a comment

Uh oh!

himkt commented Feb 18, 2021

Uh oh!

crcrpar commented Feb 19, 2021

Uh oh!

crcrpar commented Feb 20, 2021

Uh oh!

himkt commented Feb 20, 2021

Uh oh!

himkt commented Feb 20, 2021

Uh oh!

himkt left a comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

Uh oh!

Conversation

crcrpar commented Feb 14, 2021 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

codecov-io commented Feb 14, 2021 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

HideakiImamura commented Feb 18, 2021

Uh oh!

HideakiImamura left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

HideakiImamura left a comment

Choose a reason for hiding this comment

Uh oh!

himkt commented Feb 18, 2021

Uh oh!

crcrpar commented Feb 19, 2021

Uh oh!

crcrpar commented Feb 20, 2021

Uh oh!

himkt commented Feb 20, 2021

Uh oh!

himkt commented Feb 20, 2021

Uh oh!

himkt left a comment

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

crcrpar commented Feb 14, 2021 •

edited

Loading

codecov-io commented Feb 14, 2021 •

edited

Loading