Update cli tutorial #3902
Update cli tutorial #3902
Conversation
| @@ -847,7 +847,7 @@ def get_parser(self, prog_name: str) -> ArgumentParser: | |||
| "--skip-if-finished", | |||
| default=False, | |||
| action="store_true", | |||
| help="If specified, tell is skipped without any error when the trial is already" | |||
| help="If specified, tell is skipped without any error when the trial is already " | |||
There was a problem hiding this comment.
Currently, this part is printed as alreadyfinished.
| # Even so, we can invoke the optimization as follows. | ||
| # (Don't care about ``--storage sqlite:///example.db`` for now, which is described in :ref:`rdb`.) |
There was a problem hiding this comment.
I'd like to suggest removing ( and ).
…d in CLI tutorial
nzw0301
force-pushed
the
update-cli-tutorial
branch
from
8ca67ad
to
4fda221
Compare
Codecov Report
@@ Coverage Diff @@
## master #3902 +/- ##
==========================================
- Coverage 90.58% 90.57% -0.02%
==========================================
Files 164 164
Lines 12601 12601
==========================================
- Hits 11415 11413 -2
- Misses 1186 1188 +2
📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more |
|
@keisuke-umezawa Could you review this PR, please? I think we can merge this PR with a single approval since it updates the tutorial only. |
| # ... | ||
| # [I 2018-05-09 10:40:26,204] Finished a trial resulted in value: 14.704254135013741. Current best value is 2.280758099793617e-06 with parameters: {'x': 1.9984897821018828}. | ||
| # $ optuna ask --storage sqlite:///example.db --study-name $STUDY_NAME --sampler TPESampler \ | ||
| # --search-space '{"x": {"name": "FloatDistribution", "attributes": {"step": null, "low": -10.0, "high": 10.0, "log": false}}}' |
There was a problem hiding this comment.
Do we actually need to specify step and log here? If this commend can run without these arguments, we can remove them. Some people might get stuck by them.
There was a problem hiding this comment.
No, we don't. However, I believe that we use https://optuna.readthedocs.io/en/stable/reference/generated/optuna.distributions.distribution_to_json.html to obtain the json formatted distribution, which includes step and log as seen above.
There was a problem hiding this comment.
What do you think about
- removing
logandstepkey values from the example above for simplicity and readability, and - Adding a sentence to mention https://optuna.readthedocs.io/en/stable/reference/generated/optuna.distributions.distribution_to_json.html to generate json formatted distribution?
There was a problem hiding this comment.
I understand. In that case, to generate the json formatted distribution, we need to explain how we can create it. I agree on your idea.
Adding a sentence to mention https://optuna.readthedocs.io/en/stable/reference/generated/optuna.distributions.distribution_to_json.html to generate json formatted distribution?
You can explain it by the above. But, users need to run some code to generate it. The other way is that you can explain it by some sentences like:
You can choose distributions from `FloatDistribution` and `IntDistribution`, and
specify parameters by the key values of json with `low`, `high`, `log` and `step`.
If you want to use log scale, you can set `"log": true`, otherwise `"log": false`.
If you want to specify step e.g. 2, you can set `"step": 2`, otherwise `"step": null`.
The above sentences is just a quick job, so we need to refine it.
There was a problem hiding this comment.
Thanks. I think the explanation of each argument of the distribution looks unnecessary for this tutorial, so I've added only links to https://optuna.readthedocs.io/en/stable/reference/generated/optuna.distributions.distribution_to_json.html t and distributions. For the consistency with the output of distribution_to_json, I've not removed log and step from the example code.
keisuke-umezawa
added
the
code-fix
Motivation
Resolve #3032.
Description of the changes
As described in #3032, remove deprecated command examples and use ask and tell commands. However, since the
askandtellcommands in CLI module doesn't look suitable for the current example codefoo.py, I've created another scenario to use these commands.