[docs] Tune overhaul part II

[maxpumperla]

Preview: docs

Largely completes the Tune docs overhaul. Notable improvements:

• Key concepts page improved.
• Content deeply hidden in API docs now surfaced (either in guides or key concepts).
• Feature guides now collectively under "user guides"
• "Examples" are now all executable (.ipynb). Apart from the 8 good tutorials converted, we added 6 new ones.
• There are barely any untested code snippets anymore. They are either included and tested, or are notebooks and tested.
• Testing simplified. A bazel rule picks up all notebook examples and doc test snippets. New examples can now just be added by users without touching our CI (which is massive for adoption), at least for Tune.

Checks

• I've run scripts/format.sh to lint the changes in this PR.
• I've included any doc changes needed for https://docs.ray.io/en/master/.
• I've made sure the tests are passing.
• Testing Strategy
  • Unit tests

[Yard1]

Awesome work!

[richardliaw]

One nit is that the intro "pytorch optuna" code doesn't seem to run. Missing imports i think?

[maxpumperla]

@richardliaw the example is too long to show in its entirety, but the snippet is tested. we could have a link to the full example though?

[richardliaw]

Sounds good, lets link it inline

…

On Fri, Feb 25, 2022 at 10:21 PM Max Pumperla ***@***.***> wrote: @richardliaw <https://github.com/richardliaw> the example is too long to show in its entirety, but the snippet is tested. we could have a link to the full example though? — Reply to this email directly, view it on GitHub <#22656 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABCRZZMYARPA62AYCA4MXULU5BWM7ANCNFSM5PKG34KA> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[yingbaihu]

I have one question, can anybody help me?
In the example mlflow_ptl_example.rst
it imports ray.air, and use air.RunConfig command as a parameter in tune.Tuner,
is there any way that not use ray.air package? any replace function of air.RunConfig?

[richardliaw]

Whats the reason for that?

…

On Thu, Mar 2, 2023 at 12:06 PM yingbaihu ***@***.***> wrote: I have one question, can anybody help me? In the example mlflow_ptl_example.rst <https://github.com/ray-project/ray/blob/master/doc/source/tune/examples/includes/mlflow_ptl_example.rst> it imports ray.air, and use air.RunConfig command as a parameter in tune.Tuner, is there any way that not use ray.air package? any replace function of air.RunConfig? — Reply to this email directly, view it on GitHub <#22656 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABCRZZLE7O7HYJ2LXB3H4MDW2D4UTANCNFSM5PKG34KA> . You are receiving this because you modified the open/close state.Message ID: ***@***.***>

[yingbaihu]

Whats the reason for that?
…
On Thu, Mar 2, 2023 at 12:06 PM yingbaihu @.> wrote: I have one question, can anybody help me? In the example mlflow_ptl_example.rst https://github.com/ray-project/ray/blob/master/doc/source/tune/examples/includes/mlflow_ptl_example.rst it imports ray.air, and use air.RunConfig command as a parameter in tune.Tuner, is there any way that not use ray.air package? any replace function of air.RunConfig? — Reply to this email directly, view it on GitHub <#22656 (comment)>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABCRZZLE7O7HYJ2LXB3H4MDW2D4UTANCNFSM5PKG34KA . You are receiving this because you modified the open/close state.Message ID: @.>

Because my plan is only to use ray for optimization, so I just installed ray.tune package