mirror of
https://github.com/vale981/ray
synced 2025-03-09 12:56:46 -04:00
f9b71a8bf6
This PR consolidates both #21667 and #21759 (look there for features), but improves on them in the following way: - [x] we reverted renaming of existing projects `tune`, `rllib`, `train`, `cluster`, `serve`, `raysgd` and `data` so that links won't break. I think my consolidation efforts with the `ray-` prefix were a little overeager in that regard. It's better like this. Only the creation of `ray-core` was a necessity, and some files moved into the `rllib` folder, so that should be relatively benign. - [x] Additionally, we added Algolia `docsearch`, screenshot below. This is _much_ better than our current search. Caveat: there's a sphinx dependency that needs to be replaced (`sphinx-tabs`) by another, newer one (`sphinx-panels`), as the former prevents loading of the `algolia.js` library. Will follow-up in the next PR (hoping this one doesn't get re-re-re-re-reverted).
57 lines
2.3 KiB
ReStructuredText
57 lines
2.3 KiB
ReStructuredText
.. _trainer-reference-docs:
|
|
|
|
Trainer API
|
|
===========
|
|
|
|
The :py:class:`~ray.rllib.agents.trainer.Trainer` class is the highest-level API in RLlib.
|
|
It allows you to train and evaluate policies, save an experiment's progress and restore from
|
|
a prior saved experiment when continuing an RL run.
|
|
:py:class:`~ray.rllib.agents.trainer.Trainer` is a sub-class
|
|
of :py:class:`~ray.tune.Trainable`
|
|
and thus fully supports distributed hyperparameter tuning for RL.
|
|
|
|
.. https://docs.google.com/drawings/d/1J0nfBMZ8cBff34e-nSPJZMM1jKOuUL11zFJm6CmWtJU/edit
|
|
.. figure:: ../images/trainer_class_overview.svg
|
|
:align: left
|
|
|
|
**A typical RLlib Trainer object:** The components sitting inside a Trainer are
|
|
normally N :py:class:`~ray.rllib.evaluation.worker_set.WorkerSet`s
|
|
(each consisting of one local :py:class:`~ray.rllib.evaluation.RolloutWorker`
|
|
and zero or more \@ray.remote
|
|
:py:class:`~ray.rllib.evaluation.RolloutWorker`s),
|
|
a set of :py:class:`~ray.rllib.policy.Policy`(ies)
|
|
and their NN models per worker, and a (already vectorized)
|
|
RLlib :py:class:`~ray.rllib.env.base_env.BaseEnv` per worker.
|
|
|
|
|
|
Building Custom Trainer Classes
|
|
-------------------------------
|
|
|
|
.. warning::
|
|
As of Ray >= 1.9, it is no longer recommended to use the `build_trainer()` utility
|
|
function for creating custom Trainer sub-classes.
|
|
Instead, follow the simple guidelines here for directly sub-classing from
|
|
:py:class:`~ray.rllib.agents.trainer.Trainer`.
|
|
|
|
In order to create a custom Trainer, sub-class the
|
|
:py:class:`~ray.rllib.agents.trainer.Trainer` class
|
|
and override one or more of its methods. Those are in particular:
|
|
|
|
* :py:meth:`~ray.rllib.agents.trainer.Trainer.get_default_config`
|
|
* :py:meth:`~ray.rllib.agents.trainer.Trainer.validate_config`
|
|
* :py:meth:`~ray.rllib.agents.trainer.Trainer.get_default_policy_class`
|
|
* :py:meth:`~ray.rllib.agents.trainer.Trainer.setup`
|
|
* :py:meth:`~ray.rllib.agents.trainer.Trainer.step_attempt`
|
|
* :py:meth:`~ray.rllib.agents.trainer.Trainer.execution_plan`
|
|
|
|
`See here for an example on how to override Trainer <https://github.com/ray-project/ray/blob/master/rllib/agents/pg/pg.py>`_.
|
|
|
|
|
|
Trainer base class (ray.rllib.agents.trainer.Trainer)
|
|
-----------------------------------------------------
|
|
|
|
.. autoclass:: ray.rllib.agents.trainer.Trainer
|
|
:special-members: __init__
|
|
:members:
|
|
|
|
.. static-members: get_default_config, execution_plan
|