Source code for torchrl.trainers.algorithms.configs.common
# Copyright (c) Meta Platforms, Inc. and affiliates.
#
# This source code is licensed under the MIT license found in the
# LICENSE file in the root directory of this source tree.
from __future__ import annotations
from abc import ABC, abstractmethod
from collections.abc import Sequence
from dataclasses import dataclass
from typing import Any
from omegaconf import DictConfig, ListConfig
from tensordict import NestedKey
[docs]
@dataclass
class ConfigBase(ABC):
"""Abstract base class for all configuration classes.
This class serves as the foundation for all configuration classes in the
configurable configuration system, providing a common interface and structure.
"""
@abstractmethod
def __post_init__(self) -> None:
"""Post-initialization hook for configuration classes."""
@dataclass
class Config:
"""A flexible config that allows arbitrary fields."""
def __init__(self, **kwargs):
self._config = DictConfig(kwargs)
def __getattr__(self, name):
return getattr(self._config, name)
def __setattr__(self, name, value):
if name == "_config":
super().__setattr__(name, value)
else:
setattr(self._config, name, value)
def _normalize_hydra_key(key: Any) -> NestedKey:
"""Converts a Hydra list of strings to a :class:`~tensordict.NestedKey`.
A multi-element list/``ListConfig`` is converted to a tuple, a single-element
list/``ListConfig`` is unwrapped to the contained element, and any other
value is returned unchanged. This means a YAML entry such as
``action_key: [agents, action]`` parses to ``("agents", "action")``,
``action_key: [action]`` parses to ``"action"`` and ``action_key: action``
parses to ``"action"``.
"""
if isinstance(key, (list, ListConfig)):
key = tuple(key)
if len(key) == 1:
return key[0]
return key
def _normalize_hydra_keys(keys: Sequence[Any] | None) -> Sequence[NestedKey] | None:
"""Maps :func:`_normalize_hydra_key` element-wise over a sequence of keys.
Each element of ``keys`` is treated as an independent
:class:`~tensordict.NestedKey`. To express a single nested key, wrap it in
its own list, e.g. ``out_keys: [[agents, observation]]`` parses to
``[("agents", "observation")]``, while ``out_keys: [next, reward]`` parses
to two flat keys ``["next", "reward"]``.
"""
return keys if keys is None else [_normalize_hydra_key(key) for key in keys]
