Subgroups should follow '.' notation

[bobakfb]

Is there any reason you don't have subgroup options follow the '.' notation? e.g.

from __future__ import annotations
from dataclasses import dataclass
from simple_parsing import ArgumentParser, choice, subgroups
from pathlib import Path


@dataclass
class ModelConfig:
    ...


@dataclass
class DatasetConfig:
    ...


@dataclass
class ModelAConfig(ModelConfig):
    lr: float = 3e-4
    optimizer: str = "Adam"
    betas: tuple[float, float] = 0.9, 0.999


@dataclass
class ModelBConfig(ModelConfig):
    lr: float = 1e-3
    optimizer: str = "SGD"
    momentum: float = 1.234


@dataclass
class Dataset1Config(DatasetConfig):
    data_dir: str | Path = "data/foo"
    foo: bool = False


@dataclass
class Dataset2Config(DatasetConfig):
    data_dir: str | Path = "data/bar"
    bar: float = 1.2


@dataclass
class Config:

    # Which model to use
    model: ModelConfig = subgroups(
        {"model_a": ModelAConfig, "model_b": ModelBConfig},
        default=ModelAConfig(),
    )

    # Which dataset to use
    dataset: DatasetConfig = subgroups(
        {"dataset_1": Dataset1Config, "dataset_2": Dataset2Config}, default=Dataset2Config()
    )


parser = ArgumentParser()
parser.add_arguments(Config, dest="config")
args = parser.parse_args()

config: Config = args.config

print(config)

should be run with

python myscript.py --model="model_b" --model.momentum=1.2

rather than how it works now

python myscript.py --model="model_b" --momentum=1.2

which is confusing because you need to ensure that there are no args that have the same name across all subgroups and the main config...

For instance suppose this config handled the creation of two models (again of type model_a or model_b). How would you be able to distinguish which model you were setting the momentum for?