Slot Machine

Have you ever found yourself unclear over which configuration to choose?

Don't choose.

Randomize.

With slot_machine you can now sample any configuration.

Can't settle on a background color? Randomize...

Looking for a reasonable model parameter? Randomize...

Don't know what to eat on Sunday? Randomize...

With slot_machine your next choice will be one you look forward to.

Examples

Simply add your favorite samplers to the configuration:

chicken_nuggets: !SampleRange 5..20
price: !SampleUniform 9.99..20

Specify your dataclass and get rolling:

from dataclasses import dataclass
from slot_machine import SlotsSerializer

@dataclass(slots=True)
class Basket(SlotsSerializer):
    chicken_nuggets: int
    price: float

yaml_file = """
basket:
  chicken_nuggets: !SampleRange 5..10
  price: !SampleUniform 9.99..20
"""

surprise_basket = Basket.from_yaml(yaml_file)

Build your own

from dataclasses import dataclass
from slot_machine import SlotsSerializer, MappingSampler

@dataclass(slots=True)
class Basket(SlotsSerializer):
    chicken_nuggets: int
    splits: list[int]

class SampleSplits(MappingSampler):

  @classmethod
  def get_sample(cls, n_splits: str, values: str):
    n_splits = int(n_splits)
    values = list(map(int, values.split("..")))
    return sorted([random.randint(*values) for _ in range(n_splits)])
    
yaml_file = """
basket:
  chicken_nuggets: 100
  splits: !SampleSplits
    n_splits: 3
    values: 0..100
"""

surprise_basket = Basket.from_yaml(yaml_file)