Without maximum, floats are extremely constrained:

>>> create_random_float(Random(), minimum=100, maximum=None)
100.18084162477052
>>> create_random_float(Random(), minimum=100, maximum=None)
100.54924103325709
>>> create_random_float(Random(), minimum=100, maximum=None)
101.5380217963041
>>> create_random_float(Random(), minimum=100, maximum=None)
100.48749377309393
>>> create_random_float(Random(), minimum=100, maximum=None)
100.62887519044165

With maximum, they are differentiated much more:

>>> create_random_float(Random(), minimum=100, maximum=1000)
922.9521500077755
>>> create_random_float(Random(), minimum=100, maximum=1000)
594.5901441349624
>>> create_random_float(Random(), minimum=100, maximum=1000)
668.6168721529523
>>> create_random_float(Random(), minimum=100, maximum=1000)
760.5950727117361
>>> create_random_float(Random(), minimum=100, maximum=1000)
269.2787865511392

I discovered this while debugging a factory of mine. It's a ModelFactory (Pydantic), with a PositiveInt field (built-in Pydantic type), and I found that Polyfactory barely differentiated those ints. I traced that to the create_random_float function, and values passed to it.

I temporarily worked around this by using a custom Pydantic type with annotated_types.Le, but believe this is not sane default behaviour.