Live probing of generators

You can use the MOSTLY AI Python client to probe a trained generator for a number of synthetic samples in real-time without the need to generate a synthetic dataset.

python

g = mostly.generators.get('INSERT_GENERATOR_ID') # Get the generator by its ID
sp = mostly.probe(g, size=3)                       # Probe the generator for 3 synthetic samples

Based on the specified size, you get 3 synthetic data samples generated.

   age workclass education      marital-status       occupation relationship   race     sex  hours-per-week      native-country income
0   29   Private   HS-grad  Married-civ-spouse     Craft-repair      Husband  White    Male              55       United-States  <=50K
1   36   Private   HS-grad  Married-civ-spouse  Exec-managerial      Husband  White    Male              40       United-States   >50K
2   23   Private       9th       Never-married     Adm-clerical    Unmarried  White  Female              40  Dominican-Republic  <=50K

You can also request the creation of specific samples by providing a seed. For example, to generate 3 synthetic samples where the sex of the subjects is always Female, you can use the following:

python

seed = pd.DataFrame({'sex': ['Female', 'Female', 'Female']})
sp = mostly.probe(g, seed=seed)

The following are 3 samples created with a generator trained on the US Census Income dataset.

   age  workclass  education marital-status       occupation   relationship   race     sex  hours-per-week native-country income
0   43    Private  Bachelors       Divorced            Sales      Unmarried  White  Female              45  United-States   >50K
1   45  Local-gov  Bachelors       Divorced  Exec-managerial      Unmarried  White  Female              40  United-States  <=50K
2   41    Private  Bachelors  Never-married   Prof-specialty  Not-in-family  White  Female              40  United-States  <=50K