Model parameters

by: Kevin Broløs & Meera Machado
(Feyn version 3.0 or newer)

---

You can output and inspect all parameters as a python dict using the params property on a Model. It's mostly useful for programmatic parsing and relies on the positional node orderings.

For other cases, we recommend you use the get_parameters function on the Model instead. It takes the name of the input or output and returns a pandas.DataFrame with its parameter values.

If your Model has multiple inputs with the same name in different positions, this is disambiguated with their ordinal position that you can cross-reference in the Model graph by checking the number in the corner of the node.

Example

Here is an example of how to display the parameters for a model that contains categorical and numerical inputs.

import pandas as pd
import feyn

ql = feyn.QLattice()

data = pd.DataFrame(
    {
        'a': [1, 2, 3, 4, 5, 6],
        'cat': ["Three", "Two", "Four", "Three", "Two", "One"],
        'y': [4, 4, 7, 7, 7, 7]
    }
)

models = ql.auto_run(data, output_name='y', stypes={'cat':'c'}, max_complexity=3, n_epochs=1)

best_model = models[0]
best_model

Parameters of 'cat'

Parameters of 'a'

Parameters of 'y'

best_model.get_parameters(name='cat')

	cat
category
Four	0.376742
Three	0.143783
Two	-0.089089
One	-0.321966

best_model.get_parameters(name='a')

	a
scale	0.400000
scale_offset	3.500000
w	0.582286
bias	0.993797
detect_scale	0.000000

best_model.get_parameters(name='y')

	y
scale	1.500000
scale_offset	0.000000
w	2.862249
bias	0.572713
detect_scale	0.000000

---

Location in Feyn

This function can also be found in the feyn.tools module.

from feyn.tools import get_model_parameters

get_model_parameters(best_model, 'cat')