Model Configuration Loader

Loads and instantiates python classes from a configuration file.

• Free software: MIT license
• Documentation: https://cfloader.readthedocs.io.

Features

• Auto class instantiation from json configuration
• Read configuration from archives
• Load dependencies between classes
• Load configuration parameter as class or as object

Installation

pip install cfloader

Usage

Create a configuration file with parameters:

{
    "model": {
        "model_name": "SomeClass",
        "param_1": 20
    },
    "epochs": 10
}

Now you can load the configuration parameters, either as primitive dicts or as a class:

import cfloader

class SomeClass:
    def __init__(self, param_1):
        self.param_1 = param_1

loader = cfloader.open("config.json")
num_epochs = loader.load("epochs") # = int(10)
model_param_1 = loader.load("model.param_1") # = int(20)
model_configuration = loader.load("model") # = {"model_name": "SomeClass", "param_1": 20}
model_class = loader.load("model", as_class=True) # = <SomeClass object (param_1 = 20)>

For a more extensive example, see examples/pytorch/example_pytorch.py.

Note

If you want to run the example, run it as a python module: python -m examples.pytorch.example_pytorch --help

You may need to install pytorch pip install torch torchvision tqdm

Credits

This package was created with Cookiecutter and the audreyr/cookiecutter-pypackage project template.