https://github.com/andreasgrv/mlconf

A config module with machine learning in mind
https://github.com/andreasgrv/mlconf

argparse config machine-learning reflection yaml

Last synced: 9 months ago
JSON representation

A config module with machine learning in mind

• Host: GitHub
• URL: https://github.com/andreasgrv/mlconf
• Owner: andreasgrv
• License: other
• Created: 2017-09-13T23:49:12.000Z (almost 9 years ago)
• Default Branch: master
• Last Pushed: 2024-12-16T14:34:11.000Z (over 1 year ago)
• Last Synced: 2025-04-13T15:04:22.349Z (about 1 year ago)
• Topics: argparse, config, machine-learning, reflection, yaml
• Language: Python
• Homepage:
• Size: 37.1 KB
• Stars: 2
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE.txt

Awesome Lists containing this project

README

          
## [![Build Status](https://api.travis-ci.org/andreasgrv/mlconf.svg?branch=master)](https://travis-ci.org/andreasgrv/mlconf) mlconf 

Because in machine learning, sensible defaults don't last long. Especially when choosing hyperparameters:

![Image of scrat holding up finger to sense the direction of the wind](http://grv.unargmaxable.ai/static/images/scrat-air.jpg)

What's more, there are so many hyperparameters that quickly become

unmanageable as command line options. If you move them all to a config file,

it's frustrating to have to change the file each time you try something different.

So why not enjoy the best of both worlds?

**Dynamic argparse arguments with sensible defaults from a yaml file

that can be modified on the fly from the command line.**

### mlconf in a nutshell

* Dynamically set a lot of sane defaults from a [YAML](http://yaml.org) file.

* Be able to override any number of the defaults from the command line.

* Make the returned object easily accessible using . notation.

* Allows for instantiation of classes with reflection using the $classname

and $module parameters at a later time by using the blueprint's *build()* command.

### Installation

>

	pip install mlconf

### Example

For example usage see [my post](https://grv.unargmaxable.ai/posts/mlconf).

### Tests

Tests are run using tox. Versions of python tested are Python 3.6+

To run the tests, run the following:

>   

	git clone https://github.com/andreasgrv/mlconf

	cd mlconf

	pip install -r requirements.txt

	pip install .

	tox

### License

3 clause BSD, see LICENSE.txt file

### Credits

Image is a [neural styled](https://tenso.rs/demos/fast-neural-style)

portrait of Ice Age celebrity sabre-tooth squirrel, scrat.