sparkwheel 0.0.2

A powerful YAML-based configuration system with references, expressions, and dynamic instantiation

Define your workflow in YAML, instantiate it in Python

---

Installation

pip install sparkwheel

Requirements: Python 3.10+

Quick Start

Define your workflow in YAML, then use it in Python. Objects are created with their dependencies automatically resolved.

config.yaml network: _target_: torch.nn.Linear in_features: 784 out_features: 10 optimizer: _target_: torch.optim.Adam params: $@network.parameters() lr: 0.001

main.py from sparkwheel import ConfigParser parser = ConfigParser() parser.read_config("config.yaml") network = parser.get_parsed_content("network") optimizer = parser.get_parsed_content("optimizer")

Features

• Declarative: Build complex applications using simple YAML configs
• References: Link configuration values with @ for resolved Python objects or % for raw YAML values
• Composition: Combine multiple configuration files seamlessly
• Expressions: Execute Python code within configs using $ prefix

Usage

📚 Documentation →

About

Sparkwheel is a hard fork of MONAI's configuration system, stripped down to focus and improve its usability for general purposes. We're deeply grateful to the MONAI team for their excellent foundation.