v3.0.0 🎉

params-proto v3.0.0 - Official Stable Release 🎉

params-proto v3 is now stable! This is a complete redesign with a modern, decorator-based API.

Installation

pip install params-proto==3.0.0

Key Features

Decorator-Based Syntax

from params_proto import proto

@proto.cli
def train(
    lr: float = 0.001,  # Learning rate
    epochs: int = 100,  # Number of epochs
):
    """Train a model."""
    print(f"Training with lr={lr}, epochs={epochs}")

Environment Variables with OR Operation

api_key: str = EnvVar @ "API_KEY" @ "SECRET_KEY" | "default"

Hyperparameter Sweeps

from params_proto.hyper import piter

configs = piter @ {"lr": [0.001, 0.01]} * {"batch_size": [32, 64]}

Full Type Support

• Enum, Literal, Path, Tuple types
• Union types for subcommand-like behavior
• Automatic type inference for untyped attributes

Documentation

• Quick Start
• User Guide
• Migration from v2

What's New Since v2

• ✅ Cleaner decorator syntax (no inheritance needed)
• ✅ Full IDE support with type hints
• ✅ Inline documentation becomes automatic help text
• ✅ Support for functions, not just classes
• ✅ Better Union types and Enum support
• ✅ Simplified singleton pattern with @proto.prefix
• ✅ EnvVar OR operation and lazy loading