outdatedLabs 0.1.1

A secure package for training machine learning models

OutdatedLabs - Secure Machine Learning Training Package

A secure and easy-to-use package for training machine learning models with remote server support.

Installation

pip install outdatedLabs

Quick Start

1. Start the ML training server:

cd mlTrainingServer
uv run main.py

2. Use the package in your code:

from outdatedLabs import SecureModel

# Create a linear regression model
model = SecureModel.linearRegression()

# Train the model
model.train(
    dataset_hash="your_dataset_hash",
    features=["feature1", "feature2"],
    target="target_column"
)

# Get training metrics
metrics = model.get_metrics()
print(metrics)

# Download and load the trained model
model.download_model()
loaded_model = model.load_model()

Features

• Secure model training with remote server support
• Automatic dataset download and cleanup
• Progress tracking with tqdm
• Comprehensive error handling
• Detailed logging
• Support for multiple algorithms (currently Linear Regression)

Configuration

The package connects to a local ML training server by default at http://localhost:3000. You can change the server URL when creating a model:

model = SecureModel.linearRegression(server_url="http://your-server:3000")

API Reference

SecureModel

The main class for model training and management.

Methods

• linearRegression(server_url: str = "http://localhost:3000"): Create a Linear Regression model
• train(dataset_hash: str, features: List[str], target: str): Train the model
• get_metrics() -> Dict: Get training metrics
• download_model() -> str: Download the trained model
• load_model() -> Any: Load the trained model

License

MIT License