neurolite 0.2.0

AI/ML/DL/NLP productivity library for minimal-code machine learning workflows

NeuroLite 🧠⚡

NeuroLite is a revolutionary AI/ML/DL/NLP productivity library that enables you to build, train, and deploy machine learning models with minimal code. Transform complex ML workflows into simple, intuitive operations.

🚀 Why NeuroLite?

• 🎯 Minimal Code: Train state-of-the-art models in less than 10 lines of code
• 🤖 Auto-Everything: Automatic data processing, model selection, and hyperparameter tuning
• 🌍 Multi-Domain: Unified interface for Computer Vision, NLP, and Traditional ML
• ⚡ Production Ready: One-click deployment to production environments
• 🔧 Extensible: Plugin system for custom models and workflows
• 📊 Rich Visualization: Built-in dashboards and reporting tools

📦 Installation

Quick Install

pip install neurolite

Development Install

git clone https://github.com/dot-css/neurolite.git
cd neurolite
pip install -e ".[dev]"

Optional Dependencies

# For TensorFlow support
pip install neurolite[tensorflow]

# For XGBoost support  
pip install neurolite[xgboost]

# Install everything
pip install neurolite[all]

🎯 Quick Start

Image Classification in 3 Lines

from neurolite import train

# Train a computer vision model
model = train(data="path/to/images", task="image_classification")
predictions = model.predict("path/to/new/image.jpg")

Text Classification

from neurolite import train

# Train an NLP model
model = train(data="reviews.csv", task="sentiment_analysis", target="sentiment")
result = model.predict("This product is amazing!")

Tabular Data Prediction

from neurolite import train

# Train on structured data
model = train(data="sales.csv", task="regression", target="revenue")
forecast = model.predict({"feature1": 100, "feature2": "category_a"})

One-Click Deployment

from neurolite import deploy

# Deploy your model instantly
endpoint = deploy(model, platform="cloud", auto_scale=True)
print(f"Model deployed at: {endpoint.url}")

🌟 Key Features

🤖 Automatic Intelligence

• Auto Data Processing: Handles missing values, encoding, scaling automatically
• Auto Model Selection: Chooses the best model architecture for your data
• Auto Hyperparameter Tuning: Optimizes model parameters using advanced algorithms
• Auto Feature Engineering: Creates and selects relevant features

🎨 Multi-Domain Support

Computer Vision

# Image classification, object detection, segmentation
model = train(data="images/", task="object_detection")
results = model.predict("test_image.jpg")

Natural Language Processing

# Text classification, sentiment analysis, translation
model = train(data="texts.csv", task="text_generation")
generated = model.predict("Once upon a time")

Traditional ML

# Regression, classification, clustering
model = train(data="tabular.csv", task="classification")
predictions = model.predict(new_data)

🚀 Production Deployment

from neurolite import deploy

# Deploy to various platforms
deploy(model, platform="aws")        # AWS Lambda/SageMaker
deploy(model, platform="gcp")        # Google Cloud
deploy(model, platform="azure")      # Azure ML
deploy(model, platform="docker")     # Docker container
deploy(model, platform="kubernetes") # Kubernetes cluster

📊 Advanced Features

Hyperparameter Optimization

from neurolite import train

model = train(
    data="data.csv",
    task="classification",
    optimization="bayesian",  # bayesian, grid, random
    trials=100,
    timeout=3600  # 1 hour
)

Model Ensembles

from neurolite import train

# Automatic ensemble creation
model = train(
    data="data.csv",
    task="regression",
    ensemble=True,
    ensemble_size=5
)

Custom Workflows

from neurolite.workflows import create_workflow

# Define custom ML pipeline
workflow = create_workflow([
    "data_cleaning",
    "feature_engineering", 
    "model_training",
    "evaluation",
    "deployment"
])

result = workflow.run(data="data.csv")

Real-time Monitoring

from neurolite import monitor

# Monitor deployed models
monitor.track(model, metrics=["accuracy", "latency", "drift"])
dashboard = monitor.dashboard(model)

🔧 Configuration

Global Settings

import neurolite

# Configure global settings
neurolite.config.set_device("gpu")  # cpu, gpu, auto
neurolite.config.set_cache_dir("./cache")
neurolite.config.set_log_level("INFO")

Model-Specific Configuration

model = train(
    data="data.csv",
    task="classification",
    config={
        "model_type": "neural_network",
        "epochs": 100,
        "batch_size": 32,
        "learning_rate": 0.001,
        "early_stopping": True
    }
)

📈 Performance Benchmarks

Task	Dataset	NeuroLite	Traditional Approach	Time Saved
Image Classification	CIFAR-10	3 lines	200+ lines	98.5%
Sentiment Analysis	IMDB	2 lines	150+ lines	98.7%
Sales Forecasting	Custom	4 lines	300+ lines	98.7%

🛠️ Supported Models

Computer Vision

• Classification: ResNet, EfficientNet, Vision Transformer
• Object Detection: YOLO, Faster R-CNN, SSD
• Segmentation: U-Net, DeepLab, FCN

Natural Language Processing

• Text Classification: BERT, RoBERTa, DistilBERT
• Text Generation: GPT-2, T5, BART
• Translation: MarianMT, T5
• Question Answering: BERT, RoBERTa

Traditional ML

• Classification: Random Forest, XGBoost, SVM, Logistic Regression
• Regression: Linear Regression, Random Forest, Gradient Boosting
• Clustering: K-Means, DBSCAN, Hierarchical
• Ensemble: Voting, Stacking, Bagging

🔌 Plugin System

Extend NeuroLite with custom models and workflows:

from neurolite.plugins import register_model

@register_model("my_custom_model")
class CustomModel:
    def train(self, data):
        # Custom training logic
        pass
    
    def predict(self, data):
        # Custom prediction logic
        pass

# Use your custom model
model = train(data="data.csv", model="my_custom_model")

📚 Documentation

• Getting Started Guide
• API Reference
• Tutorials
• Examples
• Plugin Development

🤝 Contributing

We welcome contributions! Please see our Contributing Guide for details.

Development Setup

git clone https://github.com/dot-css/neurolite.git
cd neurolite
pip install -e ".[dev]"
pre-commit install

Running Tests

pytest tests/ -v

Code Quality

black neurolite/ tests/
flake8 neurolite/ tests/
mypy neurolite/

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

🙏 Acknowledgments

• Built with ❤️ by the NeuroLite Team
• Powered by PyTorch, Transformers, Scikit-learn, and other amazing open-source libraries
• Special thanks to our contributors and the ML community

📞 Support

• Documentation: https://neurolite.readthedocs.io
• Issues: GitHub Issues
• Discussions: GitHub Discussions
• Email: [e-mail]

---

Made with ❤️ for the AI/ML community