isage-sias 0.1.0

Sample-Importance-Aware Selection (SIAS) - Continual learning and coreset selection algorithms

SAGE SIAS (Sample-Importance-Aware Selection)

Independent package for sample-importance-aware selection, continual learning, and coreset algorithms

🎯 Overview

sage-sias provides Sample-Importance-Aware Selection algorithms for:

• Continual Learning: Efficient sample selection for continual/lifelong learning scenarios
• Coreset Selection: Select representative subsets from large datasets
• Active Learning: Importance-based data selection strategies
• Tool/Trajectory Curation: Select important samples for agent training

📦 Installation

# Basic installation
pip install isage-sias

# With PyTorch support
pip install isage-sias[torch]

# Development installation
pip install isage-sias[dev]

🚀 Quick Start

Continual Learning

from sage_sias import ContinualLearner

# Create continual learner
learner = ContinualLearner(
    buffer_size=1000,
    selection_strategy="importance"
)

# Add samples
for data, label in stream:
    learner.add_sample(data, label)

# Get selected samples
important_samples = learner.get_buffer()

Coreset Selection

from sage_sias import CoresetSelector

# Create coreset selector
selector = CoresetSelector(
    target_size=100,
    method="kmeans++"
)

# Select representative samples
coreset = selector.select(dataset, features)

📚 Key Components

1. Continual Learner (continual_learner.py)

Manages sample selection for continual learning:

• Buffer management with importance-based eviction
• Multiple selection strategies (random, importance, diversity)
• Support for experience replay

2. Coreset Selector (coreset_selector.py)

Selects representative subsets:

• K-means++ based selection
• Diversity-aware sampling
• Importance scoring
• Support for large-scale datasets

3. Types (types.py)

Common data types and protocols:

• Sample representation
• Importance scoring interfaces
• Selection strategies

🔧 Architecture

sage_sias/
├── continual_learner.py    # Continual learning with buffer management
├── coreset_selector.py      # Coreset selection algorithms
├── types.py                 # Common types and protocols
└── __init__.py             # Public API exports

🎓 Use Cases

1. Agent Training: Select important trajectories for fine-tuning
2. Data Pruning: Reduce dataset size while maintaining performance
3. Active Learning: Query most informative samples
4. Memory Management: Maintain representative samples in limited buffers
5. Transfer Learning: Select relevant samples for adaptation

🔗 Integration with SAGE

This package is part of the SAGE ecosystem but can be used independently:

# Standalone usage
from sage_sias import ContinualLearner, CoresetSelector

# With SAGE agentic (optional)
from sage_agentic import AgentTrainer
from sage_sias import CoresetSelector

trainer = AgentTrainer()
selector = CoresetSelector(target_size=100)
important_trajectories = selector.select(all_trajectories)
trainer.train(important_trajectories)

📖 Documentation

• Repository: https://github.com/intellistream/sage-sias
• SAGE Documentation: https://intellistream.github.io/SAGE-Pub/
• Issues: https://github.com/intellistream/sage-sias/issues

🤝 Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

📄 License

MIT License - see LICENSE file for details.

🙏 Acknowledgments

Originally part of the SAGE framework, now maintained as an independent package for broader community use.

📧 Contact

• Team: IntelliStream Team
• Email: [e-mail]
• GitHub: https://github.com/intellistream