PQuantML

PQuant is a library for training compressed machine learning models, developed at CERN as part of the Next Generation Triggers project. It is designed to bridge the gap between high-performance ML models and hardware constraints by providing efficient tools for quantization and pruning.

Key Features

• Dual Framework Support: Works seamlessly with both PyTorch and TensorFlow.
• Compression Layers: Replaces standard layers with “Compressed” (for weights) or “Quantized” (for activations) variants.
• Advanced Pruning: Supports various pruning methods with customizable pre-training, training, and fine-tuning steps.
• Hardware Aware: Optimizes models for deployment on constrained hardware like FPGAs.

Installation

pip install pquant-ml
# Or with specific backend support:
pip install pquant-ml[torch]
pip install pquant-ml[tensorflow]

Authors

Developed by Roope Niemi, Anastasiia Petrovych, Arghya Ranjan Das, and others at CERN and partner institutions.